Stuart C. Dodd
Systemed Studies in the Behavioral Sciences
By
Stuart C Dodd
Compiled and Edited
by
Burt Webb
Produced
by
Barbara Whitt
Systemed Studies in the Behavioral Sciences
By
Stuart C Dodd
Compiled and Edited
by
Burt Webb
Produced
by
Barbara Whitt
Systemed Studies on Interactive Symbolizing
Published by
Dodd Memorial Library
Stuart C Dodd Institute for Social Innovation
First Edition
Draft 1
March 2012
Copied with permission from originals in Manuscripts,
Special Collections,
University of Washington Archives Division
University of Washington Libraries
Seattle, WA
Not to be reproduced without the permission of Special Collections
Table of Contents
The Life and Work of Stuart C Dodd 11
Sample of Stuart C Dodd’s ideas: 13
A Preview Introducing and Evaluating the "Pan-Acts Matrices" (excerpt) 13
Things Categories of Cosmists’ Actions or Scientists’ Four Aims 15
Things Liked Most 16
Pan-Acts Cosmos Pictured as the Mass-Time Triangle 18
General Systems: A Creative Search for Synthesis (excerpt) 19
Stuart C Dodd Institute for Social Innovation 21
Purposes 22
Our Mission 22
Our Methods 22
SCDI Founder: Richard Spady 23
The Leadership of Civilization Building 24
The Forum Foundation 24
Presidential Address project 24
DVDs available from the Forum Foundation 24
Founding SCDI Executive Director: Rev. Dr. Richard S. Kirby (1949-2009) 25
World Network of Religious Futurists 25
SCDI Catalyst: August T. Jaccaci 27
2008 Thomas Jefferson Returns 27
Jefferson 2040 27
Unity Scholars 28
Futurum Grid, 46th Annual Creative Problem Solving Institute Reference Sheet 29
SCDI Dodd Memorial Library Editor: Burt Webb 31
The Nexilist Notebook 31
Dodd Memorial Library 33
Dodd Memorial Library Editor Note 33
Volumes in the Dodd Memorial Library 34
Publications of Stuart C. Dodd 39
An Isosceles Distribution of Material Entities 55
I. The scales of mass, time, and "rank" of science 56
II. The isosceles distribution of all material entities. 57
III. Implications of the "mass-time triangle." 59
Notes 68
The Scientific Measurement of Fitness for Self-Government 69
The Facts to Date 70
A Proposed Measuring Instrument for Self-government 71
Application of the Measuring Instrument 76
References 81
Techniques for World Polling 83
I. Planning for International Polling 83
II. A Matrix Model for Standardizing Polling 84
III. Review of General Publications 90
IV. The Administering Stage 91
V. The Designing Stage 93
VI. The Questioning Stage 94
VII. The Sampling Stage 95
VIII. The Interviewing Stage 96
IX. The Analyzing Stage 97
X. The Reporting Stage 98
XI. The Interrelating Stage 98
XII. Recommendations for Developing World Polls 99
Notes 100
The Logistic Diffusion of Information through Randomly Overlapped Cliques 101
I. The Problem 103
II. Background 104
III. Procedure 105
IV. Findings 106
V. Explanation 111
VI. An Application 112
VII. Research Needed 113
References 114
Notes 115
Tables of Contents of Stuart C. Dodd books 117
Social Relations in the Middle East 118
Table of Contents - The plan of the book 121
List of Plates 123
I. The purpose of this book 125
II. The limits of this book 126
III. The plan of this book. 127
Light from the Near East 134
A Controlled Experiment on Rural Hygiene in Syria 137
Table of Contents 138
Appendices 141
List of Illustrations 142
Preface 143
Acknowledgements 145
Abstracts 146
Abstract of Part I. – A Hygiene Scale 147
Abstract of Part Il. – Hygienic Status 147
Abstract of Part III. – Hygienic Progress 148
Abstract of Part IV. – Hygienic Forces 149
Dimensions of Society 150
Table of Contents 152
Science and Sociology 154
I. Purpose of This Volume 154
II. The Concept of Science (In This Volume) 154
III. The Concept of Sociology (In This Volume) 160
IV. The Concept of "Quantitative" (In This Volume) 162
V. The Concept of a System (In This Volume) 164
VI. Notes 170
Systematic Social Science 171
The Intention of This Offset Edition 172
List of Plates 176
List of Tables 182
Systematic Social Science, S = sX;s 184
Detailed Table of Contents 185
Section A: Introduction - Analyzing Observable Elements 185
Section B: Environment - Analyzing Spatial Dimensions 25 185
Section C --- People - analyzing Populational Dimensions 51 186
Section D - Characteristics - Analyzing Indicatory Dimensions 187
Section E – Change - Analyzing Temporal Dimensions 189
Section F - Introduction Synthesizing Recorded Dimensions 190
Section G -- Controls Synthesizing - Correlated Factors 191
Section H – Crafts - Synthesizing Group Controls 192
Section I -- Communities - Synthesizing Regional Crafts 193
Section J – Societies Synthesizing Coeval Communities 193
Section K Social Science Studying Social Situation 194
A Social Distance Test in the Near East 196
Abstract 196
Notes 205
The Likability Models for Predicting Probable Acts of Men 293
I. The “Likes” Model 293
A. The Problem 293
B. The Observing 294
C. The Likes Hypotheses 296
D. The Testing (Ref. 25) 298
E. The Applications 299
F. The Systematizing 299
II. The "Likables" Model 302
A. The Problem 302
B. The Observing 302
C. The Hypotheses 303
D. The Testing 304
E. The Applying 308
F. The Systematizing 310
III. The Full Likability Models 310
A. The Problem 310
B. The Observing 311
C. The Hypotheses 312
D. The Testing 313
E. The Applying 315
F. The Systematizing 315
Exhibits 318
Exhibit A: The Likes-Ratings Scales 319
Exhibit B: Ratings of Organizations 320
Exhibit Ca: Baseline Experiment 322
Exhibit Cb: Simulated Experiment 324
Exhibit D: Project Worth, Test 1 1963 326
I. What is liked Most: 326
II. Specifications of this "Likables Scaling” 328
III. Trial Scorings of Likables 329
IV. Hypothesis on Likables 329
References 357
The Life and Work of Stuart C Dodd
Stuart C Dodd was born in 1900 in Talas, Turkey, where his father was a medical missionary. He received B.S. and M.A. degrees and, in 1926, a Ph.D. in psychology, all from Princeton. Dodd developed and directed the Social Science Research Section at the University of Beirut from 1927 to 1947. He married Betty Dodd in 1928. They had two sons, Peter and Brian. During World War II, he served as Director of Surveys with the U.S. Army in Sicily. In 1947, Dodd accepted an offer to direct the Washington Public Opinion Laboratory at the University of Washington, a position he held for the next 14 years. He left the laboratory in 1961 to devote more time to his research. He retired from the University in 1971. He passed away during the Christmas holidays of 1975 while on a trip to see his son Brian in California.
In 1951, the Air Force's Human Resources Research Institute (HRRI) awarded the Washington Public Opinion Laboratory a contract, to research the effects of leaflet drops on U.S. communities under the name, "Project Revere". Dodd served as principal researcher for Project Revere which ended in 1958. Dodd's influence on mass communications was significant. He trained a generation of mass communications researchers and his writings suggest a clear and profound ideological purpose that shaped his work.
Dodd's published writings between 1939 and 1974 were extensive and varied, and they demonstrate the degree to which he sought to make social phenomena reducible to and controllable by, mathematical formulae. "It is possible with our present knowledge to begin constructing a quantitative systematic science of sociology," he declared in his 1942 text Dimensions of Society. In 1951, he attempted to provide operational definitions, through the use of mathematical equations, for such concepts as "freedom," "equality," and "democracy." In a 1951 article appearing in Educational Theory, Dodd sought "to translate the traditional concepts of the Christian religion into the terms of modern social science." Dodd explained how social scientists view good and evil, the soul, sin, prayer, and other such notions. A 1959 article which possessed direct bearing on how he viewed communication, was entitled An Alphabet of Meanings for the Oncoming Revolution in Man's Thinking. Recognizing the rapid development that had taken place in communication technology since the turn of the century, and seeing a resultant inefficiency and imprecision in most communicative acts, Dodd proposed the development of a single, international language based on mathematics. Dodd noted that his new ten-letter alphabet, referred to as "TILP," would be a perfect symbolic system, capable of expressing every human meaning, without indefiniteness or waste of energy. Dodd asserted that adoption of this new alphabet of meaning would revolutionize and streamline all thought, reduce conflicts and misunderstanding, and even shorten the extended period of time that students spend in school. Finally, according to Dodd, adoption of this new alphabet of meaning would "revolutionize human speech and thinking and thereby significantly accelerate man's cultural evolving.”
Dodd pursued his quest for the understanding of human society beyond the social sciences. This work included his Mass-Time triangle which was a diagram that included the physical, biological and social sciences in one hierarchical model arranged according to the mass of the entities involved. He originally called his cosmic model, the Epicosm. His basic concept was that ceaseless random interaction of the entities at each level of his model generated everything in the Universe. Later, he changed the name of his model to the Pan-Acts Cosmos in order to better express his basic concept.
"My lifelong quest is for greater unity, pervading and tying together all diversity. Whether divergent counsel in a group's formal discussion where I seek the synthesizing motion; or seeking a life center for emotional satisfactions as I find in Betty; or searching for a simpler yet ever more inclusive formula for all things knowable to man as I developed in the Pan-Acts equation, a/ct=1, for God c (If seen as the Creator and Ruler of All in self-creating and over-all ruling cosmos when defined as the Universal set (Uo= 1) of all things namable) - all these and much more are manifestations of my mostly subconscious quest towards integrating - always trying to systematize from chaos, forever wanting to see things more wholly and as a whole."
Stuart C. Dodd, 1971
Sample of Stuart C Dodd’s ideas:
A Preview Introducing and Evaluating the "Pan-Acts Matrices" (excerpt)
64 Exhibits of the Pan-Acts Model for the Dimensions of Cosmos
in 16 Conspectuses of 64 pages
I. What prompted this Exhibits booklet?
At the age of 74, after two heart attacks in the last four months, I have decided to record this sixteen-year inquiry into the cosmos in preprint form so that, should I leave it unfinished, other cosmists can continue to explore and develop, to test and apply, this cosmic research. I intend to flesh out these skeletal outlines as far as time and health and assistance may make possible.
Such writing up of the Pan-Acts Modeling will use:
A. these 64 pages of exhibits as the gist of the modeling, supported by
B. 400 "EpiDocs" of some 2000± pages, which are mimeographed or Xeroxed studies preparing for fuller publication (Individual copies or whole sets are available at Xeroxing cost.);
C. 140 published research articles (and ten books) under contract with Gordon and Breach for republication in four volumes in 1975-76, entitled Systemed Studies on Human Transacting.
D. 30 notebooks, chronologically ordered, of my daily "Dawning Thots" (= "DT's"), hunches, memos, early drafts, etc., etc.
E. My 5-hour video-taped autobiography (from a seminar at Brown Univ. on "Masters of Sociology."
F. 6 systematizing monographs of some 4500 pages.
I plan a volume on Dimensions of Cosmos, comparing 4 versions in parallel columns on each two-page spread. The four versions, mutually enriching each other, would try to communicate to: (a) lay generalists; (b) scientists; (c) algebraists; (d) geometrists. All readers could enlarge their understanding of this World View by reading as most congenial and informative to them, while augmenting their knowledge by glimpses thru the other languages of science.
II. What recent advances have helped deal afresh with the ancient questions of, Whence the universe? and Whither Mankind? I list in partial answer:
A. Set theory, viewing every word or symbol as a name for the set of instances of its referent;
B. Systems theory, viewing complex wholes, with logarithms as the best algorithm;
C. Semiotics, studying systems of man x symbol x thing interactions;
D. The Zero exponent, X: = 1 , for every set, word, or qualitative entity (identified by sub-scripting) which enables man to Seal with all things qualitative with mathematical rigor equal to current dealing with all things quantitative;
E. Computers, greatly simplifying complex calculating and checking of quantifiable hypotheses;
F. Stochastic processes, discovering continuous creation;
G. Combinatorics (or "Combics" for short), explaining increasingly large sections of scientific laws, processes, forces, formulas, etc., in simple terms of combinings, permutings, or repeatings (called "reiterings" here).
III. What guiding principles helped most in this cosmic inquiry?
My central quest was for better symbolizing (concepts, units, scales, formulas, hypotheses, laws, etc.) for analyzing the cosmos resynthesizably — to mirror its whole activity ever more simply and clearly to man. Mostly I used intuition, setting my subconscious to work while I slept to come up with my volumes of "Dawning Thots" on waking. But always I made explicit use of scientific methods when formulated as: "To so describe whatever is studied as to explain its past genesis, predict its future recurrence, and control its anytime changing better than hitherto."
IV. What use of this cosmic Pan-Acts Model may be expected?
For philosophers, I expect Pan-Acts Modeling dill offer a more simple yet complete, a more explicit and exact, cosmology and epistemology than any present alternative theory.
For theologians, I expect Pan-Act-Theism to help solve ancient problems of the nature of God, God's Will, Good and Evil, the Creation of all things and the Destiny of man. (See EpiDoc 303.)
For scientists, I expect Pan-Acts modeling to prove a more comprehensive yet precise, more clear and operationally testable description and explanation of the whole cosmos than any current rival theory.
For sociologists, I expect the Transact submodel for Society gradually to be used as a societal supersystem or integrative framework within which more specific theories of human society can be expressed and tested.
For people universally, I expect the Pan-Acts model to provide a World View tending to help integrate nations and people, religions and ideologies, in One World Community.
Things Categories of Cosmists’ Actions or Scientists’ Four Aims
[pic]
Things Liked Most
This “Things-Liked” theory of human behavior aims to so describe the probable acts of men as to explain and predict them increasingly and thus help to augment human self-control.
The theory starts by building, thru polls of humanity, a list of human wants which posterity now seems most likely to work and live for. This listing starts, in turn, on this page by inviting every reader to rank and revise the list above so it will best express his own system of values.
This statement of a human value-system, or "things-liked," is intended to be highly;
1. Comprehensive: sampling somewhat typically all ten institutions of any culture;
2. Universal: satisfying all in most religions, ideologies, statuses, and periods;
3. Exact: specifying, in 400 words, 200 items of desiderata
1. Understandable: words-per-syllable ratio = 400/440 = 90% of maximum simplicity;
2. Important: organizing 200 most preferred items into a standardizing hierarchy.
3. This base-line statement of values can help test such axiological hypotheses as:
4. Personal Hypothesis: If each reader substitutes one’s own more preferred items (keeping within 400 words) and rates them, then one can express his own value system.
5. Group Hypothesis: If groups, in controlled experiments, rerate each item after discussion-with-intent-to agree, then closer consensus tends to result.
6. Human Hypothesis: If people everywhere experiment thus, persistently from childhood on up, producing more consensus on more values, then a world value-system for humanity tends to emerge, that is democratically desirable, definitive and durable.
Project Value-systems, S.C. Dodd, University of Washington, Seattle, WA, USA Epidoc 138
This EpiDoc 138 expands the “Dimensions of Societal Planning” matrix, EpiDoc 314:1
|0. Our Personal Present |
|We all, the PEOPLE of our Earth, |
|want now a life of greater worth |
|for each in every place and time; |
|We seek ten means to climb: |
|and thus fulfill mankind |
| | |
|1. Hygenic |6. Domestic |
|We like TO LIVE in health— |We like TO LOVE and be loved— |
|We want less sick of any kind; |as mate or parent, child or friend, |
|We want more whole in heart and mind. |as neighbor, kin or fellow men, |
|We strive to grow both safe and strong; |as living, dead, or yet to be, |
|We year for life filled full and long |each in due ways and due degree. |
| | |
|2. Economic |7. Philanthropic |
|We like TO GET more wealth— |We like TO GIVE what’s used— |
|thru work and trade we ‘re free to choose, |to help men climb, those with least health, |
|ourselves to feed, clothe and amuse, |those least in other wants like wealth; |
|to fill all needs from high to low |and those who long have lacked the most |
|as we from child to adult grow. |of what their own groups prize the most. |
| | |
|3. Political |8. Religious |
|We like TO RULE by law— |We like TO WORSHIP well— |
|thru rulers picked by vote of all |our God, our good, our goals in life. |
|who let no rights nor freedoms fall; |We march in quest but with no strife, |
|with justice and security |for what as holy each may see, |
|in local or world community. |while free to speak of what should be. |
| | |
|4. Recreational |9. Artistic |
|We like TO ENJOY life— |We like TO BEAUTIFY— |
|each day, in work, or play, or rest |ourselves, our homes, and all around— |
|on hill or plain, with more of zest, |thru music, pictures, gardened ground, |
|with memory; at cost that’s due, |so what we touch or taste or smell, |
|with fun for us and playmates too. |so lovely feelings within us dwell. |
| | |
|5. Scientific |10. Educational |
|We like TO LEARN the truth— |We like TO TEACH wisdom— |
|to test out how —as science tries— |to each child here or not yet born, |
|things move, or breathe, or symbolize; |and all who hope from us to learn, |
|to learn to curb our fears and war, |the best of ways to live and grow |
|to progress fed by research more. |thru what feel and do and know. |
| | |
|11. Our Social Future |
|If we our FUTURE Earth |
|plan safe from bomb and dearth |
|then each in roles one plays |
|must help all win in ways |
|that best augment man’s days. |
Pan-Acts Cosmos Pictured as the Mass-Time Triangle
[pic]
General Systems: A Creative Search for Synthesis (excerpt)
JOHN CURTIS GOWAN
STUART C. DODD
During the past several years, a small number of persons*, including both authors of this paper, have been at work on a particular aspect of general systems theory involving creative transformations in the mapping or recognition of isomorphisms between one theory and another. This has been accomplished by mail, individual visits, and by meetings held particularly during the 1974 and 1975 Creative Problem-Solving Institutes at Buffalo. Other meetings have also been held through the aegis of Jeanne Rindge of the Human Dimensions Institute of Buffalo. This is the first and necessarily incomplete report of some of these activities; it is completely unofficial, and should be understood only as the personal perceptions (and possibly the errors) of two of the participants.
…
Table 3: Land-Dodd consolidation using 3 x 3 set.
|(Land) —*(Dodd) |ACCRETIVE |REPLICATIVE |MUTUAL |
|subprocesses of |combining |repeating |permuting |
|interaction |adding |multiplying |empowering |
|math operation |sum |product |power |
|output | | | |
4. In place of two discrete, empirical models, we now have one categorical model, which is a two dimensional set of three elements, capable of being mapped into other theories, and of being extended into three dimensions, such as when it may be possible to map it into the Genesa theory of Langham and the eidetic concepts of Evering.
Stuart Dodd's last written comments on this work follow in the next paragraph which was constructed from his suggestions dated September 15-October 8, 1975.
The relationship between Land's theories and the Dodd Pan-Acts model is shown in Table 3. Note that the second and third rows are the first two of four rows of the reiterating matrix (Dodd, 1975). This matrix summarizes the methodology or scientists' behavior in deriving the Pan-Acts model. Row one generates the symbols, in the Pan-Acts model, row two generates the syntax, rows three and four of the Pan-Acts model indicate the four key processes and the four cosmic submodels by tense; the three tenses are transforms of the three stages in the Land-Gowan model (by taking the middle stage as the present tense.
…
*Among these were Jere Clark, Betty and David Cox, Barbara Hubbard, Gus Jaccaci, Robin King, Lucy Krall, George Land, Derald Langham, Rendle Leatham, Jo Mock, Mort Rapp, and Jeanne Rindge.
Stuart C Dodd Institute for Social Innovation
The Stuart C. Dodd Institute for Social Innovation (SCDI/SI) is a not-for-profit, tax exempt, organization registered as such with the U.S. government. Incorporated on May 9, 1997; 501 (c) (3) application approved on September 9, 1997.
“As human systems and organizations grow ever larger, more complex, and more impersonal- in our schools, in our communities, in our churches, in our governments, and in industries and commerce-the individual shrinks toward facelessness, hopelessness, and frustration.”
Dr. Stuart C. Dodd,
"Citizen Counselor Proposal,"
The Seattle Times, November 10, 1974
Internet web address:
Office Address: 4427 Thackeray Place NE; Seattle, WA 98105-6124
Phone: (206) 545-0547; Fax: (206) 632-1975;
Electronic mail general information: info@
Web master: Webmaster@
Other e-mails: DrRSKirby@ (Director).
LUCSTEW@ (Office Manager)
Purposes
SCDI/SI encourages scholarly, interdisciplinary research in the archives of Stuart C. Dodd (1900-1975), Professor Emeritus of Social Science at the University of Washington. Scholars and associates pursue the intellectual, moral and civic legacies of Dr. Dodd in the fields of sociology, business administration, education, urban planning, sustainable communities, cosmology, statistics, and mathematics. Topics of particular interest include organization development, administrative theory, many-to-many communication, and value reporting.
Our Mission
The Stuart C Dodd Institute for Social Innovation is dedicated to using the full range of human knowledge to achieve a society that is democratic, equitable, just, compassionate, spiritual, sustainable, diverse, and fulfilling.
Our Methods
“We advance social innovation on the successive planes of theory (particularly Social Innovation Theory), organization (particularly The Stuart C. Dodd Institute for Social Innovation with help from the Forum Foundation), event (particularly our annual conferences and regional training activities), in social policy, program and experiment, and communication.
Our social experiments range from uses of the Fast Forum@ technique to the establishment of civic innovation programs in entire cities such as Slidell, Louisiana. We operate with a view to doing ideally profitable social science and social philosophy research deriving from the legacy of Stuart C. Dodd. Our national and international activities collectively act as a model social science "think-tank." We explore new horizons of excellence in self-management and organizational development, in social theory, philosophy of society and political science/philosophy/art, and in the development of breakthrough social technologies. We aim to advance the theory and the practice of social innovation, as part of the legacy of Stuart C. Dodd. We conduct research in sub-fields of social innovation such as educational innovation. We are developing an in-house five-year plan for the planes on which we operate, from the pre-theoretical to the communication/publication plane.
SCDI Founder: Richard Spady
Richard Spady, a student and practitioner of Administrative Theory, is president of the Forum Foundation. A Seattle businessman, he co-founded Dick's Drive-in Restaurants in 1954 and currently is its president. Spady is active in community affairs, and is a lay speaker in The United Methodist Church.
“I learned long ago that if something needs to get done, one has to get organized to do it. So in 1997 I helped organize the Stuart C. Dodd Institute for Social Innovation. It is a 501 (c) (3) non-profit, tax-exempt organization to encourage scholarly, interdisciplinary research in the archives of Stuart C. Dodd. Here is the story behind the Institute.
An institute to continue the work of Dr. Dodd had been a dream of mine for many years. But Dodd's work on mathematical cosmology is not for the faint of heart; few people can understand it, much less move it forward. My acquaintance with the Rev. Dr. Richard S. Kirby through the World Network of Religious Futurists told me he might be the person for the job.
I visited Dr. Kirby in London in 1994. He had finished his Masters of Divinity studies at General Theological Seminary of the Episcopal Church, New York (USA) in 1985 and his Ph.D. in Theology studies at King's College, London from 1987 to 1992. Richard Kirby is an outstanding scholar. I judged he had the capacity to pick up the traces left in the archives of Stuart Dodd.
Dr. Kirby was open to my proposal, and he wanted to return to the United States. I offered him a home and office in Seattle (owned by my business firm and dedicated to public service) and invited him to be the Stuart C. Dodd Chair in Social Innovation at the Forum Foundation. He accepted, came to Seattle in 1995, and is now an American citizen. When the Stuart C. Dodd Institute for Social Innovation was formed in 1997, Richard Kirby became its first Executive Director.”
The Leadership of Civilization Building
by Richard J. Spady and Richard S. Kirby
Administrative and Civilization Theory, Symbolic Dialogue, and Citizen Skills for the 21st Century
“As a new century dawns, this book introduces new truths about the theory and practice of civilization. It offers a fresh and imaginative analysis of the meaning of citizenship and introduces practical tools for the empowerment of people in business, government and religion. Indeed, the authors have broken new ground in the practical definition of civilization building. It invites people everywhere to play a major role in the civilizing of their own civilization. In these pages, you will find:
• An enlightened approach to leadership in organizations and society
• New theories on how to develop your powers as a citizen
• New tools to communicate more effectively through symbolic dialogue”
Available through the Forum Foundation.
The Forum Foundation
“The Forum Foundation ....Conducts futures research in the field of Administrative Theory and Many-to-Many Communication technology to discover those dynamics which tend to move organizations and institutions, universally, toward solving their problems and anticipating or adapting to changes in their environment.”
Presidential Address project
Each year, the Forum Foundation conducts a survey of high school students on the subject of the Presidential address in January for that year, State of the Union or Inaugural Address. Each student is given the opportunity to agree or disagree with every sentence in the address.
DVDs available from the Forum Foundation
Visionary Voices Series with Richard Spady/Dr. Cecil H. Bell, Jr.
The Fast Forum Opinionnaire
A New Approach to “We the People” says Dick Spady
Founding SCDI Executive Director: Rev. Dr. Richard S. Kirby (1949-2009)
Rev. Dr. Richard S. Kirby, Executive Director, former adjunct faculty of the University of Washington's School of Business Administration; former Director of Administration, International Mensa; Chief Executive Officer of the World Network of Religious Futurists; and co-author: The Temples of Tomorrow: World Religions and the Future (Grey Seal Publishers, 1993).
Richard Spady and Richard Kirby met at a conference and found that they shared interests and goals. Spady and Kirby discussed the Stuart C. Dodd Institute for Social Innovation and it was agreed that Kirby would become the Executive Director.
World Network of Religious Futurists
“Religious futures scholarship focuses on predictable occurrences in the future of religion, based on present observable trends, and past trends in religion, compounded by expectations of wild cards or quantum leaps, in the context of society's future as a whole, ranging from science to technology. What kind of science and technology excites you, what kind worries you? Get involved in bringing religious values to the future of these sciences and technologies!”
SCDI Catalyst: August T. Jaccaci
Jaccaci has worked as a futurist, artist, teacher, and international consultant to companies both large and small helping them develop successful business strategies using his METAMATRIX future mapping methodology. In 1995, Jaccaci and Susan B. Gault founded the Social Architects Associates and began their discovery of a natural evolutionary process to charitable and profit-making enterprises.
Gus Jaccaci was a personal friend of Stuart C. Dodd. They both attended annual workshops at the Creative Problem Solving Institute, University of New York at Buffalo. Gus had the opportunity to video tape and interview with Dr. Dodd before his death. In the mid-90s, Gus was instrumental in bringing together the founder of the Institute and the editor of the DML.
2008 Thomas Jefferson Returns
Letters received and recorded by Gus Jaccaci
“Mr. Jefferson is the author of the Declaration of Independence and the social architect of the Bill of Rights is a primary guardian of the American Soul. Under the terrible pressure of losing the American Experiment in self-governance to the forces of fear trending toward fascism, Mr. Jefferson has returned to empower individual Americans to rise up to their full responsibility to protect their souls and their personal and community potential for truth, beauty and goodness which together express love, the meaning and destiny of America.
Mr. Jefferson’s writings received at various moments on and after July 4th within the year 2007 are meant to be letters to all humanity. They are, however, focused on aspects of the American soul lest it journey farther from planetary potential toward planetary problem.”
Jefferson 2040
"Awaking America" is an interactive and spontaneous conversation with a revered social architect of our nation, Thomas Jefferson. Speaking from the year 2040, he addresses the challenges of our time and reveals the creative steps Americans will take in the coming decades to reinvent our politics, our economy and our relationship with all life. Mr. Jefferson combines his view of America's past, present and ideal future into a provocative, entertaining event we describe as a "transformance" since no one leaves unchanged.
Unity Scholars
Unity Scholars is a non-profit organization of social inventors who host visionary people, projects and events dedicated to enhancing healthy human evolution. The purpose of the organization is to address global social and environmental problems by applying the knowledge gained from studying nature's universal patterns and by developing the creative potential of individuals and their communities. Unity Scholars comprises a worldwide web of people who share our dedication to this purpose and who support our endeavors through the exchange of ideas, teaching and participation in Unity Scholar's projects.
Futurum Grid, 46th Annual Creative Problem Solving Institute Reference Sheet
[pic]
SCDI Dodd Memorial Library Editor: Burt Webb
Burt Webb is a computer consultant in the Seattle Washington area. He has written, contributed to or edited books and articles on alternative energy, robotics, artificial intelligence, nanotechnology ,SETI, ethno-botany, social impact of technology, current affairs, economics, sociology and psychology as well as science fiction, fantasy and adventure novels and movie scripts.
“I met Stuart C Dodd during the founding of the World Future Society Evergreen Chapter in Seattle, Washington in the early 70s. We enjoyed discussing comprehensive multileveled models of reality. In 1974, he asked for my assistance in writing up some of his material for a general audience. He was a very warm, cordial, intelligent and witty man. I enjoyed our association very much. Dr. Dodd passed away in 1975 but his work was ahead of his time and is very relevant today. After his death, all his papers were given to the University of Washington and still reside in the archive there. Access is restricted and the papers were never cataloged. It is very difficult for anyone interested in his ideas to access his writings.
I also met Dick Spady at the WFS Evergreen Chapter meetings. We had many lively conversations about futurist subjects. In the late 1990's, Dick approached me and asked for my assistance in bringing Stuart's work to a new audience. I was glad to have the opportunity to assist in this project. Due to the generous support and encouragement of his friend, Dick Spady, the valuable work of Stuart C. Dodd will now be available to scholars everywhere who wish to explore his legacy.”
Burt Webb
The Nexilist Notebook
I have been interested in interdisciplinary studies since I was a kid. So this blog will cover a wide variety of subjects including how everything is connected. Politics, society, psychology, technology, religion, myth, dreams, the arts, and winning at the game of life.
Dodd Memorial Library
The Dodd Memorial Library is a project of the Stuart C Dodd Institute for Social Innovation. The purpose of the project is to publish a series of volumes that contain the work of Stuart C Dodd. Dodd’s intellectual legacy resides in the Special Collections of the University of Washington in Seattle, Washington. Dodd willed his writings to the University upon his death. They consist of a series of file boxes that occupy about 110 linear feet of shelving at the U of W. They contain his published and unpublished books and articles. In some cases, he had self published collections of his papers. In other cases, he left detailed Tables of Contents for books that were intended as collections of his papers. The archive also contains his correspondence, notes, working documents from his years as a college professor, working papers from the Washington Public Opinion Library and other personal papers. The DML will consist of all his published work collected and edited into a series of volumes by subject. It will also contain a collection of biographical material in one volume, a research guide to his collected works at the U of W and an encyclopedia that contain representative papers that cover the breadth of his life’s work. The volumes will be announced on the Institute website as they are released and will be available for purchase.
Dodd Memorial Library Editor Note
Stuart produced 9 books, thousands of research papers and hundreds of journal articles during his 50 year career. The material occupied over 110 feet of files. Over two years, I managed to obtain copies of his books and over 700 articles and papers. During his life, he created a number of collections of articles and tables of contents for a number of new books. I found these tables of contents and proceeded to collect all the articles mentioned. I am now in the process of creating a 25 volume set of books covering the life's work of Stuart C Dodd. This series will be made available thru the Stuart C. Dodd Institute for Social Innovation. Inquiries about the series should be referred to the following contacts:
Burt Webb: Editor
(206) 729-7410
phoenix@
Volumes in the Dodd Memorial Library
Behavioral Sciences
Best Published Articles
Best Unpublished Articles
Dimensions of Cosmos
Dimensions of Society
Epicosm
Fitness for Self-government
General Organization
Horizons of Thought
Human Values
Interactive Symbolizing
Message Diffusion
On Language
Pan Acts Cosmos
Pan Acts Theism
Probable Acts of Man
Probable Acts of Men
Public Opinion Research
Dodd Research Guide
Revere Studies on Interaction
Social Relations in the Near East
Synthesizing Oneself, Society and Cosmos
Systematic Social Science
Transact Modeling
Techniques for World Polls
(Bolded volumes are currently available – email phoenix@ for details)
Systemed Studies in the Behavioral Sciences
Introduction
This set of fourteen articles was republished in the form of microfilm in the Sociological Microjournal, Copenhagen, 1970. It was necessary to contact the University of Copenhagen in order to obtain copies of the articles.
Publications of Stuart C. Dodd
|SD:70-44 | |
| | |
|* All out of print | |
|# Letters with a number at the right mark articles in the republication volumes, | |
|Systemed Studies on Human Transactions, (Gordon and. Wrench) as follows: | |
|Microfilm S = Studies in Behavioral Science, Sociological Microjournal, | |
|Copenhagen, 1970. | |
|Volume 1: V = Systemed Studies on Human Valuing | |
|Volume 2: D = Systemed Studies on Message Diffusing | |
|Volume 3: P = Systemed Studies on Opinion Polling | |
|Volume 4: I = Systemed Studies on Interactive Symbolizing | |
| | |
|Number after letter indicates the position of the article in the particular volume | |
| | |
|Books | |
| | |
|1. International Group Mental Tests Princeton University Press, 1926, 134 pp. | |
|2. Social Relations in the Near East (Tables of Contents) |S# |
|American University of Beirut | |
|1st edition, 1931, 633 pp. | |
|2nd edition, 1940, 790 pp. | |
|3rd edition, 1946, 904 pp. | |
|Arabic edition, 1947, 389 pp. | |
| | |
|3. A Post War Bibliography of the Near Eastern Mandates, (editor) | |
|American University Social Science Series, No. 1, 1933, 1,650 pp. | |
|4. A Controlled Experiment on Rural Hygiene in Syria, |S |
|American University of Beirut, Social Science Series, No. 7, 1934, 336 pp. | |
|5. Dimensions of Society, Macmillan, 1942, 944 pp. (Tables of Contents) |S |
|6. Surveys of Public Opinion Held in .Sicily, | |
|printed for Psychological Warfare Branch of Allied Forces Headquarters, March, 1944, 150 pp. | |
|7. A Pioneer Radio Poll in Lebanon, Syria and Palestine, | |
|Government Press, Jerusalem, September, 1943, 103 pp. | |
|8. Systematic Social Science, (Table of Contents) |S |
|American University of Beirut, Social Science Series, No. 16, University of Washington Bookstore, Seattle, 1947, 788 pp. | |
|9. Probable Acts of Men, Vols. I and II, 44 Collected Papers by S. C. Dodd, |S |
|Offset by State University of Iowa, 1963, 700 pp. | |
| | |
| | |
|II. Research Articles | |
| | |
|1926 | |
| | |
|1. "The Applications and Mechanical Calculation of Correlation Coefficients," The Journal of the Franklin Institute, March, 1926. |P15 |
|2. "A Correlation Machine," Industrial Psychology, Vol. I, January, 1926. |P16 |
| | |
|1927 | |
| | |
|3. "On Criteria for Factorizing Correlated Variables," Biometrika, Journal for the Statistical Study of Biological Problems, Vol. |P30 |
|XIX. Parts 1 and 2, July, 1927. | |
| | |
|1928 | |
| | |
|4. "The Coefficient of Equiproportion as a Criterion of Hierarchy," Journal of Educational Psychology, April, 1928, Vol. XIX, No. 4.|P26 |
|5. "The Theory of Factors," Psychological Review, Vol. 35, No. 3, May, and No. 4, July, 1928. |I6 |
| | |
|1929 | |
| | |
|6. "On the Sampling Theory of Intelligence," The British Journal of Psychology (General Section), Vol. XIX, Part 3, January, 1929. |I7 |
| | |
|1934 | |
| | |
|7. "Progress Inductively Defined," The International Journal of Ethics, Vol. XLIV, No. 3, April, 1934. |V6 |
| | |
|1935 | |
| | |
|8. "A Social Distance Test in the Near East," The American, Journal of Sociology, Vol. XLI, No. 2, September, 1935. |V12 |
| | |
|1936 | |
| | |
|9. "A Test of Group Preferences in the Near East," American University of Beirut, 1936, 4 pp. |V11 |
|10. "Abstract of S-Theory – A Dimensional Theory for Systematizing Quantitative Data in the Social Sciences," American University of |I23 |
|Beirut, 1936. | |
|11. "A Theory for the Measurement of Some Social Forces," The Scientific Monthly, Vol. XLIII, No. 1, No. 250, July, 1936. |I13 |
|12. "The Standard Error of a 'Social Force'," The Annals of Mathematical Statistics, Vol. VII, No. 4, December, 1936. |P24 |
| | |
|1939 | |
| | |
|13. "A Tension Theory of Societal Action," American Sociological. Review, Vol. IV, No. 1, February, 1939. |P29 |
|14. "A System of Operationally Defined Concepts for Sociology," American Sociological Review, Vol. IV, No. 5, October, 1939. |I24 |
|(U:60-166) | |
| | |
|1940 | |
| | |
|15. "The Interrelation Matrix," Sociometry, Vol. III, No. 1, 1940. |P23 |
|16. "Analyses of the Interrelation Matrix by its Surface and Structure," Sociometry, Vol. III, No. 2, 1940. |I14 |
| | |
|1943 | |
| | |
|17. "Operational Definitions Operationally Defined," American Journal of Sociology, Vol. XLVIII, No. 4, January, 1943. |I3 |
|18. "Induction, Deduction, and Causation," Sociometry, Vol. VI, May, 1943. |I15 |
|19. "Sociometry Delimited: Its Relation to Social Work, Sociology, and the Social Sciences," Sociometry, Vol. VI, No. 3, August, 1943.|I25 |
|20. "Of what Use is Dimensional Sociology? – A Report of Further Research Upon the Utility, Precision and Parsimony of Dimensional |I32 |
|Analysis," Social Forces, Vol. 22, No. 2, December, 1943. (U:61-62) | |
| | |
|1944 | |
| | |
|21. "On Reliability in Polling," Sociometry, Vol. VII, No. 3, August, 1944. 1945 |P13 |
| | |
|1945 | |
| | |
|22. "A Barometer of International Security," Public Opinion Quarterly, Vol. 9, July, 1945. |P1 |
| | |
|1946 | |
| | |
|23. "Toward World Surveys," Public Opinion Quarterly, Vol. X, Winter, 1946. |P2 |
| | |
|1947 | |
| | |
|24. "Standards for Surveying Agencies," Public Opinion Quarterly, Vol. XI, Spring, 1947. |SP4 |
| | |
|1948 | |
| | |
|25. "The Washington Public Opinion Laboratory," Public Opinion Quarterly, Vol. 12, Spring, 1948. |P40 |
|26. "A Simple Test for Predicting Opinions from their Sub-Classes," International Journal of Opinion and Attitude Research, Vol. II, |P25 |
|No. 1, Spring, 15 | |
|27. "A Systematics for Sociometry and for All Science," Sociometry, Vol. XI, No. 1-2, May, 1948. (U:60-168) |V16 |
|28. "A Dimensional Theory of Groups," by V. Cervinka, reported by S. C. Dodd, Sociometry, Vol. XI, No. 1-2, May, 1948. |I16 |
|29. "Developing Demoscopes for Social Research," American Sociological Review, Vol. XIII, No. 3, June, 1948. (U:60-171) |P7 |
|30. "Public Opinion Definitions," International Journal of Opinion and Attitude Research, Vol. II, No. 3, Fall, 1948. |P22 |
|31. "On Predicting Elections of Other Public Behavior," International Journal of Opinion and Attitude Research, Vol. II, No. 4, |P17 |
|Winter, 1948-49. | |
| | |
|1949 | |
| | |
|32. "On Measuring Languages," Journal of American Statistical Association, Vol. 44, March, 1949. (SD:63-3) |I8 |
|33. "Note on an Index of Conformity," Science, Vol. 110, No. 2853, September 2, 1949. |P27 |
|34. "Dimensions of a Poll," International Journal of Opinion and Attitude Research, Vol. 3, No. 3, Fall, 1949. |P9 |
|35. "Racial Attitude Survey as a Basis for Social Planning," Dodd and O'Brien, Journal of Educational Sociology, Vol. 23, No. 2, |P33 |
|October, 1949. | |
| | |
|1950 | |
| | |
|36. "A Verifiable Hypothesis of Human Tensions," International Journal of Opinion and Attitude Research, Vol. IV, No. 1, Spring, 1950.|V5 |
|(U58-205) | |
|37. "The Interactance Hypothesis: A Gravity Model Fitting Physical Masses and Human Groups," American Sociological Review, Vol. XV, |D17 |
|No. 2, April, 1950. (U:60-167) | |
|38. "A Call for Experimental Designs for Election Polling," International Journal of Opinion and Attitude Research, Vol. 4, No. 2, |P18 |
|1950. | |
|39. "How to Measure Values," Proceedings of the Pacific Sociological Society, Research Studies of the State College of Washington, |V8 |
|Vol. XIII, 1950. (U:50-86) | |
|40. "An Isosceles Distribution of Material Entities," Main Currents in Modern Thought, Vol. 7, No. 4, Winter, 1950. (U61-60) |SI37 |
|41. "Steps Toward a Barometer of International Security," International Journal of Opinion and Attitude Research, Vol. 4, No. 4, |P3 |
|Winter, 1950-51. | |
| | |
|1951 | |
| | |
|42. "Scientific Methods in Human Relations," The American Journal of Economics and Sociology, Vol. 10, No. 3, April, 1951. (U:61-67) |P12 |
|43. "A Measured Wave of Interracial Tension," Social Forces, Vol. 29, No. 3, March, 1951. (U:61-51) |D2 |
|44. "Predictive Principles for Polls – Scientific Method in Public Opinion Research," The Public Opinion Quarterly, Vol. 15, No. 1, |P11 |
|Spring, 1951. | |
|45. "Religion of a Social Scientist," Educational Theory, Vol. I, No. 1, August, 1951. (U:59-127) |V27 |
|46. "Tilp - A Ten Letter Alphabet of Meanings," General Semantics Bulletin, Nos. 6 and 7, Spring-Summer, 1951. |I21 |
|47. "Sociomatrices and Levels of Interaction – for dealing with plurals, groups, and organizations," Sociometry, Vol. XIV, No. 2 and |P10 |
|3, May-August, 1951. | |
|48. "On Classifying Hunan Values," American Sociological Review, Vol.16, No October, 1951. (U:59-38) |V7 |
|49. "Historic Ideals Operationally Defined," Public Opinion Quarterly, Vol.1, No. 3, Fall, 1951. (SD:63-55) |V10 |
| | |
|1952 | |
| | |
|50. "On All-or-None Elements and Mathematical Models for Sociologists," American Sociological Review, Vol. 17, No. 2, April, 1952. |I17 |
|51. "Something Out of Nothing," Main Currents in Modern Thought, Vol. 8, No. 4, February, 1952. (U:61-52) |I1 |
|52. "Testing Message Diffusion from Person to Person," The Public Opinion Quarterly, Vol. 16, Spring, 1952. |D7 |
|53. "Testing Message Diffusion in C-Ville," Proceedings of the Pacific Sociological Society, Research Studies of the State College of |D4 |
|Washington, Vol. XX, No. 2, June, 1952. (SD:63-81) | |
|54. "On Percentage Moments – interpreting the moments as percentages of their maximums," Scientific Monthly, Vol. LXXIV, No. 4, April,|I9 |
|1952. | |
|55. "Model English," General Semantics Bulletin, Nos. 10-11, Summer, 1952. (U:52-96) |I22 |
|56. "On Estimating Latent from Manifest Undecidedness: the 'Don't Know' percent as a warning of Instability Among the Knowers," with |P19 |
|Kaare Svalastoga, Educational and Psychological Measurement, Vol. 12, No. 3, Autumn, 1952. | |
| | |
|1953 | |
| | |
|57. "Human Dimensions – A Re-search for Concepts to Integrate Thinking," Main Currents in Modern Thought, Vol. IX, No. 4, February, |I30 |
|1953. | |
|58. "A Statement of Human Wants," Educational Theory, Vol. III, No. 2, April, 1953. (U:57-68) |V9 |
|59. "A Dimensional Theory of Diffusion – an analysis, modeling and testing of one-way interaction," Sociometry, Vol. XIV, No. 2, May, |D12 |
|1953. (SD:63-53) | |
|60. "Testing Message Diffusion in Controlled Experiments, Charting the Distance and Time Factors in the Interactance Hypothesis," |D5 |
|American Sociological Review, Vol. XVIII, No. 4, August, 1953. | |
|61. "Testing Message Diffusion – Verbal vs. Graphic Symbols," with Charles E. Bowerman and Otto N. Larsen, International Social |I4 |
|Science Bulletin, UNESCO, September, 1953. | |
|62. "Can the Social Scientist Serve Two Masters – an answer through experimental sociology," Proceedings of the Pacific Sociological |D3 |
|Society, Research Studies of the State College of Washington, Vol. XXI, No. 3, September, 1953, (SD:63-50) | |
|63. "The Cube Rule for Size of Legislatures," (in French) Revue Politique, Vol. IV, Paris, September, 1953. |SI28 |
|64. "Three Semantic Exercises: 'Time,' 'Quality,' and 'God,'" ETC., Vol. XI, No. 1, Autumn, 1953. |I2 |
| | |
|1954 | |
| | |
|65. "Symbolizing the Values of Others," with William R. Catton, Jr., Chapter XXXIV, Symbols and Values: An Initial Study, Conference |V15 |
|of Science, Philosophy and Religion, Harpers, 1954. | |
|66. "The Scientific Measurement of Fitness for Self-Government," Scientific Monthly, Vol. LXXCIII, No. 2, February, 1954. (SD:63-42) |SV18 |
|67. "Physical Dimensions of Social Distance," with Jiri Nehnevajsa, Sociology and Social Research, Vol. XXXVIII, No. 5, May-June, |I19 |
|1954. (U:61-37) | |
|68. "A Dimensional System of Human Values," with Chahin Turabian, Transactions Second World Congress of Sociology, International |V14 |
|Sociological Association, 1954, pp. 100-105. (U:59-98) | |
| | |
|1955 | |
| | |
|69. "Seven-Year Report, 1947-54," Washington Public Opinion Laboratory, University of Washington, Seattle, 1955. 31 pp. |P41 |
|70. "Diffusion Is Predictable: Testing probability models for laws of interaction," American Sociological Review, Vol. XX, No. 4, |D18 |
|August, 1955. | |
|71. "Model English," Chapter 11 in Locke and Booth, Machine Translation of Languages, Technology Press of MIT and John Wiley, 1955. |D33 |
|243 pp. (U:52-96 and SD:63-2) | |
|72. "The Transact Model – a predictive and testable theory of social action, interaction and role-action," Sociometry, Vol. XVIII, No.|D1 |
|4, December, 1955. (U:55-71) | |
|73. "Why Not Scientific Method in Philosophy?" ETC., Vol. 8, Winter, 1955-56. |I11 |
|74. "A Dimensional Analysis in Social Physics," Chapter 10 in Frontiers of Social Science, London: MacMillan and Co., Ltd., 1955. pp. |I18 |
|203-217. U:60-169) | |
| | |
|1956 | |
| | |
|75. "A Test of Message Diffusion by Chain Tags," American Journal of Sociology, Vol. LXI, No. 5, March, 1956. |D10 |
|76. "Testing Message Diffusion in Harmonic Logistic Curves," Psychometrika, Vol. 21, No. 2, June, 1956. (SD:63-54) |D9 |
|77. "A Predictive Theory of Opinion – using nine 'mode-and-tense' factors," Public Opinion Quarterly, Vol. XX, No. 3, Fall, 1956. 23 |V2 |
|pp. (U:61-39) | |
|78. "Raising Research-to-Teaching Ratios – A ten-point program for Sociology Departments when enrollments rise and resources lag," |I33 |
|Social Forces, Vol. 35, No. 2, December, 1956. (U:56-101A) | |
|79. "Review of 'Social Structure and Culture – Change in a Lebanese Village' by John Gulick," Wenner-Gren Foundation, 1955, 191 pp. |* |
|American Sociological Review, Vol. XXI, No. 5, October, 1956. | |
| | |
|1957 | |
| | |
|80. "Conditions for Motivating Men – the valuance theory for motivating behaviors in any culture," Journal of Personality, Vol. 25, |V17 |
|No. 4, June, 1957. (U:56-120) | |
|81. "Techniques for World Polling – a review of the methodological literature for cross-cultural surveys," Gazette, Vol. III, No. 3, |SP5 |
|1957. (U:61-53, U:56-110) | |
|82. "The Counteractance Model – for a resistive part of a whole interaction," American Journal of Sociology, Vol. LXIII, No. 3, |D22 |
|November, 1957. (U:60-136 | |
|83. "A Power of Town Size Predicts its Internal Interacting – a controlled experiment relating the amount of an interaction to the |D23 |
|number of potential inter actors," Social Forces, Vol. 36, No. 2, December, 1957. (1J:56-90) | |
|84. "A Ph.D. Defined in Three Tenses," Journal of Educational Sociology, Vol. 3, May, 1957. |V25 |
|84. "The World Association for Public Opinion Research," Public Opinion Quarterly, Vol. XXI, No. 1, Spring, 1957. |P6 |
|86. "The Scientific Measurement of Fitness for Self-Government," Chapter XIII in Shannon, Lyle, Underdeveloped Areas, Harper and |* |
|Bros., 1957. | |
| | |
|1958 | |
| | |
|87. "Ten Semantic Tangles and the Threat of War," Journalism Quarterly, Vol. 35, No. 2, Spring, 1958. (formerly U:55-72) |I34 |
|88. "The Logarithmic Relation of Social Distance and Intensity," with Keith Griffiths, Journal of Social Psychology, Vol. 48, 1958. |V13 |
|89. "Proposals to Develop Statistical Laws of Human Geography – testing theories of geographic determinism on a world scale," |I35 |
|Proceedings of the International Geographic Union Regional Conference in Japan (in August, 1957), 1958. (U:57-48) | |
|90. "Strengthening Technical Aid by Social Research," PROD, Vol. II, No. 2, November, 1958. (U58-40) |V19 |
|91. "Can We Be Scientific About Humanism?" The Humanist, No. 5, September, (U:58-2) (Abstract, U:57-105) |P34 |
|92. "Formulas for Spreading Opinions," Public Opinion Quarterly, Vol. XXII, No. 4, Winter, 1958-59. (U:55-66, Table U:59-40, Abstract |D6 |
|U:59-24) | |
| | |
|1959 | |
| | |
|93. "The Pollster and the Professor: An Exchange of Letters," The Humanist, Vol. XIX, No. 1, January-February, 1959. |I5 |
|94. "The Reiteration Rule – a cyclic system for syntax, neurograms, and all laws," Synthese, Vol. XI, No. 1, March, 1959. (U:56-143) |I20 |
|95. "An Alphabet of Meanings for the Oncoming Revolution in Man's Thinking," Educational Theory, Vol. IX, No. 3, July, 1959. |SI31 |
|(U:56-107) | |
|96. "A Proposed Barometer of International Tensions," Conflict Resolution, Vol. III, No. 4, December, 1959. (U:57-81) |V22 |
| | |
|1960 | |
| | |
|97. "Word Scales for Degrees of Opinion," with Thomas Gebrick, Language and Speech, Vol. 3, Part 1, January-March, 1960. (U:56-91) |I10 |
|98. "A Comparison of Scales for Degrees of Opinion," with Sung Chick Hong, Journalism Quarterly, Vol. 37, No. 2, Spring, 1960. |P24 |
|(U:59-74) | |
|99. "How Random Interacting Organizes a Population – exploring a simple chance model to relate diffusion theory to information theory," |D13 |
|Synthese, Vol. XII, No. 1, March, 1960. (U:59-145) | |
|100. "Formulas in Linguistics," Science, Vol. 131, No. 3401, March 4, 1960. (Cover letter U:60-74) |D24 |
| | |
|1961 | |
| | |
|101. "Ascertaining National Goals: Project Aimscales," American Behavioral Scientist, Vol. IV, No. 7, March, 1961. (U:60-99) |V14 |
|102. "The Logistic Law of Interaction When People Pair Off 'At dill'," with P. G. Garabedian, Journal of Social Psychology, 1961, No. |D8 |
|53. (U:57-36) | |
| | |
|1962 | |
| | |
|103. "The Logistic Law in Communication," with Marilyn McCurtain, The Symposia Studies Series No. 8 of the National Institute of Social|D19 |
|and Behavioral Science, Series Research in Social psychology, Washington, D.C., September 1961, pp. 1-9. | |
|104. "Can Science Improve Praying?" Darshana, Vol. I, No. 4, October, 1961, pp. 22-37. (U:56-119) |V23 |
|105. "The Concord Index for Social Influence," with Louise B. Klein, Pacific Sociological Review, Vol. V, No. 1, Spring;-1962. pp: |P31 |
|60-64. (U;62-3) | |
|106. "Like Ratings in the Prediction of Human Behavior," with Louise B. Klein, Language and Speech, Vol. V, Part 2, April-June, 1962, |V3 |
|pp. 54-66. | |
|107. "Calendar Reform," Main Currents in Modern Thought, Vol. XVIII, No. 5, May-June, 1962, pp. 111-114. (SD:33-48) |I36 |
|108. "Clique Size As a Factor in Message Diffusion," with Peter G. Garabedian, Sociological Inquiry, Vol. XXXII, No. 1, Winter, 1962, |D11 |
|pp. 71-81. (U:57-59) | |
|109. "The Momental Models for Diffusing Attributes – Dimensional Formulas for Spreading All-or-None Acts Among People in Time," |P21 |
|Darshana, Vol. II, No. 4, October, 1962. (U:56-118). | |
|110. "How Momental Laws Can Be Developed in Sociology – by deducing testable and predictive 'Actance' Models from Transacts," Synthese,|D20 |
|Vol. XIV, No. 4, December, 1962. (5D-62-16) | |
| | |
|1963 | |
| | |
|111. "A 'Doubling and Halving' Technique for Measuring Causation Among Opinions," with Sung Chick Hong, Language and Speech, Vol. 6, Pt.|SD31 |
|2, April-June, 1963. pp. 68-87 (SD 62-5) | |
|112. "Probable Act of Men," Iowa Lectures, in offset Vols. I & II, Department of Sociology, Iowa State University, 1963 |I29 |
|113. "A Measure of Man’s Maturity," Journal of Human Relations, Vol. 11, No. 1. (SD:63-39) |P36 |
| | |
|1965 | |
| | |
|114. "The 'Concordance' Models for Social Control," With Louise B. Klein, Darshana International, Vol. 5, No. 25, April, 1965. |P32 |
|115. "A Far Goals Strategy Towards Resolving Deadlocks and Reducing Conflicts," with Stefan C.; Christopher, Journal of Human Relations,|V21 |
|Vol. 13, pp. 320-21, Third Quarter, 1965. | |
|116. "The Likability Models for Predicting Probable Acts of Men: A Theory of Evaluative Procedures," Systematics, Vol. 2, No. 4, March, |SV1 |
|1965. (SD:63 | |
|117. "The Logistic Diffusion of Information Through Randomly Overlapped Cliques, with Marilyn McCurtain, Operational Research Quarterly,|SD25 |
|Vol. 16, pp. 51-63, March, 1965. (U:59-196) | |
|118. "Rejoinder to Nicosia re: "Logistic Diffusion of Information ...," (with M. McCurtain), Operational Research Quarterly, Vol. 16, |D26 |
|No. 1, 1965. | |
|119. "A Testing of the Modes Theory," with Ronald Anderson, Pacific Sociological Review, Vol. 8, pp. 28-34, Spring, 1965. (SD:64-29 |V4 |
|rev.) | |
|120. "Research Note on 'The Law of Forecast Feedback'," American Statistician, Vol. 19, No. 57, April, 1965. |P20 |
|121. "Two Consensus-Forming Experiments: Demonstrating Transactional Models for Modernizing," Transactions of the International |D28 |
|Conference on the Problems Modernization in Asia under auspices of the Asiatic Research Center, Kobe-University, Seoul, Korea | |
|122. "Letter to Editor re: Solutions for Phone Solicitation Abuses of Polling," Public Opinion Quarterly, Vol. 29, No. 1, Spring, 1965.|* |
| | |
|1966 | |
| | |
|123. "Products Predict Interaction Where Sums Do Not," with Stefan C. Christopher Sociological Inquiry, Vol. 36, No. 1, Winter, 1966, |SI12 |
|pp. 48-60. | |
|124. "Scient-scales for Measuring Methodology – and Rating Scientific Excellence of Researching Behavior," supplemental issue of |I27 |
|American, Behavioral Scientist, July, 1966. | |
| | |
|1967 | |
| | |
|125. "Use Scientific Methods in Planning," Arab Journal, Winter, 1967. |P35 |
|126. "Three Momental Models for Predicting Message Diffusion," Journal of Broadcasting, Vol. XI, No. 4, Fall, 1967 |D14 |
|127. "Models for the Scientific Study of Prayer," Religious Humanism, Vol. I, No. 4, Autumn 1967. |I32 |
| | |
|1968 | |
| | |
|128. "The Reactants Models – Forecasting some probable acts of men," with Stefan C. Christopher, Chapter in Memorial Volume for George |D15 |
|A. Lundberg, Behavioral Research Council, Great Barrington, MA, 1968. | |
|129. "Introducing 'Systemmetrics' for Evaluating Symbolic Systems – 24 criteria 126 for the excellence of scientific theories," |I26 |
|Systematics, Vol. 6, No. 1, June, 1968. | |
|130. "Scientizing Society and Cosmos" – Abstracts and Exhibits of 16 Lectures P42 Sociological Microjournal, Copenhagen, Vol. II, 1968, |P42 |
|pp. 40. | |
|131. "Dimensions of Lundberg's Society as the Foundation of Dodd’s Sociology, a case study of a professional partnership," Sociological |P39 |
|Microjournal, Copenhagen, Vol. II, 1968, pp. 40 | |
| | |
|1970 | |
| | |
|132. "How to Produce Consensus," with S. C. Christopher, in Journal of Human Relations, 1970. |D29 |
|133. " (Republished) "A Power of Town Size Predicts an Internal Interacting," in Social Demography, by Thos. R. Ford, Prentice-Hall, |* |
|1970. pp 146-153 | |
|134. "The Epicosm Model for the Material and Mental Universes," Transactions of the International Congress of Cybernetics, Blackburn |I38 |
|College, Blackburn, Herts, England, London, September, 1970, Gordon & Breach | |
|135. "On Producing the Epicosm Model to Mirror the Cosmos," Transactions of the Joint Conference in Denver, May, 1970, of the |I39 |
|International Society for General Semantics and the Society for General Systems, Gordon & Beach | |
|136. "Deducing a Reiterant Model for the Material and Mental Universes – How the Epicosm Theory Explains and Predicts the Cosmos," Sixth|I40 |
|International Congress on Cybernetics, Namur, Belgium, September, 1970. | |
| | |
|1971 | |
| | |
|137. "Three Causes of-Achieved Consensus Controlled in Two-Person Groups," Systematics, 1971. |V20 |
| | |
|Articles Unpublished as of January, 1971 | |
|138. "Hypotheses Defining Scientific Humanism - a reformulation of humanist principles for testing by scientific methods." |V26 |
|139. "Pan-Act-Theism," Conference of the Society for the Scientific Study of Religion, New York, October, 1970. |V28 |
|140. "Synthesizing: Oneself, Society and Cosmos," Institute on Systems Education, San Jose State College, San Jose, California |V29 |
|141. "The 'Step-and-Parts' Model for Polling – a methodological theory for 29 public opinion research." |P8 |
|142. "The Course Critique Corrector – a feedback system for evaluating course critiques." |P38 |
|143. "A Proposed Campus Poll or Opinion Gauge." |P37 |
|144. "The Likability Theory – a transactional theory of values-in-context." |P28 |
|145. "Modifying Logistic Diffusion by Acquaintance, Sex, and Sampling Factors." |D27 |
|146. Tetramatrix for Modeling Macrosociology." D16 |D16 |
|147. "The Moment Laws in Macrosociology." |D21 |
|148. "Transact Analysis Producing Models for Consensus." |D30 |
|149. "Pathways, A World Game." |V30 |
An Isosceles Distribution of Material Entities
By
Stuart Carter Dodd
Washington Public Opinion Laboratory
University of Washington
Main Currents, Vol. 7, No. 4, Winter 1950
When all material entities in the universe are distributed on co-ordinates of time and space, a regular pattern emerges. The entities range from sub-atomic particles to galaxies of stars and include all living things and human groups. If these entities are defined and classified as described below, then they are found to be distributed within an isosceles triangle (see Figure 1). To see the evidence for this conclusion, the "entities" that fall into the isosceles distribution and the coordinates and units that produce the observed regularity will be presented first.
I. The scales of mass, time, and "rank" of science
Our discussion concerns any material entity in the universe at one instant of time. Entities include quanta, electrons, atoms, molecules, cells, animals, people, plants, planets, stars, nebulas, or the entire universe. A material "entity" is here defined as anything at a given moment which has mass, or alternatively, one which occupies space. An entity may be a single object, a class of objects or a class of classes in ascending more inclusive hierarchy such as a particular gold atom; or the class of gold atoms (i.e., all the gold in existence); or the class of atoms of all kinds; etc. If entities within a class vary in mass, their average is taken to represent the class. Thus the average size of all human groups with their myriad subvarieties is perhaps best estimated as between 2 and 10 persons, i.e., in the class-interval of 100 tons (in Figure 1). The entities are the points, a few examples of which are plotted, within the triangle in Figure 1.
Next, distribute these entities along the horizontal scale of mass from lightest to heaviest. This abscissa in Figure 1 is an enormous scale, one hundred units long, where each unit is a full logarithmic cycle. It is a "centicycle" scale written as 10100 or a "1" with one hundred zeros after it. Its base unit is conveniently taken as a metric ton (though a proton or any other entity would serve). The ton is almost at center so that taking it as origin point makes the simplest equation. On either side of it this cosmic scale extends downward to 10–50 tons into the microcosm and upward to 10+50 tons into the macrocosm. On this vast scale, differences in density of different entities become relatively slight. Thus a correlation of .98 (±.01) was observed between the mass and the size (diameters) of those entities for which these data could be secured. This means that mass and size are roughly interchangeable on this cosmic scale. Each entity was plotted according to its mass as more determinate, but one may also think of it as ranked along the horizontal scale in Fig. 1 according to its approximate size.
Next, distribute the entities along a vertical scale where the ordinate maybe either the "rank" of the sciences or its correlated time dimension. This "rank," called R in Figure 1, is simply putting the sciences into an order. This is the order of their inclusiveness:
| |"Rank," R |
|Social Sciences |0 |
|Human Sciences |1 |
|Biological Sciences |2 |
|Physical Sciences |3 |
|(Astronomical Sciences?) |(4?) |
The entities of each science are included in the entities of the sciences below it as biological entities are a part of all physical entities and human entities are included in biological entities, but not vice versa. (The symbolic logician writes this series of inclusions or subclasses with rigor as Social= human= biological= physical.) Also each science uses the distinctive principles of the sciences below it but not the principles of the sciences above it. Thus biologists will use physical principles of "light," "heat," etc., but not human principles of "purpose," "planning," etc. The principles rank the sciences in a reversed series of inclusions, turning the logician's inclusion symbol around (⊃). Both considerations together confirm the order of the sciences which agrees with the time-honored use in classifying the sciences in a hierarchy as above in universities and in popular thinking.
For an alternative vertical scale, which converges to identical conclusions, time may be substituted for the order of the science. Entities can be distributed by their evolutionary ages, i.e., into five broad periods of emerging in evolution, assigning them to the midpoints of those periods to minimize errors. With the logarithms of a century as the unit and the present as zero point, the five receding periods of the past and their chief characteristics emerge as follows:
Between a century and 10,000 years ago–
Civilization largely developed 1001 +
Between then and a million years ago–
Man largely evolved 1002 +
Between then and a million centuries ago–
Life largely emerged 1003 +
Between then and ten billion years ago–
Solar system was born 1004 +
Between then and a trillion years ago–
Universe existed 1005 +
This time scale correlates highly with the chief characteristics but not perfectly, for life began before a million centuries ago (even though most of its development was since then). Just how high this correlation is will depend both on more exact data and on the definitions of "life," "men," etc.
II. The isosceles distribution of all material entities.
The material entities in the universe, when plotted on these axes, fall regularly inside of an isosceles triangle. We have been unable to find any exceptions falling outside the triangle. Lines 1 and 2 in Figure 1 bound the minimal and the maximal entities at each level (i.e., at each rank or period). This cosmic pattern of regularity is most firmly and exactly fixed by the left-hand straight line on which the minimal entities in each science fall. In simplest form the equation of this line is:
Log M = -11R, the mass-rank relation,1 where M, the unit of mass, is metric tons. (If a "macrogram" = 1011 metric tons, i.e., 1017 grams, and m' is its exponent, then m' = –R.) This gives the following minimal entities:
when R = 0, the mass of a pair of persons, the lightest human group, (the entity studied in the Social Sciences) as in the class-interval of 1 ton ……………………..100 ton
when R = 1, the mass of a fertilized human ovum, the lightest whole human individual (the entity studied in the Human Sciences) as around……………………10–11 tons
when R = 2, the mass of viruses, the lightest living entities (studied in the Biological Sciences) as around. ……………………………………………………......10–24 tons
when R = 3, the mass of an electron, the lightest particle having "rest" mass, (studied in the Physical Sciences). ………………………………………………….…..10–33 tons
when R = 4, the mass of radio photons around the boundary between the lightest quanta and waves, (about where physicists shift from particle to wave terminology)……………………………………………………………………10–44 tons
The minimal entities are neatly aligned. Their (log) differences in mass are all equal. The differences are all in jumps of ………………………………………….10–11 tons
The right-hand line (with slope symmetric to the left-hand line of minimal entities) seems to bound the maximal entities of each science. Its equation is log M + 11R or, if m' is the exponent on mass, m' = + 11R. All living persons and humanity as a whole (at the R = 2 level) and all individuals and classes of flora and fauna (at the R = 3 level) are within this boundary, though not on it. At the R = 4 level it "predicted" in 1941 the mass of the universe as 1074 protons, when only Eddington's estimate of 1079 was known to the author. Professor Harlow Shapley reported to the author in 1944 that his former estimate of 1073 had to be revised (because of the discovery of ten times as much dark as luminous matter), making his best estimate then 1074.
The isosceles triangle, defining this "isosceles mass-time relation" by m = ±11R or by m = ±11 (log T – 1.5), seems highly invariant. It is not altered by:
1. Measuring the horizontal axis in either units of mass or size;
2. Measuring the vertical axis in either units of time or order of inclusiveness of the sciences;
3. Shifting the origin points, size of units, or number of class-intervals;
4. Any finding that the age of "man," "life," etc. is greater than previously estimated. (Since greater ages only depress entities downward more surely within the triangle, making, perhaps, the fit of Line 1 to the minimal entities less exact.)
For some thousands of years still, even though the origin point, the present moment, moves forever onward, the triangle will remain an orderly description of the facts in its field. Beyond that, changing the definition of the top level to choose some later date to bound the beginning of the symbol-using human group, will keep the triangle an accurate semantic device to bring out "law and order" in the cosmos.
III. Implications of the "mass-time triangle."
Among the controversial questions in the diagram is the question of whether it should include the astronomical level E (in dashed lines and question marked in Figure 1). The phenomena of level E are known to us only through radiation as energy while the phenomena of level D are also known to us through molar matter having "rest" mass, such as meteors. But this boundary between pure energy vs. matter seems to some critics to be not as clear cut as boundaries between the higher levels. If level E is excluded, the isosceles triangle becomes an even more accurate picture of regularity, but only within our solar system.
If, on the other hand, level E is included and the ranking of the sciences by the inclusiveness principle is the vertical coordinate, then the correlation of this ranking with evolutionary age becomes much more speculative in level E. Did radiant energy precede any nucleation of it in "vortices" which became electrons, protons, neutrons, etc.? Did matter "build up" out of radiation such, perhaps, as the cosmic rays? The diagram seems to imply this. But we have no evidence that far back into time where even our time scales become uncertain and fade into speculative mist. But physicists prefer not to speculate where no data exists. Consequently we have clearly labeled level E as questioned in Figure 1. Without level E the diagram portrays the isosceles regularity within our solar system; with level E the regularity extends to the universe. If the ranking is the basic ordinate no difficulty results. But if time is the ordinate, the priority of level E is unknown. Partly for this reason we prefer to use the ranking as the firmer ordinate or basic phenomenon the evolutionary ages as an imperfectly correlated variable, or epi-phenomenon.
Another controversial question concerns the accuracy of our data. What becomes of the isosceles regularity if scientists should greatly revise their present estimate of ages or sizes of the entities? If the revisions are to make things older and nearer the midsize (which is one cubic meter, or one metric ton of water) the isosceles distribution, defined by m = ±11R, is unaffected. For such revision keeps the entities inside the isosceles triangle by moving them downward and inward. But if any revised estimates of entities on the boundary lines make them much younger or much lighter (if on the m = –11R boundary) or much heavier (if on the m = +11R boundary), then the isosceles distribution will be affected. The scientist would then modify this isosceles distribution theory to fit the new facts – assuming the revised estimates were different enough and sure enough to justify adjusting the isosceles hypothesis summarized in m = ±11R. To see how this might be done consider the most likely revision – a new estimate of the mass or age of the universe. Suppose the universe is found to be younger than 10 billion years, thus moving it upwards in Figure 1 into the level D class-interval. The scientist in general may react to the new discrepancy by one of three rules:
1. By modifying his definitions of the variables so as to eliminate the newly found inconsistency, or
2. By modifying his statement of relations between variables, i.e., refining his formula, so as to integrate the new fact, or
3. By suspending judgment while he investigates further, searching for deeper harmonizing or synthesizing principles.
In the case in hand by the first rule, the "universe" might be redefined to mean, not its present form if that proves to have been "born" less than 10 billion years ago, but the form it had prior to such a cataclysmic "birth" event. The isosceles regularity would then still hold. Alternatively, by the second rule, the scientist could modify the relation, m = +11R, to another slope of line such as m = + x R where x is such as to make the line bounding the maximum entities at each level go through the new point representing a younger universe. The triangular regularity of distribution of material entities would still hold but it might not be isosceles any longer. Alternatively he may follow the third rule and suspend judgment pending further research. He may say m = ± 11R are only lines of best fit and the fit at one of the nine points of Figure 1 appears to be less close than it appeared in 1950. His further research may lead to an utterly new formulation abandoning Figure 1, or it may lead to a refinement in Figure 1. For one such refinement, suppose the universe were found to be much heavier than 1044 tons which was Shapley's estimate in 1944. This point would be outside the present isosceles triangle if the age of the universe were put at the point of the time class-interval which is level E in Figure 1. But since no beginning of time is known, it is a crude approximation or even an error to date the age of the universe which may have existed in some form indefinitely back into the past. If a finite class-interval E is used, it would be a refinement to place the universe at the bottom of it as at least existing in some form a trillion years ago. Moving the universe pint down to the 1006 line permits estimates of its mass to rise from 1044 to 1050 and still be within the present isosceles triangle.
All these techniques and many others enable the scientist to describe and predict phenomena by means of symbolized regularities which he calls scientific laws when firmly established.
The clearness of the isosceles distribution depends, as usual in science, on the unambiguous definitions of terms and classes consistently adhered to. Here mixed entities such as material culture, or man-made objects can yield confusion. One way to avoid ambiguity is to classify entities by their content in nature as unprocessed by man, as then they can be readily sorted into the physical-inorganic level, the living level, the human level, and the level of the symbol-mediated cultural group.
The validity of this hypothesis, that all material entities are included within the isosceles triangle on these coordinates of mass and rank-in-inclusiveness or age-in evolution, may be tested by searching for a single exception. Is there any entity which when properly classified as to level has a mass outside the limiting lines 1 and 2? None has yet been found. Every alleged exception thus far seems to use other definitions of the terms and hence be inadmissible evidence.
Perhaps the most controversial definition is to consider human groups as one class of entity, rather than as many classes. This entails a single point which is fixed by the rule that wherever members of a class vary in mass their average is taken. This average is probably in the single log class-interval of 2 to 10 persons since the innumerable pairs or trios of conversers are vastly more numerous than larger groups such as nations.
Two considerations dictated this choice in defining the human groups as a single entity. One consideration is that this definition yields law and order – the isosceles distribution – whereas any other spreads the apex of the triangle destroying any scientific law. Scientists in seeking laws, i.e., symbols developing regularity in phenomena, should choose those terms and definitions of them which best achieve this scientific goal. Another consideration is consistency with definitions of other entities. All entities are defined as masses observable at any one moment debarring their history at previous moments lest each mass be unlimitedly repeated. Most classifications of human groups involve the history of persons. For a human group, defined as people interacting in stimulating and responding to each other are mostly latent since at any one moment each person can be interacting with a limited number of others usually in one group and all his other groups are at that moment inactive to him. Classifying groups by regions or institutions, or other bases all involved this "history error" of not sticking to a moment's cross-section of time and hence of counting a person many times. Rigorous consistency in definitions requires a census of everyone's interacting at one moment to observe all the human groups of that moment (taken as a sample of any and all moments). "Everyone interacting with others at a moment" is the one class of group to be observed. Sub-classifying this class into many classes or entities merely means each is a partial or inadequate classification of the field.
This difficulty in classifying so as to avoid reduplicating is highly peculiar to human groups. Man as the time-binding animal lives in many groups in succession within his lifetime or even within each day. For people are simultaneously members of many groups though usually attending to only one actively interstimulating set of persons at a given moment. Duplicating by adding entities together in more inclusive classes as "gold atoms" plus "other kinds" yielding the class "all atoms" is permitted and done at every level in Figure 1 as one moves rightward on it. But duplicating by an alternative classification is a semantic manipulation which combines different symbols for classes without having correspondingly different referents. This "duplication fallacy" partly results from the "history error" in considering groups at more than one moment in time and so violating the definition of material entities here.
The rigorous logic in classifying human groups is difficult to compress into a few sentences here. The reader may try it and submit his results to fellow scientists in all fields whether philosophers and logicians or physicists or sociologists (as the author has done). No other classification than the one finally adopted here (to take all interacting sets of persons at a moment as one omnibus class and average its memberships of these sets) seems rigorously consistent with the definitions and classifying at the other levels.
It is possible to make predictions from the graph as hypotheses awaiting the test of time. These inferences lack any rational explanatory theory at present. But by noting intersections of the various mid-level lines A, B, C, D, and E, in Figure 1, with the sides of the triangle, various highly speculative inferences can be made. One inference was the mass of the universe noted above which remains an interesting coincidence until further facts may make predictions from the diagram more probable. Other inferences are:
1. The maximum number of people possible
i) on our earth,
ii) in the solar system,
iii) in the universe;
2. the maximum mass of living matter similarly;
3. the convenient boundary between our nebula (level D) and what is beyond it (level E);
4. the convenient boundary between particle terminology and wave terminology in the radio band (level E);
5. the convenient boundary for biochemists to adopt between what to call "living" and what to call "non-living" in the range between protein molecules and cells.
Further periodicities may emerge with more research. If a sheaf of lines is drawn through the apex, making equal angles with each other, they may classify entities usefully with hitherto unsuspected regularities emerging. Systematic properties may be discovered as in the Mendeleyev periodic table of the atomic elements. Suggested leads to explore in future research are:
1. The average mass, in level D, of: electrons, atoms, molecules, colloids, crystals...(gap?) ... meteors, satellites, planets, stars, nebulae.
2. In level E, radio photons, visible photons, cosmic ray photons, etc.
3. In level C, viruses, cells, protozoa, chordates, vertebrates, primates, maximal mammals.
4. In level B, critical points in the embryonic development of the human ovum, blastula, fetus, baby, adult.
The final question usually asked is for an evaluation – is the triangle a "scientific law" (if corroborated, as reported here, by other scientists) or is it a semantic device, a diagram merely? The answer depends in part on definition of terms. If "law" in science means man's statement (in some kind of symbols) of a uniform aspect he has observed in phenomena – as scientists are increasingly viewing their laws – then the equation m = ±R is a (candidate) law. The symbols chosen here2 were logs of mass, and of past centuries, or a logical ranking of the sciences, and they developed the simple regularity m = ±R. They "manufactured" it in Eddington's phrase.
Scientists discard old concepts and generalizations and adopt those new ones which create law and order – by means of man's better adapted symbolic responses to his environment. Thus it may be argued: define the lower limit of the "living" the average human "group," the boundary between "our nebula" and the galaxies beyond it, the boundary between particles and waves in relativity physics (if these terms are in controversy) by means of the equation, m = ± R, so as to make a more orderly and law abiding system out of our scientific knowledge.
Figure 1: Isosceles Distribution of Material Entities Diagramming the data of the Sciences of Their Masses Cross-Classified by Their Evolutionary Ages
[pic]
Metric Tons = Mass
[pic]
Metric Tons = Mass
Lines 1 and 2: Log M = ± 11R or Log M = 11(Log T – 1.5)
Data from: W.O. Kermack and P. Eggleton, The Stuff We’re Made Of, Science Book Club, 1938.
Harlow Shapely, Flight from Chaos, McGraw Hill, 1930
Figure 2. A Mass-Time Triangle
[pic]
[pic]
Notes
1. This equation simplifies and corrects its misprinted form in Philosophy of Science, Vol. II, No. 4, October, 1944. The equations there are coordinates of the critical points on the triangle and should have the equality sign replaced by a colon. The editor was not at fault as the proofs never reached the author who was cut off in the Mediterranean war zones, and his correction reached the editor after the journal was published.
The present paper also presents new evidence on this isosceles distribution.
2. This choice of symbols resulted from the author's "S-system" of dimensional Sociology which starts classifying all observables into the four classes of "time," "space," "people," and "all else." A second level of classifying was by their exponents or logarithms. Cross-classifying these classes on a sheet of paper discovered the isosceles triangle. All this dimensional Sociology is developed in the author's Dimensions of Society (Macmillan, 1942 pp. 944) and more simply with more applications, in his Systematic Social Science (Department of Sociology, University of Washington, Seattle, 1947, pp. 785), a typed offset edition for trial and critical revision by colleagues, in advance of publication.
The Scientific Measurement of Fitness for Self-Government
Dr. Dodd is research professor of sociology and director of the Washington Public Opinion Laboratory, University of Washington. He received his B.S., M.A., and Ph.D. from Princeton University and was a National Research Fellow at London University 1926-27. Dr. Dodd served in the U.S. Army during World War II as director of surveys, and as consultant with the Psychological Warfare Branch, Allied Force Headquarters, Sicily.
Reprinted from THE SCIENTIFIC MONTHLY Vol. LXXVIII, No. 2, February, 1954
WHEN are a dependent people fit for self-government or independence? The mandate system and now the trusteeship system are intended to give a dependency either self-government or independence in the course of time, increasingly because of the fitness of the dependency and decreasingly because of international power politics or imperial interests of the control-ling power. To carry out this policy of democratic concern for the welfare of the inhabitants requires that fitness be so specified that the inhabitants know what efforts or achievements by them will result in specified advances toward full self-government. There is need for a procedure to make this transition from dependence to independence a pacific and constructive one and to provide an alternative to the appeal to force that was the sole resource of the dependency in the past. If fitness can be defined and measured and degrees of it set as goals, the trusteeship system is likely to work with less friction and probability of bloodshed in the future.
The Facts to Date
The officially declared purpose in the United Nations Charter and Trusteeship Agreements that the dependency shall be developed toward eventual self-government or independence has taken cognizance of this problem and suggested its solution. The outlook for its solution is made more hopeful by the increased willingness, occasionally evident in the modern world, of some nations under certain circumstances to give up their dependencies and sincerely work toward developing them, not for exploitation, but for the welfare of the inhabitants.
The first step toward the achievement of self-government for dependencies is to set up criteria for measuring their fitness. We analyzed this question to find the criteria that had been agreed upon by both the dependency and the controlling power in the cases of the Philippines, India, and Iraq. There resulted agreement on some eleven criteria, five dealing with internal affairs and six with inter-national obligations of the dependency. These criteria included such requirements as the ability of the dependency to maintain
its civil government,
b) its territorial integrity,
its financial solvency,
a competent judiciary,
a public opinion wanting self-government;
to guarantee the rights of foreigners and minorities and
to abide by other international obligations of debts, treaties, and so forth, legally contracted.
These criteria thus analyzed were the identical sets that were later adopted by the Mandates Commission of the League of Nations. This official declaration of the criteria of fitness for independence is the first step in working toward a solution of the problem of peaceful transition from dependency to self-government.
The analysis was materially aided by techniques developed by the social sciences in the last decade or two for measuring such complex and qualitative phenomena. From these facts a scheme for furthering self-government peacefully may be developed—which is the purpose of this paper.
A Proposed Measuring Instrument for Self-government
How to measure the criteria for self-government.
The instrument here described for measuring fitness for self-government is based on several assumptions which it is well to state, clearly at the start. It assumes that:
a) self-government is often a desired goal;
b) measuring fitness can be a useful step toward self-government;
c) this step can reduce conflicts; and
d) fitness can be measured by suitable statistical indices.
This paper is mostly concerned with giving evidence in support of the last assumption—that fitness can be measured.
Before describing the indices for measuring fitness a review of the theory of measurement on which they are based may be useful for political scientists.
Our theory of measurement, which is applicable to any social phenomena and is not limited to dependencies nor even to governmental phenomena, may be described in three cycles: the qualitative, the quantitative, and the correlative.
This dimensional theory of measurement and the other statistical techniques in this paper are more fully developed in the author's Dimensions of Society (Macmillan, 1942, 944 pp.) and Systematic Social Science (Department of Sociology, University of Washington, Seattle, 1947, 785 pp.), a typescript volume offset for criticism and revision before full publication.
For the qualitative cycle, one begins by naming classes of phenomena and their subclasses in a classification scheme. A subclassification may be carried to as many levels or degrees of fineness as may be needed.
For the quantitative cycle one begins by observing whether a given qualitative class is present or absent in a particular situation. This makes it an all-or-none variable which may be assigned the numerical values of one or zero and then may be weighted later as desired in combining with the other variables. Wherever possible such all-or-none variables should be observed more exactly in an ordinal series, or ranks, such as in stating whether there is none, some, more, or most of that variable in a (riven situation. This ordinal variable in turn should be still more exactly observed wherever possible by converting it to a cardinal variable which is a multiple of equal and standardized units. Ordinal variables are designated "first, second, third . . . ," while cardinal variables are designated "one, two, three..." When the cardinal variable has been calibrated, that is, when its limits, reliability, validity, and so on, have themselves been measured, it becomes the most exact type of variable and is the ideal in the quantitative cycle.
For the correlative cycle the indices of the quantitative cycle, whether all-or-none, ordinal or cardinal, have all their intercorrelations calculated by appropriate techniques. This determines the structure or pattern of the indices or classes of phenomena that are involved in the total situation under study. In proportion as the determinant of the table of intercorrelations vanishes, the system of observed variables can be considered a closed system, since any one variable within it can be completely predicted from the other variables, that is, its multiple correlations with the others is unity. This serves to measure how completely or adequately, for the predictive function of science, the variables in the situation have been observed.
Now apply this theory of measurement which, in its broad outlines, can be shown to subsume all measurement that exists in any science, to the problem of measuring fitness for self-government. The eleven criteria adopted by the Mandates Commission were subclassified through seven levels into a resulting 350 subclasses,' or items, of objective observable phenomena (Table 1). These 350 items were then converted into quantitative indices by observing them either as present or absent or present in ordinal degrees or present in cardinal amounts. Consider, for example, three sample indices at these stages of precision. These three indices measure, for instance, the three major internal criteria of a competent judiciary, a favorable public opinion, and financial solvency.
These items were developed by Felicia Fedorovicz in a Master's thesis under the author's direction
Table 1: 15 Illustrative Types of Indices Selected From the Set of 350 Such Items*
Civil Government (indices 1 to 135)
5. Personnel Administration
(b) Classification
33 Is there a uniform job classification?
34 To what percentage of the civil servants is it applicable?
35 Are the standards of admission to each category defined?
36 Are salary scales for each category established?
37 Are they actually applied?
8. Health and Sanitation
(a) Health Work
71 What is the total score on the "Appraisal Form for City (and Rural) Health Work"? (Adapted from the American Public Health Associations scale)
(b) Personnel
72 What are the per capita numbers of doctors, of nurses, of midwives, of sanitary inspectors, of pharmacists?
II. Military
2. Military Power
(a) Personnel
141 What is the per capita number of military personnel?
142 What is the percentage of native personnel (senior officers; line officers; junior officers; staff officers)?
III. Order and Security
1. (b) Public opinion on security
161 How many times in the last month have you heard of a . . . (each crime below in turn) ... in this community (village, tribe, etc.)
% hearing Mean number of
in sample rumored crimes
Theft ________ _____________
Cattle theft ________ ________
Burglary ________ ________
Rape ________ ________
Murder ________ ________
IV. Financing
4. International Obligations
196. What is the proportion of per capita foreign debts to per capita yearly national income?
V. Judiciary
1. Impartiality
201 Are there any special privileges for religious, racial, economic, or other groups?
202 Are there any foreign concessions?
203 Are the court fees reasonable or are the costs prohibitive for the poor?
VI. Public Opinion
3. Political maturity
(a) Activities
281 What percentage of the electorate voted in the last national election? Municipal election?
For a qualitative index converted into an all-or-none quantitative index, consider the criterion that the dependency must possess a judiciary commanding the confidence of the people. Alongside the courts of the controlling power, let parallel native courts be set up with alternative jurisdiction as formerly was the case in various Near Eastern states. Let the native courts try those cases where both litigants agreed to take their case there. At the 2nd of each year, the percentage of all the cases in the court that went to their own native courts could be calculated and judges appointed for the next period, such that the native judges would be that percentage of all the judges. Thus, as people gained or lost confidence in their native courts, the percentage of native judges in the country would increase or decrease. When they attained full confidence in their own judiciary, all judges would automatically become nationals; and the dependency would have achieved complete control of its judiciary by demonstrated confidence in it. The action of taking the case to the native court is a qualitative item of behavior. Coupled with its absence, the not-taking-the-case-to-a-native-court becomes an all-or-none item of behavior. The arithmetic mean of such items is the percentage of cases taken to the national court. (Every percentage is the arithmetic mean of an all-or-none frequency variable.) For an illustration of an ordinal variable, consider a public opinion poll determining the criterion that there shall be a predominant public opinion in favor of self-government or independence. Here, instead of asking the question, "Do you approve or disapprove?" the question could be asked more exactly in ordinal degrees specifying the degree of self-government that is favored or as the degree of intensity of approval or disapproval of self-government. Of course, such a question should be care-fully phrased in alternative forms, penetratingly analyzed by supplementary questioning to make sure that the opinion expressed on the question at issue is adequately reflected in all its complications and is independent of the particular phrasing of the question. The questions should elicit the "give/get" ratios or price of independence which the citizens are willing to pay. This cost of what they want can be measured in eight types of indices that we have developed.' Thus indices of willingness to pay higher taxes, have military service, and put up with possible specified degrees of lowered efficiency in government services, can gage the strength of the public opinion in terms of what the people will give for what they want to get.
For a third illustration, this time a cardinal variable, the financial solvency of the budget can be determined in cardinal units of money.
For the correlative cycle, the 350 indices need to be combined in successive stages until eventually they yield a single index or scale of total fitness. This combination is usually made additively with some kind of weights. Ideally, the weights should be regression weights determined by multiple correlations, which is calculated by correlating the indices against some accepted measure of dependence-independence such as the difference between a set of dependent areas and another comparable set of in-dependent areas. As a first approximation, these weights may be secured as ratings by a panel of competent specialists.
For this weighting, a decimal system has been developed that distributes 100 points among the subclasses of each class wherever it may fall in the hierarchy of classification which breaks "fitness" down into subclasses and sub-subclasses in successive levels. The number of percentage points given to each subclass is according to the judge's estimate of its relative importance among the other sub-classes of its class in contributing toward fitness for self-government. In this way the weights of all the subclasses of any one class add up to 100. Provision may be made for an extra unidentified subclass in the event the judges feel that the given specified subclasses are an inadequate analysis of their class.
The weight assigned by the judges to the unidentified class measures the degrees of inadequacy of other subclasses. With this decimal weighting item, the net weight of any class, anywhere in the hierarchy on any of the seven levels, toward total fitness is simply the product of its percentage and percentage of all classes above it in the hierarchy. This yields indices for every item or combination of items up to the total fitness so that the whole, well as any of its parts, is measurable. This decimal weighting system is flexible in permitting the insertion or removal of any items above or below it, with consequent shift of weighting only in the subclasses.
It remains to be seen whether the ratings would transcend the raters, that is, to what extent the ratings assigned by different panels of judges with different ideologies would differ from each other or give a constant result. Thus, in constructing attitude tests it has been found by Thurstone that panels of judges with very different attitudes themselves will yet agree on rating the meanings or scale positions of statements of attitude so that the attitude scales they construct transcend or are independent of the attitudes of the judges who constructed hem. If it should be found that the ratings do not transcend the ideologies of the raters and that they yield significant differences, then the scale could still be applied but with specified weighting systems, for example, Communist weightings, Republican weightings, or a UN Trusteeship Council weightings system.
The reliability of this weighting system has been experimentally measured. Shannon3 secured weightings from social science faculty and student groups. He found no significant differences in mean weights when he subclassified his data into expert raters and inexpert raters. They correlated 0.9 together (± 0.008). The ratings differed significantly from chance, which would assign equal weights in every subclass in a class. On repeating the ratings, correlations of ratings of individuals between first and second ratings ran as low as 0.41, but correlations of the mean first ratings and the mean second ratings ran up to 0.997 (± 0.002). The split-half correlation of odd versus even items was 0.93 (± 0.-006). These studies proved that the weights could be highly reliable if they are the mean weights of a panel of judges.
(b) Reliability, validity, and norms. The next steps still to be taken in constructing the scale of fitness are to determine its reliability, validity, and norms.
Reliability can be determined for one index of it as the correlation coefficient between one determination of the scale and a second determination of it on presumably the same phenomena, as when one set of a person's determination of the indices is correlated against another set's determination of the same indices. The more the two sets of observers represent different viewpoints or cultures, the more their degree of agreement will establish the objectivity or rerating reliability of the scale, of course. To the extent that this correlation is high, say above 0.9, the scale is judged reliable; that is, it is free of errors of observation due to the observer. Many of the indices are fixed by censuses, school, tax, budget, and other statistics that would not vary much if reobserved. But other indices which are mean ratings by experts or polls of samples of the population might vary more on re-observing them. Hence their degree of reliability must be experimentally isolated and measured. '
Validity is definable as the correlation coefficient between the scale and an accepted index of dependence-independence. How high does the scale, when applied to a group of independent and a group of comparable dependent areas, correlate with the all-or-none variable of independence or lack of it? The higher this correlation, the more valid the scale, that is, the more closely it is proved to measure what it claims to measure in society. At present the validity of only a few indices has been explored. Shannon} has found. correlation scattergrams with correlations in the 0.2 to 0.5 range between dependent or independent states as the one criterion variable to be predicted and illiteracy, gold reserves, and other factors as independent or predictor variables. These preliminary findings suggest that by multiple correlation techniques a scale can be built and proven to measure more accurately the degree of independence.
For this purpose of determining the norms, a "para-nation technic" is proposed. It may be sketched here in oversimplified summary; eventually the Trusteeship Council might work out details. Let the degree of fitness demanded of a candidate dependency, in order for it to qualify for full self-government or independence, be some preassigned amount such as the average degree of fitness of ten para-nations, which are nations comparable in as many respects as possible to the candidate area. Let these nations be measured by the scale, arid their arithmetic mean determined. Call this point "100 percent fitness" for the candidate nation (Table 2). This means that the dependencies must be more fit than the less competent independent nations of its class, yet need not measure up to the most competent, but only attain the average.
With this 100 percent point on the scale of fitness determined for a given nation, subdivisions of it, when reached, could determine when that candidate would reach specified increments of autonomy Thus, it could progress gradually toward self-government by peaceful, rational, and predetermined steps.
Application of the Measuring Instrument
a) Possible administration of testing. In order to try out the extent to which a measuring instrument can contribute to the peaceful evolution of a dependency toward self-government, the instrument must be applied and developed further under field conditions. Administrators of dependent areas must become more interested in applying social science techniques to their practical problems. It is to be expected that such interest will grow sporadically but may be encouraged by clear statement of the goal and available means. The goal here would be a controlled experiment for testing the hypothesis implied in the second and third assumption above.* This fitness scale hypothesis might be stated as: It is expected that in proportion
(1) as a people's fitness for self-government is well measured;
(2) increments in fitness are pre-announced as the precondition for increments in self-government; and
(3) increments in fitness are at least partly attainable by the people's own efforts then, self-government is likely to be achieved sooner, with less conflict, and with more satisfaction to most people than without this measurement and motivation plan.
The proposals here may seem a little naive to some political realists. The author recalls, however, proposing In Geneva in 1927 that criteria for terminating a mandate should be specified. Sir Austen Chamberlain replied that that was a job for the academic people before political administrators could move. The fact that when criteria were then worked out with a fellow academician (Professor Ritsher) these were the ones adopted by the Mandates commission, gives hope that now in measuring these criteria the statesmen may again find the social scientists’.
In order to test this hypothesis scientifically, controlled experiments would be designed, ideally, but in practice only crude approximations may be possible. These may be less crude, however, if the administrators have the plan of a controlled experiment in mind, even though they may lack the power to execute it cleanly. A controlled testing of the hypothesis implies finding two similar or matched dependencies. One of them would then have the aforementioned measurement and motivation plan applied to it while the other would not (or at least would get much less of it.) In a few decades the satisfactory and peaceful progress of the former toward self-government should be visibly greater than in the latter, if the hypothesis holds.
(b) Possible administrators of testing. There are four types of authorities, each of whom could con-tribute a part to the full testing of the hypothesis These authorities are the United Nations Trustees, national colonial offices, bureaus for research or government, and political scientists with statistical training.
(1) Trustees of United Nations trusteeships should be the first to apply the hypothesis and its implementing scale to their trusteed lands. Thus a technic is offered that can help fulfill the official purpose of trusteeships—to prepare the trusteed lands for self-government or independence. It offers a substitute for agitation, rebellion, and the appeal to force which have all too often in the past been the only course open to a dependent people wanting their freedom. It offers an evolutionary method of gradual growth toward self-government in place of revolution. It provides a definite constructive procedure which can motivate the inhabitants to those efforts and achievements that fit them best for taking over the management of their own affairs. In short, it offers a way to make the trustee's task easier. Of course, this fitness is not the only consideration in granting self-government or independence. Considerations of strategic defense and other self-interests of the trustee may still play a part, but the measuring and isolating of fitness removes the question of fitness from the realm of controversy and makes it less possible for any trustee to rationalize imperialistic policy by the claim that the trusteed land is not yet fit for autonomy Above all, the scheme would stimulate the development of fitness as a conscious, implemented goal.
This is a bit of the scientific method applied to administration, namely, the isolating of factors so that, with the causes and effects of each factor separately known, any one factor can be controlled even when other factors are still uncontrolled.
Trusteeships should begin this process o planned, peaceful transition to self-government by writing an article to this effect into the charter that set them up. The principle of measured and motivated fitting of oneself for self-government could be established by some such article as the following:
In order that the inhabitants may be constructively guided in their efforts to fit themselves for self-government by peaceful steps and without recourse to force the Trusteeship Council shall advise and recommend to the Administrative Authorities as to the procedure for determining: the inhabitants' current degree of fitness for self-government, and what efforts or achievement by them, and what other conditions, are considered prerequisite to specified increments in self-government.
2) Farsighted colonial administrators can use the fitness yardstick to simplify their problems. Nationalistic agitation tending toward demonstrations involving bloodshed can be headed off by agreement, at an early stage, upon fitness goals with the representatives of the inhabitants when such agreement is more possible than later when nationalist emotions and demands go to the extreme. With political, economic, educational, and social development mapped out in definite detail and the energies of the nationalists harnessed to cooperate in achieving them, the colonial administrator will have intelligently solved his future problems in large part by preventive action.
Considerable progress in this direction has been made. The Trusteeship Council has prepared elaborate and searching questionnaires for securing detailed information from the trustees. But the motivating of effort by the nationals is little developed, since what they must achieve to earn specified increments in autonomy is not stated definitely and in advance.
3) Bureaus for research on government and governing could test the hypothesis and also expand it. They should research to refine the scale of fitness toward making it a more realistic and exact instrument for prediction and control—the functions of science, whether political science or any other. Further, they could expand the scale from government of dependencies to any governmental unit as many of the indices would be equally pertinent in a self-governing or a non-self-governing area. By appropriate choice of indices, variant versions of the scale could measure the excellence of functioning of any governmental unit.
Indices of municipal political functioning could be developed and synthesized into a scale for the efficiency of municipal government. Indices of county or provincial or state governing could be collected more systematically than hitherto by the technics developed here to form scales for county or state government. Appropriate indices of national or international government could eventually yield scales for measuring the excellence of national and world governing. The method of scale construction, based in part on our theory of measurement, is applicable to any size of governmental unit. It may well be that self-government is best developed on a local scale first and later extended to larger units.
This method of scaling is also applicable to any department or function of government for which separate diagnosis and treatment may be wanted. Subscales for administrative, legislative, judicial, or party functioning, or any of their subfunctions, can be progressively developed by research. Bureaus
Table 2: A Schematic Distribution of Nations P Fitness Scores
(x = 1 para-nation)
Number of x
nations x
x x x
x x x x
0 50 100 150 200 250
Fitness score, a weighted average of 350 indices
mean
score
Range taken as standard
and divided into percents
0% 100%
totally % fully
unfit of fit
point fitness point
for research on the government process to greatly improve that process in the long systematic research to measure it, so that ratings of excellence can be factually determined diminishing controversy and frustration for the voter.
Political scientists, authorities on government more than in government, also should be helping to prove this hypothesis. They should get beyond the description of government and achievement to that prediction and control that science can bring. Toward this end, more exact generalized laws of political behavior are required, and exact observing involves a precision in things such as our scales of government. To the extent that political scientists expand the indices for self-government to measure the excellence of any government, the resulting governance can help to make political science an exact science.
With such ends in view, this paper is intended to do more than report on the construction of a scale of fitness for self-government. It is also a blueprint for further quantitative research in political science. It is an invitation to any political scientists to cooperative research in further developing the government scale and a more exact science of government.
References
1. Ritsher, W. H. "Criteria of Capacity for Independence," Social Science Series, p. 152. Lebanon American University of Beirut, 1934.
2. Dodd, S. C. American Sociological Review 16, No. 5 (1955).
3. Shannon, L. "Measuring Capacity for Self-Government." Unpublished M.A. thesis, University of Washington, 1947.
4. Shannon, L. "The Non-self-governing Political Entity." Unpublished Ph.D. thesis, University Washington. 1951.
Techniques for World Polling
A Review of the Methodological Literature for Cross-Cultural Surveys
By
Stuart Carter Dodd
Washington Public Opinion Laboratory,
Seattle U.S.A.
I. Planning for International Polling
In 1953 UNESCO contracted with its Non-Government Consultant Organization, WAPOR (the World Association for Public Opinion Research) to survey the journal literature on polling methodology. The resulting 600-page report is previewed here in some 6,500 words. To keep this paper within limits of a journal article its many pages of bibliographic references are omitted here. The reader is referred to their fuller citing and reporting in the chapters of the volume corresponding to the sections of this paper.
The long-range objectives of this research contract were to further the development of a Barometer of World Opinion in some form. Such a barometer would periodically poll a representative sampling of an ever-increasing part of the world's population on as wide a range of topics about man's chief institutional fields of interest as might be currently desired and financed. It could measure more and more fully as time went on the pollable opinions of mankind — their asserted aspirations and needs, their feelings and degrees of satisfaction, their knowledge and expectations, their personal and public or organized actions. For UNESCO an opinion barometer could mean early measurement of tensions in the minds of men where wars begin and guidance for preventive steps. For the behavioral scientists a barometer of world opinions could mean an instrument for observing mankind as a whole, transcending all present limitations to fractions of humanity and fragments of human culture.
The short-range objectives were to survey cross-cultural methodology in order to increase its comparability or standardization. The research contract was expected to help standardize polling techniques internationally. It should tell how pollers, by standardizing their instrument, are improving the measurement of opinion as to its five chief dimensions, namely:
its universality when done in any country (in the space dimension)
its reliability when repeated at a later date (in the time dimension)
its validity when applied to any kind of speech behavior (in the activity dimension)
its representivity when sampling all kinds of people (in the population dimension)
its utility when directed for any goals or purposes (in the values dimension)
Towards these objectives Project Demoscopes was expected to report to UNESCO, within two years, upon a bibliographic search of the literature on polling techniques, for international use especially. The search had to be limited chiefly to articles from two dozen journals during the past thirty years — though some books and unpublished materials supplied by pollers were also abstracted and covered in the review. An Advisory Committee of sixteen eminent pollers from eight countries and a panel of fifteen assistant editors writing signed sections were organized on a volunteer basis. With an assistant director, Professor Jiri Nehnevajsa, and a staff of four graduate students the bibliography of 1,103 titles was selected and the larger part of it abstracted onto form sheets. These were then written up following a detailed outline of some 140 topics, each of which had been rated by the Advisory Committee. The twelve chapters of the finished report, entitled Techniques for World Polls, are reviewed in the twelve sections of this paper.
II. A Matrix Model for Standardizing Polling
In order to systematize both the survey and the field of polling methodology, a theory of polling was developed in Chapter 2. This theory is presented in the form of a matrix model with 192 cells. Each of these 192 cells, or 'step-parts' as they are called, attempts to standardize one source of instrument variation and so eliminate one kind of error. For every set of data from any poll can be analyzed as a product (and not as a sum!) of two factors (not two addends). Every set of polled data (or observations in any science, for that matter) is always some kind of logical, mathematical and empirical product of
1) an instrument factor whereby we observe and
2) the phenomena observed.
The phenomena in polling mean the true opinions or speech behavior of respondents if unbiased by the standardizing interview situation. The basic hypothesis that is assumed for testing in controlled experiments in this 'step-parts' model, or operationally defined theory of polling, is expressed in the equation:
A = a X α Eq. 1 the methodology equation for all science.
This says in effect “the observations (= A) are always a product of the observing (= a) and the observed (= α)” or more simply here “Polls can be factored into polling and opinion or into research acts of pollers and speech acts of respondents.” Each of these three variables is a composite, a set of many elements. The kind of resulting product may be a logical product of classes, or a conjunction of sentences in Symbolic Logic, or an intersect of sets in the algebra of sets, or an arithmetic or algebraic product, or a product moment of statistical variates, or a matrix product, or other kind of product appropriate to the form of the data — but not the corresponding kind of sum (i.e., A ≠ a ± α).
This subhypothesis that products predict better than sums (under conditions specified by a 'vanishing result' test) can be verified by controlled experiments at any time. Thus the Washington Public Opinion Laboratory has shown that in spreading an opinion (or other item or all-or-none state) from person to person with equal opportunity for everyone, the percent of persons contacted or diffused will necessarily grow in an S-shaped logistic curve. If the interactors (who are either knowers (= p) or non-knowers ( = q) of the item, so that p + q = 1) are multiplied together to get their product, pq — which expresses the joint probability of their interacting in meeting each other — these products cumulate in t successive periods to fit the observed logistic curve excellently. The data and this logistic hypothesis (expressed by pq) showed a correlation of .99 between their in-crements; and the discrepancies were not significant by the chi square test at the five percent level. Then the identical data were added together to get their sum, p + q — which expresses the alternative probability of their separate occurrence. This sum between a constant at unity correlated necessarily at zero with the observed data of spreading the item by conversations in pairs. Thus merely multiplying versus adding identical data results in either almost perfect prediction of the observed and re-observable facts or the worst possible prediction such as pure chance gives. This repeatable classroom experiment is typical of the way in which our step-parts theory can be rigorously tested. Each of its 192 step-parts is an hypothesis as to a source of error that can be tested in controlled polling experiments. This Chapter 2 of the report maps out hundreds of research projects for pollers and graduate students' theses for many years ahead. These should systematically improve our polling tools, our demoscopes, until they match the accuracy in specifiable respects of the best microscopes or telescopes in other sciences.
To develop this step-parts model, the polling process was first classified into eight 'stages' or standard subprocesses. These are:
1. Administering a polling agency, including organizing it;
2. Designing a poll, including all administrative planning for it;
3. Questioning, including development and pretesting of questionnaires;
4. Sampling, including preparation, execution, and evaluation;
5. Interviewing, including recruitment, training, supervision, and evaluation;
6. Analyzing, including all editing, tabulating, and statistical computing;
7. Reporting, including all recording and filing as well as publication;
8. Interrelating the parties, including pollers and sponsors, respondents and publics.
These eight stages are reviewed in eight chapters in the report. They form major rows in the step-parts matrix shown in the Conspectus attached (Figure 1).
Figure 1
|The Variables in the Model: |Every poll (A) is analyzed into 2 factors, i.e., the public’s opinion in the α and the poller’s procedures |
| |(a): A = α x a. |
| |The opinions (i.e., speech reactions in a poll) proposed as behavioral criteria (B) for testing this model |
| |internationally are: assertions (as in a census) of age, sex, schooling, income, occupation, nationality, |
| |religion, and languages spoken. |
| |The procedures of polling (i.e., role-actions of pollens) are analyzed here into 8 stages and 24 substages |
| |called "steps" in polling. Each step may have 8 type parts (which are the dimensions of any human behavior) |
| |yielding 192 "step-parts" in the cells. |
|Its Central Hypothesis: |Standardizing the step-parts improves the comparability and validity of polls. |
|Its Testing: |Reliability (rAA’) and validity intraclass correlations (rAB, BA') from repolling the criteria' opinions, |
| |with each step-part in turn varied alone over poll and repoll values, can test how standardizing may improve |
| |validity. r AA’, A’A = 1 if σa = 0 and r BA. = r BA',AB |
|Stages or |Sub-Stages or Steps |Eight Type-Parts |
|periods | | |
| | |Acts |Actors |Values (wants, goals, desider-ata) |
| | |(Behav-ior|(pollers, |V |
| | |) |pollees, |Why? |
| | | |publics) | |
| | | | | |
| | |A |P |jj3 |
| | |What is |By Whom? | |
| | |done? | | |
| | | |jj2 | |
| | |jj1 | | |
| |Executing 22j | | | |
| |Completing 23j | | | |
|Question-ing |Preparing 31j | | | |
| |Executing 32j | | | |
| |Completing 33j | | | |
|Sampling |Preparing 41j | | | |
| |Executing 42j | | | |
| |Completing 43j | | | |
|Interview-ing |Preparing 51j | | | |
| |Executing 52j | | | |
| |Completing 53j | | | |
|Analyzing |Preparing 61j | | | |
|statistically | | | | |
| |Executing 62j | | | |
| |Completing 63j | | | |
|Reporting |Preparing 71j | | | |
| |Executing 72j | | | |
| |Completing 73j | | | |
|Interrelat-ing all parties |Preparing 81j |831 |
|Any instructions to interviewer in light italic type | | |
VII. The Sampling Stage
The enormous literature on sampling could not be adequately reviewed within the manpower resources and space limitations of the report in this survey of methodology for cross-cultural polling. Our limited canvass, however, justifies, we believe, the following five generalizations about sampling techniques for world polling:
1. Sampling consists in selecting and using a part to represent some whole.
2. The 'whole' in polling usually means a specified human population but may be measured along any other dimension such as in time sampling, areal sampling, activity sampling, and opinion or attitude sampling (as when testing a sample of statements for their scalability by the Cornell technique).
3. Representativeness of a sample is mathematically best assured and measurable by probability sampling, especially when both the sample and its 'universe' are large. The representativeness of quota sampling is usually less or unknown. Availability sampling is still worse.
4. Probability sampling's exactness is somewhat offset by its greater expense, administrative difficulty often, and possible biases if imperfectly executed.
5. Research is progressing, but much more is needed, towards combining the accuracy of probability sampling with the ease and lower cost of quota or other sampling.
A few items of this research may be glimpsed. Kendall and the British pollers have worked on combining the best features of each. Deming's new duplicate sampling technique (reviewed in this chapter) greatly reduces computational administrative costs in probability sampling. The technique of weighting respondents in proportion to the number of evenings they were at home in the preceding seven days helps correct for the `not-at-home' bias in probability sampling. Stapel's 'birthday' technique of asking to interview the person with the nearest birthday randomizes the choice of interviewee within the household more simply than by using Kish's tables for this. Other techniques for better and easier sampling include development of: more complete population lists, maps showing every dwelling unit, previous master samples available for purchase, tables of random numbers and other mechanically randomizing devices, etc.
VIII. The Interviewing Stage
The review in Chapter 8 of the journal articles (and a few books) on the interviewing stage in a poll is broken down into the three substages, and each of these is further subclassified by the eight type-parts. The fullest way to picture this chapter's review, when it has to be further condensed in this paper to a paragraph, is to reproduce the table of contents for the descriptive part of this chapter. (Its author, Peter Mazur, adds an evaluative part giving his appraisal of the literature on each substage.)
Chapter Section Type-part
I. Preparing the interviewers (Pre-stage)
A. Setting qualifications for interviewers Acts and values
B. Recruiting suitable interviewers Population
C. Preparing instructions for training interviewers Symbols
D. Using interviewer equipment and calculating costs Material
E. Choice of a suitable place for training Space
F. Training interviewers in time Time
G. Other conditions Circumstances
II. Conducting the interviews (Mid-stage)
A. Setting standards for an interview Acts and values
B. Matching interviewers and respondents Population
C. Keeping interviewer records Symbols
D. Using material means in an interview Material
E. Placing the interview Space
F. Scheduling the interview hours and days Time
G. Other circumstances in an interview Circumstances
III. Evaluating the interview (Post-stage)
A. Measuring fulfillment of standards Acts and values
B. Measuring incompatibilities among interviewers Population
C. Researching on semantics of interviewing Symbols
D. Measuring costs of interviewing Material
E. Measuring spatial factors Space
F. Measuring duration of interviewing Time
G. Appraising the context of the interview Circumstances
IX. The Analyzing Stage
The two authors of the report, Techniques for World Polls, followed the ratings of WAPOR'S Advisory Committee as to the relative amount of reviewing space to be given to each step in the polling process. The consequent abstracting of the vast literature on statistical analysis was limited here to what might be most useful and distinctive to cross-cultural polls leaving the international poller to read up on most of the statistical techniques in the many available textbooks and manuals.
For cross-cultural purposes, the chief techniques are the editing of filled-out questionnaires or schedules, the coding and preferably pre-coding of responses, their transfer to punch cards (or edge-marked cards for hand tabulating), and the tabulating of the frequencies of response in as many breakdowns or subclasses as desired. Once the data is on a punch card, it is in an entirely international statistical language whatever the idiom of the interview. The commoner statistical analyses such as getting any desired indices like percents, other means, variances, raw and multiple correlations, chi squares, factor loadings, tests of significance levels, of confidence bands, of excellence of fit, etc. are all completely independent of the language of the questionnaire or the culture of the respondents.
A note may be made of the coming mechanical and electronic instruments which will vastly increase the volume, speed, ease, and accuracy of international polling especially (where a world sample may run up to 100,000 interviews eventually). Thus one unpublished proposal blueprints a wire recording device the size of a cigarette case. It would be operated by the interviewer's fingers of one hand punching in the responses as heard. It would flag omitted questions before the interview was over, thus editing out such gaps. A thousand items of .response could be recorded as magnetized or non-magnetized points a millimeter apart on each person's meter of wire — which would be equivalent to one punched card. One hundred spools could unwind 100 parallel meters of wire on a scanning table every second. This would yield tabulations of 100 cases per second in each of 1,000 frequency counts which would be rung up in 1,000 counters. The readings of the bank of 1,000 counters would be photographed every second recording all data cumulated to date by successive samples of 100, permitting quick testing of reliability. With I,000 'electric eyes' reading the 1,000 magnetized points through 'scoring screens,' any selecting and combining of data becomes possible for item analyses, personal scores, distributions, scattergrams, curve fits, factor analyses, variance analysis, etc. A thousand items of response could conceivably be tabulated for a million persons in one day by one machine. If a world census were taken on one day (as some countries now do it), it could be all tabulated the next day and printed on the third day! A future society that wants to predict and control its affairs most fully could organize and pay one randomly sampled person in every thousand to work as a paneller for ten minutes a day. Each paneller would spend his daily ten minutes recording into a telephone his 'statistical diary' consisting of his responses to a questionnaire on some hundred items of diverse behavior each of which he did or did not do that day. Central machines could then tabulate, for the world or for any sampled fraction of it, each of as many thousand human actions or attitudes as desired. Each of these or any combinations of them could then be mechanically translated into daily forecasts for any periods ahead. All this is now physically possible whenever society wants it enough to pay for it. The cost might be a small fraction of the billions of dollars now paid for military armaments.
X. The Reporting Stage
The reporting of polls is subdivided in Chapter I10 of the report under review here into its three substages, or steps, by tense, as planning the reporting, writing it, and publicizing it.
The reporting should be planned to inform its readers and so its language and length will vary according as it is addressed in the appropriate national language(s) to the general public, special publics, or persons technically competent in the content and/or methodology. Often a three-level report is useful for mixed audiences – telling its story in an abstract first, then more fully in the body of the report, with technical appendices of details for specialists.
Reports of polls are increasingly being written in four versions woven together. In order to communicate most fully to most people, reporting in sentences is necessary; numbers and tables supplement words with greater accuracy; graphs make these easier to see and remember; cartoons or pictured meanings may make reports still better understood and believed. The prose should be objective in style and operational in its definitions, permitting repolling and verifying. It should tell what it wants to say in well-known and oft-used short words – as proved by a words-per-syllable rate of .6 or more. (This ratio is .93 in the last sentence and .5 in the previous sentence, for example.)
A set of standards for reporting polls, as endorsed but not adopted by AAPOR in 1948, is:
Poll reports should state:
1. The purpose of the survey
2. For whom and by whom the survey was conducted
3. The universe investigated
4. Size and nature of the sample
5. Time spent on field work
6. Nature of the interviews
7. Control methods used
8. The phrasing of the questions
9. The bases of percents
10. The distribution of responses
For publicizing reports on polls the chapter gives a list of ways. The media for publication, such as news releases, mimeographs, printed articles, books, are all used, of course, but these are not peculiar to polls.
XI. The Interrelating Stage
The last 'stage' of interrelating all the parties to any poll is not so much a final process as one which pervades all the previous stages and is only discussed last in summary. For interrelating the parties means much more than the usual concept of public relations. It includes the two-way communicating and cultivating of attitudes and actions in all combinations among the following six parties. The six parties, directly or indirectly related to a poll, are:
1) the sponsors who start a poll,
2) the pollers at all levels in the polling agency which executes the poll,
3) other pollers as competitors or fellow professionals,
4) the respondents in the sample,
5) the public that is sampled, and
6) the publicists (press, broadcasters, etc.) who may mediate between any of these.
A seventh party in the future may be poll auditors who on request will investigate with appropriate measuring rods the honesty and competence of any poll or pollster and certify their findings to the public much as chartered accountants do for a firm's financial bookkeeping. Such accrediting may become available and much utilized in international polling especially.
The chapter reviews some principles of public relations in polling, a list of 'effective policies' for developing public confidence in a polling agency, and the rudiments of a proposed code of professional ethics for pollers.
XII. Recommendations for Developing World Polls
The agreements between UNESCO, WAPOR, and the Washington Public Opinion Laboratory as subcontractor for this survey of polling methodology specified that in the volume reporting it the authors' evaluations should be clearly separated from their descriptions of the literature reviewed. Accordingly, the last chapter in the report presents the authors' personal evaluations in the form of recommendations for action. These recommendations are grouped by the chapters of the report. They are variously addressed to UNESCO, to WAPOR, to pollers, to sponsors of polls, to publicists, to graduate students, and to the public. They add up to pointing out steps for relevant parties to take whereby polling and international polling especially can be developed as a more exact science and as a more useful tool for society.
For the specific recommendations the reader is referred to the report, Techniques for World Polls, which will be in UNESCO'S files.
The present paper is a preview or appetizer for that review of polling techniques.
Notes
1. The dimensional formula [denoted by square brackets] specifies the exponents or level of development by self-multiplication for each type-part of any transact (= B) or behavior-in-situation thus:
[B = Aa • Pp • Tt • Ll • Vv • Mm • Ss • Cc] Eq. 2
The statistical formula is expressed in our interdisciplinary and standardizing notation by corner scripts. In addition to the exponents these four corner scripts also specify the origin and other points, the units and indices by means of scripts on the other corners of any variable, thus:
units s s exponents
X
points s s indices
Dimensional formulas are useful for generalizing as in classifying data, formulas, and situations into families of dimensional models expressing hypotheses and laws; while statistical formulas are useful for particularizing as in specifying local details of terminal points, class-intervals, and indices of each variable as used in a particular study.
The Logistic Diffusion of Information through Randomly Overlapped Cliques
Stuart C. Dodd and Marilyn McCurtain
Department of Sociology
University of Washington
Seattle, Washington
Operation Research Quarterly, Pergamon Press 1965. vol. 16, pp. 51 to 63. Printed in Great Britain
The logistic model expects: If an item is told and heard in a set of people and periods under conditions of steady, pairing off, with equal opportunity, then At = kptqt, i.e. increments in knowers are proportional to joint probabilities.
Unequal opportunities result (among many causes) from "clique effects" where people communicate more within their daily circles of contacts (in homes, work, transit, eating and in leisure or other activities) than between such cliques.
We hypothesized that if cliques are randomly overlapped in membership, then as clique size increases from 2 to all N persons, the diffusion-retarding effect of clique barriers will tend to vanish. "Larger cliques accelerate diffusion."
Previous experiments confirmed this hypothesis up to 4-man cliques and contradicted it thereafter, as larger cliques exceeded logistic expectations, and did so systematically. This was due to a constraint of "seeking out non-knowers" which was eliminated in the present experiments.
With three replications from a population of playing cards, simulating increasing sizes of cliques, the curves rose steadily towards the logistic as upper limit just as hypothesized. The parameter rie measuring agreement of observed with expected increments rose from zero for the 2-man cliques up to ric= 0.98 for the N-man clique-of-the-whole.
From this and other experiments, we infer that logistic diffusion of items is likely to be approximated in so far as populations are homogeneous with very diversely overlapped cliques which are larger than pairs. As cliques enlarge, the diffusion curve approaches the simplest logistic model. At cliques of four persons, acceleration from "seeking non-knowers" offset deceleration from clique barriers.
I. The Problem
The Monte Carlo experiments reported here deal with the problem: Given a population structured into cliques, what effect does clique size have on the diffusion of a message through the population under the conditions of steady random pairing ?7-10
In one of our earlier experiments we found that as the circle of possible diffusion grows from a family to a community to a nation that the average person's circle of contacts grows numerically but shrinks in relative size to the total population?' Does this relative shrinking of the circle of contacts constrain the diffusion of information? Does the creation of cliques also create culs de sac in the communication net?
However variable, shifting and vaguely bounded the cliques may be in a population, their structuring of the population may be described and measured in terms of:
1. the average size of cliques;
2. the dispersion of cliques around the mean size; and
3. the degree of overlapping memberships or correlations between cliques.
Using these three descriptive statistical moments of the distribution of cliques as a starting point, we can seek generalizations about the effects of clique structures on the diffusion of a message through a population by steady random pairing, asking for systematic answers to such questions as the following:
1. How does the mean size of cliques affect diffusion? e.g., Does diffusion improve as cliques become larger?
2. How does the dispersion around the average clique size affect diffusion? e.g., Does the predictability of diffusion, as explored here by a logistic curve, change as dispersion shrinks? If so, what is the function connecting the dispersion (a) with the predictability correlation (r) as clique size (Pc) varies, i.e. does r = f(σ, Pc)?
3. How does the linkage of membership, or overlapping of cliques, affect diffusion? e.g., Does random overlapping (as explored here) tend to eliminate the clique boundaries which retard diffusion and thus tend to approach the unstructured logistic situation of a single homogeneous population offering to every person equal opportunities to diffuse the item? Does random overlapping of clique memberships operate to reduce the diffusion-blocking effects of cliquing more and more as cliques enlarge? We suspected that this double condition holds and tested it as a hypothesis in a previous experiment2,14 and in the present experiments.
Our experiments should yield universal laws of group behavior which are limited neither to the samples observed, nor even to their universes in an observed culture or subculture. Our hypotheses will be in statistical terms such as: "If cliques enlarge (under specified probabilistic preconditions), then the correlations of the observed increments of message knowers with the expected increments of knowers predicted by the logistic model will improve." Hypotheses of this type should hold regardless of what society is under observation, or what cultural or other factors have caused the enlargement of the cliques under observation just so long as the preconditions specified are present and operative. Hopefully, continued testing and use of these laws will develop techniques of measuring, and later creating and manipulating, the necessary preconditions for changing clique size.
To further the mapping of clique effects on diffusion, the experiments reported here were a Monte Carlo test in a simulated population of strictly randomized numbers representing persons. The "clique-size hypothesis" expected: As clique size enlarges from 2 to N, the observed diffusion of an item by steady, random pairing will approach as an upper limit the linear logistic diffusion curve, defined by cumulating the most probable increments, Δt = kpiqi.1
This clique-size hypothesis imposes a structuring on the formerly homogeneous population which is assumed by the linear logistic hypothesis, namely: "In so far as a set of actors approach equality of opportunity to transmit an item steadily from man to man, in just so far the course of this co-operative social process is predicted by a logistic curve which cumulates the joint probabilities of knowers meeting and telling non-knowers." In simplest terms the logistic hypothesis expects: "Free pairing produces logistic diffusion of an item."
II. Background
In a previous experiment" this hypothesis of logistic diffusion in the entire population (N) as the upper limit of clique size was tested with ambiguous results – which called for the present more exact testing to resolve.
A population of playing cards had been structured in the previous experiment into randomly overlapped cliques (as described below) of 2, 3, 4, 5, 6, 7 and 8-man cliques and a clique-of-the-whole which was the linear logistic limit. Ten per cent of the simulated "persons" were chosen by lot as "starters". Partners were chosen in each period of interacting with the aid of a book of random numbers (i.e. pairing off). The resulting communicational networks were traced through the matrix of responses which comprised a row for each person, a column for each period, and a cell entry identifying the row person's partner in the column period. The increments in knowers were totaled for each period at the foot of each column.
The graph for the 2-man clique was much below the logistic curve, the graph for the 3-man clique was nearer, and the graph for the 4-man clique almost coincided with the logistic expectation. But the curve for the 5-man clique exceeded the slope of the logistic curve, and those for the larger cliques exceeded it more and more. The hypothesis that diffusion curves would more nearly approximate the logistic curve as clique size increased was confirmed up to 4-man cliques, while its opposite was confirmed for larger cliques. Clearly some
further systematic factor(s) operated, overlaying the purely random causation of the linear logistic model.
The disturbing factor was found to be a constraint that had been built into the model to simulate life situations. In life, people seek non-knowers of a news item to tell it to as "news" in preference to retelling it to a knower. They ask, "Have you heard . .?" and proceed differently according as the respondent has or has not heard the news item at issue. Accordingly, in drawing for each person's partner within his clique, the procedure had limited the drawings to the nonknowers in his clique. This accelerated the diffusion as each knower talked only to non-knowers until all his clique were knowers.
This constraint of "seeking out non-knowers" accelerated the diffusing and counteracted the decelerating constraint of barriers between cliques. For the effect of cliques is to concentrate diffusion within cliques, decreasing diffusion between cliques, and thus make for inequality of opportunity to interact among the actors in the total population. The relative strength of the two opposed constraints just balanced each other here in 4-man cliques.
This finding suggested two further lines of controlled experiments. First, the researcher could vary each constraint alone while holding the other constant and thus measure in turn the isolated effects of clique size and of seeking-out-non-knowers. (This line of experimentation is what the present paper reports.) Secondly, the researcher could vary both constraints together in varied relative shares and thus measure each as a social force in terms of the amount of the other it could neutralize. This technique could be extended and generalized to adding constraints one at a time and assessing the conditional effects of each amount of each constraint as influenced by each amount of each other constraint that might be varied in controlled degrees in the total experimental design up to that point.
III. Procedure9-14
The simulated population used in these experiments consisted of the numbers from 1 to 60; (N = 60). These "persons" were organized into randomly overlapping cliques; in the first experiment the cliques were of 2 persons each. In the second experiment the cliques were of 3 persons each, and each succeeding experiment increased clique size by 1 person up to cliques of 8 persons each, and finally a clique-of-the-whole (60 persons) for the last experiment. Arrangement of the population into cliques was done by drawing random numbers from a large random number table of one million digits. The drawing of cliques was governed by the following restrictions:
1. Number of cliques: A clique was always drawn for each of the 60 persons, i.e. 60 cliques, regardless of size of clique. Drawing was made with replacement so that a person could appear in several cliques, subject to the restrictions below.
2. Connectivity between cliques: Each "person" must appear in a clique of at least one other person, i.e. there must not be any "isolates" (persons having a clique but not being in another clique).
3. Non-repetition in a clique: No person appears more than once in any one clique. (He may appear by chance in from one to N different cliques.)
4. Non-repetition in a pair: No self-meeting is allowed.
For each clique size (2, 3, 4, 5, 6, 7 and 8-man cliques and a clique-of-the-whole) a "diffusion matrix" was made. In this matrix the row number represented the person to whom the clique listed in the stub of the row belonged. Each column represented one time period of interaction and was labeled T0, T1, T2, ..., Tt. Each cell of the matrix recorded the partner of the row person in its column period.
Ten per cent of the population (6 persons) were selected at random to be the original knowers of the message, or "starters". In each period each person in turn had a partner drawn by random numbers from his clique. Thus all pairing was randomized within the randomly overlapped cliques. There was no "seeking out" of non-knowers. If a knower met a knower the meeting did not add to the observed increment for that period.
In this diffusion matrix the "descendants" or hearers of any person can be read off in the rightward columns. Any person's "ancestors" or previous tellers are read off in the leftward columns.
The number of persons learning the message in a time period is totaled at the bottom of that column. These totals simulate the observed increments (= Δt) in knowers. Cumulative totals are carried from column to column below the increments.
Three sets of observations were made, replicating each of the eight matrices (2, 3, 4, 5, 6, 7 and 8-man cliques and clique-of-the-whole). By drawing for three different sets of starters and tracing their descendants rightward through the columns of the matrix, one can average the three totals of replicated increments, Δ', Δ", Δ'", in each column The observed increments in knowers were thus averaged over the three replications (in the diffusion matrix for each clique-size) in order to stabilize sampling fluctuations somewhat. The observed cumulative proportions of knowers were similarly averaged. The cumulative averaged-observations were then graphed in comparison with the logistic model. Interclass and intraclass correlations were computed between each set of observed increments and the expected increments for the whole population (N = 60) persons.2 Since cumulation compels some spurious correlation, using correlations of increments is the more exacting test of agreement between fact and theory in growth series.
IV. Findings
The correlations are reported in Table 1 below. Figure 2 below shows a summary of selected graphs for the 2, 3, 4 and 8-man cliques, and the clique-of-the whole as compared to the logistic model.
Table 1: Correlations between Observed and Expected Increments as Clique Size Increased
|Clique size |Interclass |Intraclass |
| |r |r |
|2-man |0.000 |0.000* |
|3-man |0.931 |0.842 |
|4-man |0.948 |0.913 |
|5-man |0.958 |0.927 |
|6-man |0.964 |0.949 |
|7-man |0.963 |0.930 |
|8-man |0.993 |0.982 |
|Clique-of-the-whole |0.986 |0.980 |
*The expected increments were computed from the logistic model Δt = kptqt with the values of p and q for T0 set at p = 0.10, q = 0.90, k = 1. Note that both the observed and the expected increments of knowers are for the entire population (N = 60), not for each clique separately. The formulas used for the interclass and intraclass correlations were:
Where = expected mean square between groups. r = ,
= expected mean square between groups. r =
Table 1 shows increasing correlations between the observed increments and the logistically expected increments as clique size increases. Interclass correlations ranged from 0.000 for the 2-man clique (over 14 periods) to 0.993 for the 8-man clique (over 14 periods) and 0.986 for the clique-of-the-whole. Intraclass correlations ranged from 0.000 for the 2-man clique to 0.982 for the 8-man clique and 0.980 for the clique-of-the-whole. The greatest increases in both the interclass and the intraclass correlations were shown between the population structures in 2-man cliques vs. in 3-man cliques.
Figure 2 shows in detail the increasing approximation to the logistic model as clique size increases. The 5, 6 and 7-man cliques are eliminated from the graph so that the picture is easier to see. As in Table 1, the most significant increases in approximation to the logistic curve are between the 2 and 3-man cliques and between the 3 and 4-man cliques. The diffusion pattern for the 2-man clique is not logistic in shape at all; it is a straight line increment of one person per period until a knower is encountered and that chain terminates (which did not happen in the fourteen periods graphed in Figure 13). The 3-man clique permits a logistic type of diffusion growth, but divergence from the model occurred in the third period of interaction and the rate of growth remained considerably below that of the model. Beginning with the 4-man clique, the diffusion patterns were close to the logistic model until approximately one-half of the diffusion was completed, then they tended to fall below the model.
Figure 1: Logistic diffusion when clique size varies and non-knowers are "sought out".
[pic]
Cumulative growth curves of knowers as the size of cliques, randomly overlapped in
membership, went from 2 to 6 persons, under conditions of "seeking-out" non-knowers
from within one's clique.
|Legend |
|Logistic model |-X- |
|2-Man cliques |-2- |
|3-Man cliques |-3- |
|4-Man cliques |-4- |
|5-Man cliques |-5- |
|6-Man cliques |-6- |
The logistic curve describes and predicts the spreading of an item under conditions of random, steady, pairing by the formula Δ = KPQ
where:
Δ = An increment in
P = The proportion of knowers and
Q = The proportion of non-knowers (P+ Q = 1)
K = "Potency parameter" = hearers per teller and period. (K = 1 here.)
The findings thus tend to confirm the hypothesis here. As clique size increased, the diffusion curves rose (i.e. accelerated) towards the linear logistic as the upper limit of just one "clique" of size N.
Figure 2 illustrates the findings of the clique size experiments reported in this paper and the findings of the previous series of experiments are shown in Figure 1. The differences in the two graphs demonstrate clearly the effect of eliminating the "seeking-out" factor for the current experiments, as compared to the effect of seeking-out when it was a required condition of the earlier experiments. With seeking-out, the logistic model of diffusion was surpassed by the 5 and 6-man cliques: That is, diffusion in 5 or 6-man cliques was more rapid than predicted by the model when knowers sought non-knowers to meet with. When the seeking-out was eliminated (and meetings were more completely randomized) the original hypothesis was confirmed. As clique size increases, the pattern of diffusion of a message through a population approximates the logistic curve more closely, and the logistic curve is the upper limit of the diffusion pattern.
Figure 2: Logistic diffusion when clique size varies.
[pic]
Cumulative growth curves of "knowers" as the size of cliques, randomly overlapped in membership, went from 2 to 60 persons.
|Legend |
|Logistic model |-X- |
|Clique of Whole |-W- |
|Clique of 8 |-8- |
|Clique of 4 |-3- |
|Clique of 3 |-2- |
|Clique of 2 |-6- |
|N = 60 |
|Time periods (of interaction) |
People interacting at will within cliques approximated random interaction more and more closely as cliques enlarged. The five intraclass correlations, Ric, measuring the closeness of fit of the observed increments to the logistic expectation were:
Clique of the whole = 0.98 Clique of 5 = 0.93
Clique of 8 = 0.98 Clique of 4 = 0.91
Clique of 7 = 0.93 Clique of 3 = 0.84
Clique of 6 = 0.94 Clique of 2 = 0.00
V. Explanation
Towards a rationalized explanation of these empirical findings we offer two subhypotheses. We hypothesize that the decelerating of diffusion as cliques get smaller is due to: (a) Earlier dropping out of the communication network; and (b) lower overlapping of clique memberships.
a. Earlier drop-outs. Since each person tells the item only within his clique of Pc persons, he ceases to diffuse the item as soon as all his clique become knowers, for thereafter he merely overtells (i.e. repeats the news) at all his further meetings. The smaller the clique, the sooner each person tends to drop out as an effective diffuser. Thus at the two limits, in 2-man cliques each person always drops out in the next period after his first telling, whereas in the N-man clique every person continues to be a potential. teller as long as there are any nonknowers in the whole population.4
b. Lower overlapping of clique memberships. The number of direct linkages possible for a given clique is the sum of the outside contacts of its members through their own separate cliques. Therefore, as the size of the clique decreases, the number of direct linkages to the total population also decreases proportionately. As the number of linkages to the total population decreases, the effectiveness of the clique members as message diffusers also decreases. (An algebraic description of this function is not yet available. This is a problem that can also be investigated by computer simulations.)
In the tests of the 2-man cliques, for example, the population was actually split, by chance, into two mutually exclusive groups of cliques. In one test the randomly selected "starters" all fell into one of these groups, and the diffusion of the "message" could not be completed because a segment of the population could not be contacted. This sort of isolation of single cliques or groups of cliques may well occur in a real population, and it suggests the need for a further refining factor in the final models for diffusion.
To distinguish between these two decelerative mechanisms (and any others) and to test hypotheses about each will require analytic or algebraic formulation first and further controlled experimentation next, both on randomized populations (as here), on social groups operating at will, and possibly through computer simulations of variously structured populations.
For such analysis, our concept of social forces1-4 should prove useful. We defined a unit social force, isomorphically with physics, as one person celerated one unit of change per period, per period. Such a social dyne is here one person changed to being a knower in one period if starting from scratch. Each alternative decelerative mechanism could then be measured as to how much increase or decrease in celerating the diffusion it produced in a constant set of people and periods.
F = APT-2, a social force. Equation 2
Future experiments should show (in a situation of diffusing with X units of potency, or dynes of social force) just how many dynes were cancelled by the dropping out, or by the overtelling, or by the rebuffing mechanisms that decelerate diffusing.
For fuller interpreting of the findings here it must always be kept in mind that sampling fluctuation definitely affects the stability of the observations reported here. When the experiment was set up, an N of 60 was chosen as being comparable with the populations used in the previous experiment with cliques and as being small enough to facilitate construction of the cliques and tracing of the diffusion patterns. An increase in the population would result in a disproportionate increase in the amount of labor involved in constructing cliques and tracing and checking diffusion patterns. For this reason, an effort was made to compensate for the small population and sample fluctuation by using replications of observations and by averaging the observations. It must be remembered that further replications (only three were used) might have resulted in further stabilizing of the diffusion patterns reported here. For instance, it might separate more clearly the differences in diffusion pattern between the 4 and 8-man cliques. At present their diffusion patterns are intertwined. Fuller control of sampling fluctuations would, of course, accrue if the randomizing were done by programming the IBM 709 suitably, and it would accrue most fully by analytic formulation.
VI. An Application
The basic application of this logistic cliques model lies in the field of most human communication systems other than the mass media. The diffusing of items and sets of items of information, action, property or culture through any population will tend to follow the logistic S-shaped curve in proportion as its three necessary and sufficient preconditions exist, namely, steady, random, pairing off. All degrees of irregular departure from these conditions in any combination are of course possible. But far from invalidating the model, such heterogeneous irregularity, if it goes far enough, reapproaches the model. For as heterogeneity of preconditions becomes in effect a situation due to many small uncorrelated influences, it becomes the chance-like situation denoted by the word "random". We therefore may expect the logistic diffusion model to hold either in a controlled laboratory experiment or in a population large enough so that all the relevant variables tend to be randomly distributed or related throughout the population.
A recent study of 60,000 U.S. families, for example, notes the tendency of families to structure themselves in congenial cliques of about five like-families.15 This provides a primary group, between the nuclear family and the neighborhood, which has like-values and habits and which supports each family in socializing their children.
This study by Harvard sociologists Zimmerman and Cervantes suggests how multiplexly overlapped cliques tend to reinforce American families in rearing a normal oncoming generation. We see this logistic cliques model as simply describing a regular probabilistic structure which can be the upper limit (or metropolitan pole) of societal structuring, where irregularities approach the many small uncorrelated influences which may be called "chance influence" or "multiplex causation" or "equal opportunities for everyone".
VII. Research Needed
In order to develop and test the simple logistic theory described above into a set of laws of social diffusion, we see at least the following research steps as desirable to follow up the first step which is reported in this paper.
a. To validate diffusion laws as related to the mean size of cliques requires:
i. Testing the Monte Carlo demonstrations (reported here) by further tests on human populations which interact at will when pairing off with equal opportunity within their randomly overlapped cliques of like-size.
ii. Expressing the dependence of diffusing on clique size in an analytic formula of fullest possible generality.
b. To develop diffusion laws as dependent on the higher moments of cliques requires:
i. Formulating algebraic models and testing them on human groups who act spontaneously within the equalized opportunities when the sizes of the cliques within one population and one testing vary systematically such as in a normal curve with dispersions systematically increasing.
ii. Similarly modeling and testing when cliques are linked in membership with average linkage (i.e. per cents of common members) systematically varied in controlled experiments so that r„, = 0, 0.1, 0.2, ..., 0-9, 1-0. (When the average intercorrelation of cliques is maximal so r„= 1.0, there can be only one totally inclusive clique which is the case of the linear logistic upper limit again.)
c. To develop diffusion laws as dependent on the further transact factors and conditions requires controlled experiments testing models which systematically vary:
i. Non-random overlaying variables (see Reference 14 for a list of 52 such circumstantial dimensions correlated to diffusion);
ii. Non-steady acting in the time dimension;
iii. Non-pairing structuring of the population dimension as when meeting, not in pairs within cliques, but in triplets, quartets, quintets, up to N-tets;
iv. Non-constant varying of the other transact dimensions of space, valued objects and other diffusing actions.
Such a research program of controlled experiments will develop, we predict, a firm set of transactive laws in social mechanics.5
References
1. Dodd, S.C., (1942) Dimensions of Society, Macmillan, New York.
2. Dodd, S.C., (1947) Systematic Social Science, American University of Beirut, Social Science Series, No. 16. University of Washington Bookstore, Seattle.
3. Dodd, S.C., (1936) "A theory for the measurement of some social forces", The Scientific Monthly, 43, Nos. 1 and 250.
4. Dodd, S.C., (1936) "The standard error of a ‘social force’", Annuals of Mathematical Statistics, 7, No. 4.
5. Rapoport, Anatol, (1951) Nets with distance bias. Bulletin of Mathematical Biophysics, 13.
6. Dodd, S.C., (1952) "On all-or-none elements and mathematical models for sociologists", American Sociological Review, 17, No. 2.
7. Dodd, S.C., (1953) "Can the sociologist serve two masters? An answer through experimental sociology", Research Studies of the State College of Washington, Vol. 21.
8. Dodd, S.C., (1955) "Diffusion is predictable: Testing probability models for laws of interaction", American Sociological Review 20.
9. R. J. HILL (1955) "Temporal aspects of person-to-person message diffusion", Unpublished Ph.D. Dissertation, University of Washington.
10. Dodd, S.C., (1956) "Testing message diffusion in harmonic logistic curves", Psychometrika, 21, June.
11. Dodd, S.C., (1957) "A power of town size predicts internal interacting – A controlled experiment relating the amount of an interaction to the number of potential interactors", Social Forces, 36, No. 2.
12. Dodd, S.C., and Garabedian, P. G., (1961) "The logistic law of interaction when people pair off at will", Journal of Social Psychology, 53, 143-158.
13. Garabedian, P. G. and Dodd, S.C., (1962) "Clique size as a factor in message diffusion", Washington Public Opinion Laboratory Document U: 59-75. Sociological Inquiry, 33, 71-81.
14. Dodd, S.C. and McCurtain, M., (1961) "The Logistic Law in Communication", Series Research in Social Psychology, The Symposia Studies Series No. 8 of the National Institute of Social and Behavioral Science, Washington, D.C., September.
15. Zimmerman, C.C. and L. F. Cervantes. L.F., (1960) Successful American Families, Pageant Press, New York.
Notes
1. The logistic growth formula is: Δt = kptqt,
where: pt = the current proportion of knowers
qt = 1– pt = the current proportion of non-knowers
k = the "potency parameter" specific to the message, the population and the situation. (For these experiments k = 1.) It is the number of new hearers per knower in a unit period of interaction.
The logistic growth formula, Δt = kptqt), qt, predicts that when an all-or-none variable (such as an item message) is diffused through a population, by the three conditions: i.e.
a. steady
b. random
c. pairing of the units of the population it will show a cumulative S-shaped growth curve (a logistic curve).
This formula is an illustration of the elementary probabilistic law of dyadic interaction, i.e. the probability of a knower (p) meeting a non-knower (q) in any one time period. The curve is constructed by cumulating the "most probable" finite increments (pq) in successive time periods. This is a case of the law of joint probability (pq).
2. This correlation was observed as 0 over the first 14 time periods (the number involved for the other clique sizes). It was r = 0.713 (interclass) and 0.392 (intraclass) when observed up to complete diffusion which required 29 time periods before all three 2-man cliques closed themselves out by encountering only knowers.
3. This accounts for a difference between ric = 0.00 for the 2-man cliques in the 14 periods graphed in Figure 1 vs. ric= 0.39 for the 2-man clique when computed over the first 28 periods which were required to terminate the three chains here and complete the diffusion (at less than 100 per cent of the population becoming knowers).
4. This "dropping out" of active tellers as cliques become fully diffused has much in common with the random net model developed by Rapoport (Reference 5). He posited a terminating potency of k hearers per "teller-ever", so that each actor drops out after telling k hearers. This seems almost our former model which had the constraint of seeking out non-knowers (Reference 14). There each actor dropped out of action as soon as he had told the message item to the t fellow members (t = k here) of his clique in the first t time periods (or fewer periods in so far as some fellow members would hear through other contacts). In the present Monte Carlo experiment, since all the pair meetings are random ones but are limited to one's own clique, and since all cliques were overlapped and interconnected, the diffusing could go on as long as even one person remained a non-knower (i.e. for t periods).
The possible overall duration of diffusing can be estimated for the simple case of "full" diffusing where k = 1 (as here), i.e. where each teller meets and tells just one person in each unit period. The probability, Pi, of a teller's meeting a particular fellow clique member is inverse to the number, Pc–1, of such fellow members, i.e.:
pi = 1/Pc–1 Equation 1a
qi = 1 – 1/Pc–1 Equation 1b
5. Note the research program of modeling and experimentation, as sketched in paragraphs 7(a), (b) and (c), typifies our strategy to develop exact laws in this "social mechanics" branch of Sociology by means of dimensional analysis. (For a complete discussion of dimensional analysis see References 1 and 2.)
Tables of Contents of Stuart C. Dodd books
Social Relations in the Middle East
A Textbook in Citizenship prepared for the freshmen at the American University of Beirut
(Third English edition. Revised and enlarged)
by
Stuart Carter Dodd, Ph. D.
Professor of Sociology
and
Assistants in the Course
American University of Beirut
1946
American Press,
Beirut, Lebanon.
AMERICAN
UNIVERSITY OF BEIRUT
PUBLICATIONS
OF
THE FACULTY OF ARTS AND SCIENCES
[pic]
SOCIAL SCIENCE SERIES
No. 17
English Editions:
First edition 1930
"Social Relationships in the Near East" With Teachers' Manual pp. 633
Second edition 1940
"Social Relations in the Near East" With Teachers' Manual pp. 790
Third edition 1946
"Social Relations in the Middle East" With Teacher's Manual pp. 800
Arabic Edition 1947
"Social Relations in the Arab World" Dar el Kitab, publishers
Table of Contents - The plan of the book
Part I. People of the Middle East – Distribution
Chapter I. Sectors of Society –
People and their Interrelations in Space and Time 3
Chapter 2. Demography 15
Part II. Interrelations of People — Institutions
A. Domestic relations – between Men and Women
Chapter 3. The Family 31
Chapter 4. Status of Women 43
Chapter 5. The Home 71
Chapter 6. Child Welfare 83
A. Scholastic Relations – between Children and Educational Specialists
Chapter 7. Educational Aims 93
Chapter 8. Educational Problems 103
B. Economic Relations – between Consumers and Production Specialists
Chapter 9. Production 115
Chapter 10. Agriculture 129
Chapter 11. Money 143
Chapter 12. Standards of Living 175
Chapter 13. Cooperatives 159
B. Political Relations – between Citizens and Governmental Specialists
Chapter 14. Government 187
Chapter 15. Law and Crime 203
Chapter 16.. Parliamentary Procedure 215
C. Religio-Ethic Relations – between Laymen and Religious Specialists
Chapter 17 . Judaism 227
Chapter 18. Christianity 255
Chapter 19. Islam 279
Chapter 20. Other Religions 305
D. Hygienic Relations – between the Sick and Medical Specialists
Chapter 21. Public Health 317
A. Philanthropic Relations. – between Subnormals and Welfare Specialists
Chapter 22. Defectives 351
Chapter 23. Social Insurance 367
H. Recreational Relations between Recreators and Recreational Specialists
Chapter 24. Recreation 381
Chapter 25. Social Service 395
H. Aesthetic Relations – between People and Artistic Specialists
Chapter 26. Fine Arts 409
I. Scientific Relations – between Benefiters and Scientific Specialists
Chapter 27. Science 423
I. Communicational Relations – between the Public and Communication Specialists
Chapter 28 Language 441
Chapter 29. Agencies 459
Part III. Spatial Relations of People – Regional Communities
A. Environmental Relations – between People and their Habitat
Chapter 30. Ecology 475
B. Desert Relations T– among Nomads
Chapter 31. Bedouins 493
C. Rural Relation4 – among villagers
Chapter 32 and 33. Village Problems 511
D. Urban Relations – among Townsmen
Chapter 34. Town Planning 539
E. National Relations – among Countrymen
Chapter 35. Patriotism 549
F. International Relation's – among World Citizens
Chapter 36. East and West 559
Chapter 37. The United Nations 593
Part IV. Temporal Relations of People – Social Change
A. Past Relations between the Living and Their Ancestors
Chapter 38. History 611
B. Present Relations – between the Living and the Living
Chapter 39. Social Processes and Forces 621
C. Future Relations – between the Living and Their Descendants
Chapter 40. , Social Planning 635
List of Plates
|Plate |Chapter |Topic |Page |
|1 | |Changing Social Relations | |
|2 | |Old customs... Frontispiece - New customs... | |
|3 |Demography |Some Populational Characteristics |16 |
|4 |Demography |Distribution of Hygiene &Pores | |
|5 |Demography |What Population Pyramids Show |18 |
|6 |Demography |Correlation of Birth and Death Rates |19 |
|7 |Status of Women |Work of a Visiting Nurse |72 |
|8 |Child Welfare |Playgrounds Reduce Crime |73 |
|9 |Aims of Education |School Classes of the Past and Present |94 |
|10 |Aims of Education |Modern Times Demand Modern Schools |95 |
|11 |Production |Shoemaking—by Hand and by Machine |120 |
|12 |Production |Productivity and Working Hours |121 |
|13 |Agriculture |Methods of Irrigation |138 |
|14 |Transportation |... Bridges |139 |
|15 |Transportation |Travel Time of a Letter |168 |
|16 |Standards of Living |Distribution of Income |224 |
|17 |Parliamentary Procedure |Parliamentary Procedure |225 |
|18 |Religion |Jerusalem |240 |
|19 |Public Health |Health Indices |323 |
|20 |Dependents |Palliatives - Alms to a Beggar |362 |
|21 |Dependents |Curatives and Preventives |363 |
|22 |Social Insurance |The Mortality Table | |
|23 |Recreation |A Cafe and a Park |390 |
|24 |Recreation |A "Women's Club" and a Playground |391 |
|25 |Fine Arts |A Rug and a manuscript |412 |
|26 |Fine Arts |The Taj Mahal |413 |
|27 |Science |Research Needed |438 |
|28 |Language |Languges of Commerce |446 |
|29 |Language |Esperanto at a Glance |447 |
|30 |Language |A Simplified Arabic Script |452 |
|31 |Language |Basic English Vocabulary |453 |
|32 |Language |Model English |457 |
|33 |Ecology |Civilization and Climate |488 |
|34 |Bedouins |Nomads and Motors |489 |
|35 |Village Problems |Forested and Treeless Hills |528 |
|36 |Village Problems |Goats and Tree Nurseries |529 |
|37 |Town Planning |Urbanization |540 |
|38 |Town Planning |Jerusalem Town Plan Map |541 |
|39 |Town Planning |Street Shade—Arches |546 |
|40 |Town Planning |Street Shade—Trees |547 |
|41 |East and West |An American University in the Near East |574 |
|42 |East and West |Personalized Rural Health Insurance |475 |
|43 |United Nations | |598 |
|44 |History |Mass Time Triangle |618 |
|45 |Social Planning |A Graph of Civilization |634 |
|46 |Social Planning |The Proposed World Calendar |635 |
|47 |Social Planning |A Time Curve and a Schedule |638 |
|48 |Social Planning |A Budget and a Five-Year Plan |639 |
|49 |Social Planning |Human Welfare |645 |
The purpose of Chapter 1 is to show the student-citizen a systematic plan for studying Society, and to whet his appetite for the chapters ahead.
I. The purpose of this book
This book, through introducing the student to the social sciences, is a course in citizenship. Whether as an introduction to further study of the social sciences, or as a cultural survey of the field for those students who will specialize in other fields, a fundamental purpose of this book and this course is to develop more intelligent and more ardent citizens. We shall study the problems upon which all citizens should have some information, and on which they should try to form intelligent opinions. In whatever profession or occupation each of us specializes, we are all citizens of some local community, of some nation, and of the world. As citizens, it is the duty of every one of us to think about, and to try to improve conditions in our community, in our nation, and among nations. This book aims to teach and stimulate thinking on our common duties as citizens. This course aims to fulfill part of the larger -purpose of this University to develop thinking citizens and leaders of society,
The central thesis of this book may be put in three statements as follows:
A. Social relations are changing.
In the modern world change is the rule if one thinks of change as varying in time rate or speed from very fast (i.e. in a few months) to very slow (i.e. in several centuries). It is difficult o think of any social relation that is not changing at some rate.
B. Social relations are imperfect.
They are currently far short of the ideal that people want Furthermore their changing may be towards greater perfection or may be the opposite.
C. Therefore social relations must be studied towards controlling them.
In order to make the changes desirable ones people must learn how to direct change towards whatever seems to them to be more perfect. The student citizen should study social relations then in order to cause progress. For progress, defined as change desired by the people, does not happen of itself but only comes in proportion as people work towards goals.
II. The limits of this book
A. Relations in the Middle East.
In this book, we shall study people and how they live together. But the study of people living-together is far too large a subject to be studied fully in one course. We shall have to limit our study. There are more than two billion people (2.000,000,000) in the world, and their life together is very different in the Sudan and Siberia, in China and in Europe, in the desert and the city, among the poor and among the wealthy.. Thus we cannot study all the different kinds of people and all the kinds of group life in the world. Consequently, we shall limit our study to people in our own environment in the Arabic-speaking Near East. In studying how these two percent of the people of the world live together, we shall study the principles common to people everywhere; but their application will be made chiefly to local populations. We shall study our own social environment as a. sample of human life everywhere.
B. Current relations.
We shall further limit, in three more ways, this large subject of people-living-together, which we call society.. A second limit is, that out of the thousands of millions of people who have ever lived in the Arabic-speaking East, roughly between Khartoum and Baghdad, Mecca and Aleppo, we shall only study the life of the forty odd million who are living now.
We shall leave to courses in history the study of the background of our present society. History describes the .development of our society to date; sociology studies the principles operating in society around us. Therefore, in this introductory course in sociology, we shall study people as they live together at present. A major purpose in studying both history and sociology, both the past and the present, is, of course, to aid in determining the future society and in determining how people may live together in the years ahead.
C. Social relations.
A third limitation is, that we shall study the social relations of people, not their physical relations; that is, we shall study the relations between people, not the relations between people and material things. We shall study the relations between person and person, between persons and a group, and between groups: These interhuman influences are social relations. The relations of people to the climate and the soil, to plants and to animals, is left to courses in the natural sciences – astronomy, physics, chemistry, biology and their subsciences. This book studying, as it does, social relations, i.e. how people live together, as an introduction to the social sciences.
A survey course.
This brings us to the fourth limitation of this volume, namely, that it is only an introduction to the social sciences, not a thorough study of them, or of any one of them. It is a brief survey of the field covered by the various social sciences – economics, political science, social psychology, anthropology and sociology. Some students will major in some one social science and in later years at college, will study that field more thoroughly, but for many freshmen who specialize later in medicine or the natural sciences, or in the languages, this survey course will be their last study of social relations in a systematic way.
Relation to other courses.
In fulfilling the purpose of this course, namely, to help in developing thoughtful citizens, the relation of this course to other courses of freshman and sophomore :years should be clearly realized. The freshman courses in history, the physical sciences, and the social sciences, mutually supplement each other. Through them the student should get a more comprehensive view of man's past, of his relations to physical things, and of his relations to himself and to his fellow men. The other freshman courses are necessary tools for further study –languages are tools for studying any subject, and mathematics, "the language of science", is an essential tool for scientific studies. In sophomore year, in addition to starting specialized courses in a major field, the philosophy course aims to deepen the student's understanding of fundamental questions underlying history, the natural sciences and the social sciences. In philosophy, the student will study the deeper issues of the purpose and meaning of man and of the purpose and meaning of the universe as far as man can know it. In history and in the natural and social sciences, we study how man is related the past and how he is related to material things and to other men, but in philosophy, which means "love of wisdom", we ask why these relations exist at all. In these freshman courses we seek to know the truth about man's past and present environment; in philosophy we seek the deeper truth underlying all man's attempts to know at all.
III. The plan of this book.
A. Analysis of sociology by its four sectors.
We have noted the limits of this book which define negatively our study of human society, by showing what we shall not study, and what we shall leave to other courses in college. Now let us outline positively how we shall go about our study of society. Human society means people-living-together. Of the many different possible ways of analyzing such a large subject as human society we shall analyze, it in terms of the four sectors which can be most scientifically determined.
Society may be analyzed into four sectors, namely, space, time, people, and their interrelating characteristics. The first sector; space, is the familiar geographic surface of the earth. The spatial plan of our town and of the rooms of our houses are a most fundamental aspect of our environment. But man also lives in time. Generations of our ancestors have been born and have died: in the past, and these cycles of life continue through the ever-sliding present on into the future. We remember our past experiences, we feel what we experience at present, and we try to predict and control our future experiences. The aim of modern science is to predict and control phenomena in the future. Man's life is intimately involved with time. Life itself is an activity, a something-happening-in-time, not a static thing. How people live together cannot be photographed at any one instant of time; it requires a cinema film showing their existence during a period of time.
People in time and space are, then, three basic units for sociology. Physics, which studies things, rather than people, similarly employs, since Newton's day, three basic units of-space, mass and time, measured, in the metric system, in centimeters, grams and seconds. Space and time are fundamental concepts for sciences, and "people", a population-mass instead of a mass of matter, distinguishes the social sciences from the natural or physical sciences.
But these sectors are not sufficient for an understanding of society. There are also all the millions of characteristics of people and of their environment, which we need to understand. It is these characteristics which relate people to each other and to their sub-social environment. Characteristics may be defined as every which is not the number of people, space, nor time. Characteristics include, broadly all the other things we know about people and their environment. Thus defined, everything that can be known about people and their environment can be described in terms of these four sectors – space, time, people, and their characteristics. A more definite notion of what "characteristics" mean and a classification of them into their different classes and subclasses, is given later on. But first, note why these four sectors have been chosen to aid us in our analysis of society.
These four sectors of society enable us to make a scientific 'study of society, instead of a literary or a philosophical study.
A scientific analysis of society is immediately chosen because this course is an introduction to the social sciences, and more ultimately, because the usefulness to man of scientific methods have been so great in the last few centuries, that their application to society itself, also promises to be highly useful.
What is meant by the adjective "scientific"? The meaning of scientific knowledge will be studied more fully in science courses and in philosophy. We shall mention here only three properties of scientific knowledge. Knowledge is scientific when it is exact, verified, and systematic. Exact knowledge is knowledge that is accurate, or has been secured by measurement. If something can, be measured it can be known most exactly. Progressively as the social sciences advance, more characteristics are being reduced to measurement. People's intelligence, their education, their attitudes and beliefs of many kinds, their health, their wealth, their behavior in some religious and recreational situations, and many aspects of their environment, can be measured by various tests and surveys. These four sectors yield, therefore, a more scientific analysis of society, because three of them can be exactly known and the forth though much less exact, is, by research, becoming; little more so every decade.
Among many techniques for verifying knowledge one important one is, that different persons who independently observe phenomena, must agree in their reports of those phenomena. Not until more than one astronomer sees a new star, is its discovery verified Not until more than one doctor reports successful treatment of patients with some new vaccine is such a report verified. Not until several competent observers working separately, agree on some item of knowledge is it considered to be verified, i.e. to be "true". This test of the agreement of competent observers has been applied to this analysis of social phenomena into the four components. Experiments have been made in which observers independently analyzed hundreds of sets of social data into these four sectors and achieved 97% of agreement in the formula they wrote. Therefore, this analysis into these four sectors is one that can he verified by scientific experiments and has been partly verified by a few such experiments.
A third property of scientific knowledge, is that it be systematic; that is, that that field of knowledge be organized in an orderly system such as in a related set of scientific laws or in a classification. We believe that four sectors of space, time, people and their characteristics, provide an orderly system of social knowledge. The system can be expressed in an algebraic equation. (See Dimensions of Society by Stuart C. Dodd, Macmillan, 1942.) This equation provides a comprehensive classification for social phenomena. It can be demonstrated by experiment, that this equation classifies any set of social data into some definite class in this classification, according to the amount of each of the four sectors that exist in those social data. The four parts of this book (see Table Of Contents) "People", "People with interrelating characteristics", "People in spatial relations", and ."People in temporal relations", illustrate a systematic or orderly classification of knowledge about society.
There are other properties of scientific knowledge besides these properties of exactness, verification and system. However, in this introduction to the social sciences it is perhaps enough to simply suggest that these three properties are possessed by the four sectors which have been chosen to aid us in making a scientific study of society.
B. The first sector—people. A prospectus of Part I.
In Part I of this book, Chapter 2 on Demography introduces two sectors, the people in the Middle East today, and some of their more important characteristics. "Demography" means "writing about people". In this chapter, we shall study how populations change in size by births, deaths, and migration, and how far these changes can be controlled.
Would a larger population make your nation greater and more powerful? Or, would it mean over-population and increased poverty? How do you know? Chapter 2 helps to answer such questions.
C. The second sector — interrelational characteristics. A prospectus of Part II.
In Part II of this book, Chapters 3 to 30 discuss part of the characteristics sector of society. As characteristics are extremely numerous and varied, a classification of them is useful to make of them an orderly system of what, otherwise, seems a complex, confusing, universe. The characteristics of people and of their environment may be briefly classified for sociological purposes as follows:
I. Physical characteristics (inorganic)
A. Soil and minerals
B. Water
C. Air
D. Energy radiations — heat, light, electricity, radio waves
E. Other natural processes — gravity, tides, chemical action, etc.
II. Biological characteristics (organic)
A. Plants (Botany)
B. Animals (Zoology)
1. Germs to apes
2. Human beings
a. Anatomical (i.e. structural) characteristics
b. Physiological (i.e. functional) characteristics-,'.,
i. Neural-glandular characteristics (Physiological Psychology).
ii. Other physiological characteristics digestion, respiration, etc.
III. Cultural characteristics (superorganic)
A. Physico-cultural characteristics (Engineering )
Shelters, clothing, tools, machines, man-made things.
B. Bio-cultural characteristics (Agriculture)
Domesticated animals and cultivated plants
C. Personal-cultural characteristics (Psychology)
A person's knowledge, feelings, and actions (i.e. all behavior) as far as these are conditioned by other. people (i.e. acquired in one's own lifetime).
D. Group-cultural characteristics (Social sciences)
Combinations of the above characteristics, organized into systems called "institutions- for social, control. The eleven chief institutions are
1. Domestic – the family _and the_ home
2. Scholastic – the school, library, etc.
3. Economic – farm, factory, shop or office
4. Political – the government, parties
5. Religious-Ethic – the temple, synagogue, church,. mosque
6. Hygienic – the hospital, clinic, etc.
7. Philanthropic – charity organization, social welfare and insurance
8. Recreational – the cafe, theatre, playground, etc.
9. Artistic – the museum, music hall, garden, etc.
10. Scientific – laboratory, institute, etc.
11. Communicational – the press, the radio, the forum
This classification of characteristics divides them roughly into; the physical sciences, the biological sciences, and the social sciences, or into characteristics roughly dealing with matter, with life, and with man. The cultural differences are man-made ones, i.e. any physical or biological phenomenon which man has modified in any Way. Culture is defined in Sociology and Anthropology as all the knowledge, customs and equipment produced by man. Out of all the cultural characteristics, however, Part II of this book will only those characteristics which interrelate people to each other in each of the eleven chief institutions. As pointed out above, this course is limited to the social sciences which study the social institutions comprising class III. D of the classification above. An institution is a segment of culture, a part of the civilization of a people. An institution can be defined in cultural terms as the knowledge, customs, and equipment which interrelate people in a group in achieving a major life-purpose.
The Institution of the family (Chapter 3-6) is all the knowledge, customs and equipment of the home interrelating the parents and their children in creating and rearing a new generation.
The educational institution (Chapters 7-8) consists of all the knowledge, customs and equipment used by teachers and pupils in transmitting the culture of the old to the young.
The economic institution, (Chapter 9-13) consists of the knowledge, the customary ways of behaving, the symbolic and material equipment which interrelate people while earning their living. The economic institution is made up of what people know, do and use together in producing goods and services for exchange in satisfying wants.
The political institution (Chapters 14-16) is all the knowledge, the customs and the equipment in use by citizens and government officials, (who are political specialists) in governing a people.
The religious institution (Chapters 17-20) consists of all the knowledge, including beliefs, the customary attitudes and conduct, and the symbolic and physical equipment of the group of adherents who find in their religion the supremely valuable meaning of life.
The health institution (Chapter 21) is all the hygienic and medical knowledge, the customs for healing and preventing sickness, the equipment of pharmacies, of hospitals and of other health agencies used by medical specialists and laymen in achieving a great degree of physical and mental vigor.
The philanthropic institution (Chapters 22-24) is all the knowledge customs and requirement for dealing with the unnormal classes such as orphans, criminals, the insane, the suicidal, the feebleminded, paupers, prostitutes and perverts, vandals, refugees, flood and famine victims.
The recreational institution (Chapters 25) involves whatever we know and habitually do and use in recreating or refreshing ourselves. in leisure hours.
The aesthetic institution (Chapter 26) similarly includes whatever we know and customarily do and use in enjoying whatever we consider beautiful.
The scientific institution (Chapter 27) is all that we know, do, or use in order to predict and control phenomena of any sort. What we "do" here is called "research", or applying scientific methods of observing and inducing.
The communicational institution (Chapter 29) is our knowledge and habits of language and the modern mass methods of telling things to people through the printed page, movie film, or broadcast.
One further element in defining an institution should be emphasized. The essential characteristics of institutions are the interrelations of people. The persons in an institutional group interact with each other. They stimulate and respond to one another. One buys from another who sells. One teaches another who learns. A doctor operates on a patient. The players in a game each interact as the ball or his turn conies to him. The persons in each institutional group are interrelated to each other in complex ways. In each institutional group some persons specialize in carrying on the work of that institution for other people. Teachers specialize in educating the young, thus freeing the parents during school hours. Doctors specialize in health: religious leaders specialize in religious functions for their followers and government officials specialize in ruling all the people. We depend on specialists for almost .everything we know or use. The specialists in turn depend on their clients in carrying on their own specialty. The essence of each institution is the interdependence or interrelations of the people in it. This is why the title of Part II of this hook is : 'People with interrelating characteristics – Institutions".
D. The third sector—space. A prospectus of Part III.
After studying the eleven chief institutions in sociology we shall go on to study combinations of these institutions in local communities as in a village, a city, or a nation. These different kinds of regional communities are classified according to the relation between people and area. If the number of people is divided by the area of the land they live upon, the result is the "man-land ratio" which is also called the population density :
Persons/ Square kilometers = Density, or P/K2 = D
This ratio is an index of the type of civilization of a region from most primitive, to most complex. In a hunting and pastoral culture such as among the nomadic Bedouins in the desert, the ratio. is around one person per square kilometer (D = 1). In primitive agriculture, such as is found among the fellaheen of the plains, the ratio on an average, is 2. It is larger in a commercial culture and rises to more than 200 persons per square kilometer in a highly industrial country. The size of this man-land ratio is a fundamental index to the type of community. In Part III we shall study nomadic, village and city communities, each in turn. Then we shall study .regional communities where both the population and the area increase in size, first in nations, and finally in continents, where we shall study briefly, the mixing of the cultures of the East and of the West.
E. The fourth sector–time. A prospectus of Part IV.
Up to this point, our study of society, analyzed by the preceding three sectors, has been a study of current society, of people as they live together at present. In Part IV of this book we shall go on to survey society as it moves and changes in time – its past, its present, and its future.
In Chapter 38 on History. our past development – the birth of our sun and our earth, the evolution of life, of man, and of civilization – will be surveyed. When and how was our earth created?. When and how did life begin upon it? When and how did man appear? "When and how did modern society come into being. Questions such as these will be discussed briefly, with suggestions for studying them more fully in advanced courses.
In Chapter 39, we shall study the social processes and the social forces that operate in society at any time. Processes such as how men compete or cooperate, conflict or unite in all their institutions and communities, will be studied. The meaning of "social forces" and "social control" will be explained. An exact, verifiable and systematic theory for making social forces measurable will be presented. A little understanding of these dynamic processes and forces in society will be sought in this chapter which is an invitation to deepen this understanding later on in advanced courses in Sociology.
Finally in Chapter 40 on "Social Planning", we shall study a little the future of society. We shall look at some of the visions of the ideal society that men have dreamed and worked for. What is the ideal society? What is your vision of it for your own people and for all people in the world? What are the means towards reaching our social ideals? Such prediction and control of man's future is the purpose of science and of .citizenship, and to deepen the student's understanding of social science and of citizenship, is the purpose of this book.
It is suggested that this introductory chapter should be reviewed during the course and carefully restudied at the end of the course, and upon finishing the book. Much of what has been said here will be only partially grasped by the beginner. Its fuller significance will be better grasped after studying the rest of the book. This chapter is both an introduction and a review.
Light from the Near East
George A. Lundberg
Columbia University
A Controlled Experiment on Rural Hygiene in Syria by Stuart C. Dodd, Beirut, Lebanon Republic, American University of Beirut Press, 1934. 536 pp. Also available from Near East College Association, 50 West 59th Street, New York City, New York
I regard this work as one of the finest sociological monographs which has appeared up to the present time. In its own particular field, namely, the technique of measuring social change, it is beyond question superior to anything heretofore published. To one who has had occasion recently to declare that the entire literature of social surveys, including his own contribution, to that subject, is of very minor scientific significance, it is a great pleasure to come upon, in this unpretentious volume, a model of what a scientific piece of work in this field should be. The full merits of the book can be realized only through careful study of it. But I shall call attention to the two main grounds upon which I accord it the above high estimate, namely,
1. the rigor of its scientific procedure and results and
2. its contribution to a phase of fundamental sociological theory.
The problem which Professor Dodd undertook was to measure the hygienic status of the population of certain Syrian villages and the change in that status over a period of years. That purpose is stated in the first paragraph. Th: author then proceeds to define precisely all the units and terms to be employed in the achievement of his purpose. He next addresses himself to the problem of constructing a scale with which to measure hygienic status. The theoretical specifications of such a scale with respect to validity, reliability, and practical administration are first discussed with a detail and a lucidity and yet with a brevity that it would be impossible to improve upon. A trial schedule is then prepared and rigidly tested on all the points stated in the theoretical specifications. In addition to the usual test of sampling error, the schedule is tested for seasonal error, for reliability of the informant, reliability of the interviewer, error in scoring, ambiguity of the schedule itself, objectivity and quantitativeness of the questions, and the ease and simplicity of the administration of the schedule.
It is impossible to give an adequate account of the ingenuity-, perseverance, and meticulous care with which the schedules and scales were constructed and revised to conform to the requirements of a scientific instrument. Anyone interested in techniques and methods of sociological study will find here a fascinating story of weeks spent in villages of Egypt, Palestine, Iraq, and Persia to gain first hand cultural background; the painstaking formulation of lists of items, patterns, practice, or environment probably related to health; the preparation of a trial list of questions on these subjects; the thorough criticism of this list by clinic physicians and nurses as well as by others; the trial of the schedule in the field and resulting revisions. From this process there emerged the first complete form, a schedule of 2.70 questions with approximately 1,000 possible answers for checking. A detailed manual of instructions for interviewing with this schedule was also prepared. It was then administered to 345 families in order to gain data for still further improvement of the instrument. Frequency graphs for the scores of each of the 270 questions for each of six sample populations were then studied to locate ambiguous and subjective questions and other defects in the schedule, as well as for characteristics of the populations to be studied and the reliability of the procedure in general. On the basis of this experience and these data, subjected to the most searching analysis, the final schedule, abbreviated and corrected for all the defects of the first, was finally formulated. In short, no pains were spared to make the instrument as perfect as possible under the circumstances; to measure the unreliabilities which remained; and, finally, to report every step in the process. The full description of this procedure will stand for a long time as a brilliant example of scientific technique in sociology.
As a piece of scientific reporting alone the book is notable. I do not recall having encountered in the whole literature of social research a superior account of exactly what was done, the results achieved, and a full exhibit of the instruments used. Sociologists will do well to consider the book from this point of view alone. Those who have been urged by their university presses to eliminate even the more important statistical material and who have been compelled to exclude entirely an adequate methodological ac- count, thus violating the most elementary and fundamental scientific requirement, will appreciate the unusual character of Dodd's work from this point of view alone.
While the methodological significance of this book is of far greater importance than any contribution to the knowledge of contemporary hygiene in the Near East could possibly be, the author never allows himself to become so fascinated by methodological considerations as to forget his practical objectives. As a result, we have here also an excellent account of one of the culture complexes of Syria. I hasten to reassure those who have been discouraged by my emphasis upon the scientific aspects of the work that there is also a score of excellent photographs and much juicy "case" material. The important difference between Professor Dodd's use of this material and its current use in surveys is that Dodd is fully aware of the purely illustrative and entertainment value of the episodes portrayed and at no time confuses them with his scientific data or makes random incidents the basis of his conclusions. This "warm," "human," "interesting," "readable" material is always kept in its proper place with reference to the rigorous requirements of scientific methods. When used in this way such material is, of course, very valuable. But too often a serious study is so deflected and diluted by a desire to cater to tabloid tastes as to destroy its scientific value. Journalists, novelists and other sociological clairvoyants which are all too frequently scientifically illiterates are still regarded as proper review and critics of sociological monographs. Their lamentations about "worlds on paper," "encountering no human beings," and the failure of sociological writings to conform to the stereotypes of contemporary cartoons, are taken quite serious and are not infrequently used by sociologists themselves as legitimate sociological criteria. Obscurity should never be confused with profundity, but neither should scientific rigor be sacrificed to popularization. There is no more reason why a scientific treatise on sociology should be pleasant Sunday afternoon reading for the scientifically illiterate than that a work on physiology or chemistry should appeal to the lay public. In this respect again, Professor Dodd's treatise is a fine example. It is a marvel of lucidity and coherence on its own level. But it makes no cheap compromises with tabloid journalism in either style or content. It is a brilliant illustration, also, of how a statistical approach results in the only type of insight into a culture complex which is of any scientific significance.
To those who regard the field of social measurement as but a naive and passing aberration, even the high quality of the work under review will seem of little sociological importance. To those who believe that a large part of future sociological research must consist of a painstaking scientific check of the generalizations which today fill the Introductions to Sociology, Dodd's book will be an encouragement. It may be that a hundred years from now sociology will still be nothing more than a formidable collection droll stories, impressionistic generalizations, and windy dialectics. On the other hand, it may be that we shall then have a substantial set of generalizations the probability of which, under given conditions, can be stated with some mathematical accuracy and reliability, as it be the case in every true science. Professor Dodd's volume proceeds on the later assumption. It demonstrates a type of scholarship which does not rest entirely upon the ponderousness his bibliography, and which is as yet not very common in sociology. Nor will the customary charge that studies of this kind neglect the theoretical side apply to Dodd's work. "The Theory of the Measurement of Social Forces" submitted in the Concluding Part (Part IV) is one of the ablest and most stimulating treatises that has yet appeared on the subject. Briefly, there is here submitted a logical theory for the reduction of the concept of social forces to an equation based on measured entities.
I have touched here on only a few of the original and ingenious features of this unusual and excellent monograph. Nor is a more detailed account necessary, for the intrinsic merits of the book will sooner or later compel recognition. In so important a work, however, even minor flaws in the way of misprints must be noted: The correlation coefficient in the second paragraph on page 73 should be – .30 instead of +.30. The formula at the bottom of page 210 should read
Ai–iv = (((Siv – Siii)/Yiv–iii) – ((Sii – Si)/Yii–i)) / .5Y(iv – iii – ii – i)
In addition to these minor and self-evident slips, an inferior job in book manufacturing must be noted with regret. But these flaws shrink to insignificance when the extraordinarily high quality of the work as a whole is considered. No student or teacher of research methods can afford to neglect this model monograph.
A Controlled Experiment on Rural Hygiene in Syria
A Study in the Measurement of Rural Culture Patterns and of Social Forces
Stuart Carter Dodd
Associate Professor of Sociology
American University of Beirut
Beirut, Lebannon
[pic]
American Press
Beirut, Lebanon Republic
1934
Table of Contents
Page
Preface..................................................................................................................................vii
Abstracts of Part I-IV .............................................. ...............................................................x
PART I: A Hygiene Scale –The Measuring Units....................................................................1
Definition of Hygiene and the Unit, s..........................................................................1
A. Construction of Form A..................................................................................................1
1. Specifications to be met ..........................................................................................2
2. Techniques of construction.......................................................................................7
3. Techniques of application........................................................................................12
B. Studies on Form A .......................................................................................................20
1. Item analyses...........................................................................................................20
2. Reliability analyses .................................................................................................24
3. Validity analyses......................................................................................................30
4. Summary as the degree to which Form A meets the specifications .......................30
C. The Revised Scale—Form B ........................................................................................34
1. Procedure in revision................................................................................................34 2. Exhibit of Form B ....................................................................................................39
D. Studies on Form B.........................................................................................................59
1. Plotting of distributions for comparative samples.....................................................59
2. Reliability studies.....................................................................................................62
a. Sampling error...................................................................................................62
b. Seasonal error...................................................................................................63
c. Informant error...................................................................................................63
d. Interviewer error.................................................................................................67
e. Scorer error .......................................................................................................68
f. Schedule error ...................................................................................................69
g. Form of equation expressing all errors in the scale ..........................................69
3. Validity studies .........................................................................................................72
a. Correlation of hygiene score and health indices .................................................72
b. Correlation of hygiene score and income ...........................................................76
c. Intercorrelations of hygienic culture complexes ..................................................74
4. A brief scale ................................................. ............................................................81
5. Summary of the extent to which Form B meets the specifications............................82
PART II: Hygienic Status—The Measurement of Sample Populations....................................86
Definition of "Hygienic Status", Si
A. Description of the Sample Populations ..........................................................................87
1. Geographic situation ................................................................................................87
2. Demographic situation .............................................................................................88
3. Historical situation ..................................................................................................89
4. Economic situation ..................................................................................................92
5. Religious situation ...................................................................................................96
6. Domestic situation ...................................................................................................97
7. Educational situation ...............................................................................................98
8. Recreational situation ..............................................................................................99
9. A natural area .........................................................................................................99
10. Bika sample...........................................................................................................100
11. The village unit ....................................................................................................101
12. The urban sample .................................................................................................102
B. Comparative Rural-Urban Distributions of Hygiene Data from Form B.......................104
1. Vital. Statistics (Section I) ..........................................................................Tables 8-23
2. Remedies for Sickness (Section II)............................................................Tables 24-31
3. Infant Hygiene (Section III) .......................................................................Tables 32-37
4. Food and Cleanliness (Section IV)............................................................Tables 38-44
5. Insects (Section V) .....................................................................................Tables 45-52
6. Housing (Section VI)..................................................................................Tables 53-65
7. Sex hygiene (Form A) ........................................................................................168
8. Seasonal differences (Form A) ..............................................................................169
C. Summary of Rural Hygiene .........................................................................................170
1. Outstanding good features of rural hygiene ..........................................................170
2. Outstanding unhygienic features of Alaouite rural life compared with the urban sample ...................................................................................................................172
PART III: Hygienic Progress—Remeasurement after a Time Interval ...................................176
Definition of Progress, C = Sii — Si, Sii > Si - ........................................................176
A. Evidence that increase of hygienic score constitutes "progress" for these villagers........177.
1. Verbal evidence.......................................................................................................177
2. Behavior evidence...................................................................................................178
B. Plan of a controlled experiment to produce and measure progress............................... 179
1. Outline of the experiment.........................................................................................180
2. Relation to a larger project to change rural culture..................................................180
3. Criteria for equating the experimental and control villages......................................181
C. The progress produced..................................................................................................184
1. Qualitative evidence of progress from verbal testimony and other indicators...........184
2. Quantitative evidence of progress.............................................................................187
D. Conclusions from the data on progress .........................................................................205
PART IV: HYGIENIC FORCES—Measurement of the Factors Producing Progress................207
Definition of Force, F = AP .... ............................................................... 207
A. A theory of the measurement of social forces ...............................................................208
1. A proposed system of concepts defined by equations ........................................... 208
a. The three basic units ...........................................................................................208
1) Population, P .............................................................................................208
2) Indicia, I .....................................................................................................208
3) Years, Y .....................................................................................................209
b. Compound units ..................................................................................................209
1) Change, C ..................................................................................................209
2) Rate of Change, R........................................................... ..........................210
3) Acceleration, A ...........................................................................................210
4) Momentum, M ............................................................................................211
5) Social force, F ..........................................................................................211
6) Impulsion, Im ..............................................................................................213
7) Other possible compounded concepts........................................................213
8) Probable errors, PE ....................................................................................214
2. Discussion of the proposed system of concepts........................................................214
a. The problem of the plurality of indicia................................................................. 214
b. Assumptions underlying the system of concepts ................................................216
c. Relation of these concepts to Eubank's system of sociological concepts ...........217
B. Application of the theory to the present experiment in hygienic forces .........................218
C. Agencies generating forces ...........................................................................................223
1. Itinerant Clinic of the Near East Foundation.............................................................224
a. Personnel ........................................................................................................ 224
b. Equipment ........................................................................................................ 225
c. Budget................................................................................................................227
d. Purposes ...........................................................................................................229
e. Policies .............................................................................................................229
f. Program ............................................................................................................232
g. An appraisal .......................................................................................................233
2. Other possible agencies and forces ........................................................................233
Appendices
I. Exhibit of Form A Schedule Card and Instructions to Surveyor.
Key for Form A and Form B question numbers................................................................239
II. Exhibit of Brief Scale, Form B ..........................................................................................277
III. Exhibit of clerical worksheets (correlation scatter-gram) and costs for this experiment..283
IV. Exhibit of distribution graphs of two Form A questions illustrating the method of comparison of samples in the 960 unpublished graphs ...................................................287
V. Statistical summary of the activities of the Clinic of the Near East Foundation: an analysis of an agency generating hygienic forces ..........................................................................291
VI. Excerpts from the Narrative Reports of the Clinic.............................................................297
VII. Hygienic Status of Bedouin Samples ...............................................................................309
VIII. Selected and Annotated. Bibliography .............................................................................313
IX. Topical Index ....................................................................................................................321
List of Illustrations
Facing
The technic of cultivating good-will for survey.................................................Page 16 and 17
Map of the Alaouite Villages in the Hygiene
Surveys....................... ...............................................................................................Page 87
No. 23. Diarrhea....... .................................................................................................Table 25
No. 37. Baths.......... ..................................................................................................Table 36
Nos. 38 and 39. Baby Care ...................................... .................................................Table 37
No. 44. Amount of Water ...................................... .....................................................Table 42
No. 45. Source of Water ...................................... ......................................................Table 43
No. 57. Food. Covered from Flies...................................... ........................................Table 47
Vermin Control ...................................... .....................................................................Table 51
Nos. 68, 69. Housing ...................................... ...........................................................Table 58
No. 77. Courtyards ...................................... ...............................................................Table 65
Abundant sunlight ...................................... ................................................................Page 170
Progress in diet and water supply...................................... .......................................Page 186
Hygienic Agencies......................................................................................................Page 233
Preface
The aim of this study is to contribute to quantitative sociology. A field of study becomes a science when it employs. the scientific method. This in general starts with observation, proceeds through induction to systematization of the phenomena, and then verifies its hypotheses experimentally. Observation, to be of the rigorous sort demanded by science, must get beyond the stage of qualitative concepts and develop quantitative units and instruments of measurement. The present study aims to do this in one little corner of the field of rural sociology —that of hygienic culture patterns in Syria.
While much of this hygiene scale is of local interest only, the analysis of errors in social surveying and the techniques for: measuring these errors may be of more general value. Similarly the system of concepts, defined in equations, leading up to the formula for social forces may prove to be useful in all the. social sciences.
Three types of readers were kept in mind in the presentation of the data of this study; There is first of all the scientific reader who is interested in the techniques and the theory primarily. There is the social worker or health officer who is interested in the practical utility of this scale in working upon hygienic problems. There is the student of local conditions who wants to know more about the life of peasants in Syria. The scientific reader will find most of interest in Part I, dealing with the measuring instrument, and in Part IV, dealing with the equations constituting the theory of social forces. The social health worker may turn to the exhibits of the scale and the sections giving administrative directions for using it and the norms that are graphed in Part II, Hygienic Status. The description of the agencies (Part IV C) which generated the social forces will also interest the social worker. The student of local culture will be best repaid by skipping the technical Part I entirely, and beginning with Part II with its general cultural background and detailed graphs with interpretative comments on each hygienic culture pattern. He may then care to go on and see the extent of progress after two years of education (Part III).
To aid readers of diverse interests in finding what they want readily, an extensive index is provided. An attempt has been made to consign less important detail to footnotes or to the appendices in order not to cumber the text with too much 'technical minutiae. A large part of the data is omitted here. All the graphs of the data from the preliminary Form A, which are not presented in this volume, may be consulted in bound volumes of typewritten manuscript in the library of the American University of Beirut. (See Department of Sociology Yearbook, Vol. III) .
A warning to the reader is needed. He should not assume that these rural samples and this urban sample represent the entire peasantry or city folk of Syria. No such claim is here made. It is probable that these rural samples represent the lowest extreme of Syrian peasantry and the urban sample was expressly selected to represent the upper extreme of Syrian society. For, if a scale can differentiate satisfactorily at each extreme, it will be adequate for the intermediate ranges. Consequently the reader is warned that the conclusions drawn apply to these particular samples and may or may not represent rural and urban life more generally. In the text and the graphs the labels "rural" sample and "urban" sample are repeatedly used. It would become absurdly wearisome to qualify them every time.
This warning is further needed in that the samples studied are very small. From the point of view of an actuary they are entirely inadequate. It was experimentally found, however, on taking further samples that increasing their size did not appreciably increase the stability of the results. With limited funds and personnel it was thought best, therefore, to make intensive studies of eight small samples. The trends discovered among these evaluated the items of the hygienic scale and refined it to a simpler and more accurate instrument with which more extensive studies on larger samples can profitably be made.
Acknowledgements
The Near East Foundation through its travelling clinic, in: charge of Miss Anne E. Slack, R.N., cooperated throughout with this controlled experiment. As the agency conducting the Welfare work, it generated the social forces and produced the progress which this study sought to measure.
Appreciation is expressed for the courtesy of the officials, who facilitated this study—M. Burnier, the Commissioner for Refugees from the League of Nations, and Captain Vuilleoucl, the French resident Advisor in the province of Masyaf.
The author of this study is greatly indebted to the following students and to others who as surveyors gathered most of the data during the six expeditions :-
Drs. A. Aboushdid, Kassabian, and A. Mufarrij; Misses. X. Calliondji, L. Hakim, H. Howie, M. Howie, H. Inedjian, E. Nasir, P. Rubeiz, E. Senekjian, V. Shaghougian, and H. Witzeman; Messrs. A. Badeer, E. Epstein, G. Farrokhzad, H. Ghalib, F. Khuri, M. Melkonian, Y. IVIugrditchian, R. Salbi,. T. Sallum, and Z. Shakashiri; and Mr. and Mrs. J. Kunzler.
Abstracts
Abstract of Part I. – A Hygiene Scale
Hygiene in this study means all the knowledge, practices; and environmental conditions, which are under the control of the family and which tend to increase health. It is more precisely defined by the questions of this "Hygiene Scale for Families."
Form A of this scale, comprising 27o questions which were scored on a r000 point scale, was constructed. It was administered, by techniques suited to the Near East, to 345 village families. The analysis of g6o frequency distributions, of the resulting data led to the selection of 55 questions for the revised Form B. This form was given to various rural groups, totaling 256 families.
The scale score of Form B showed a correlation of:
.76 with an index of mortality
.52 with an index of morbidity
.65 with an index of longevity .81 with an index of income
.94 with each of the 5 Sections of the scale, on the average.
Constant errors, in the form of a statistically significant difference between mean scores, were proved not to exist between different samples of plains villages studied, between different informants in one family, between different interviewers, or between different scorers.
Variable errors resulted in the following correlation coefficients in a standardized rural range:
.91 between male and female informants in the same family
.94 between different interviewers of the same informants
1.00 between different scorers of a set of schedules.
All seasonal errors were eliminated by comparing data from one season only.
The Form B schedule contains full and simple instruction& for surveying, interviewing and scoring in the peculiar conditions of primitive Syrian villages.
A Brief Scale of 20 questions yielding 500 points of score was selected for quicker and more extensive surveys. The reliability correlation of this was .90 and its correlation with the full scale was .98.
Abstract of Part Il. – Hygienic Status
The hygienic status of a group is defined as its average hygiene score. As a background to the data on the hygienic conditions in the Syrian villages included in this study, those villages are described. Geographically the seven villages are in the Alaouite State, in a plain bounded by a mountain range, barren uplands, and the Orontes River. The death rates of various samples ran well over 5o per z000, and the high infant mortality kept the average longevity well below t to years. The historical vestiges in the region and the immediate events leading up to this study are described.
The rural income is from agricultural and pastoral sources, averaging around zoo Syrian pounds annually per family ($80.00 at par) . The land belongs to feudal lords. The taxes range from 1/8 to 1/5 of the gross income. Debts are large. The houses are windowless mud huts, shared with the animals.
The peasants are mostly Shiite Moslems except in one village where they are Armenian Christian refugees. Polygyny, wife-beating, and barter for wives exist. These Alaouites are almost totally illiterate, but the Armenians average several years of schooling.
For surveying, the village is a natural, self-contained unit as only Bedouins, but no peasants, live between villages. An urban sample of well-to-do, well-educated families in the port city of Beirut are described as representing the opposite extreme of Syrian society,
Following this description of the background there are presented frequency distributions of the answers to all the fifty five questions on hygiene of Form B of the scale. These graphs show the data from zoo "normal", rural families, identified by number, compared with the data from the urban sample. The graphs are classified in the following sections : Remedies for Sickness, Infant Hygiene, Food and Cleanliness, Insects, and Housing. The unpublished data from Form A are drawn upon to add some notes on sex hygiene and on the differences between hygienic practices in the spring and in the summer, as determined from a resurvey of one village.
In conclusion, the outstanding hygienic features of the villages are summarized, and a semi-tabular comparison of the chief unhygienic features of rural life compared with upperclass urban life is presented.
Abstract of Part III. – Hygienic Progress
A social change in a population may be measured as any reliable difference between its status at one date and its status at a later date. If the change is in a direction desired by the population undergoing it, it is here defined as progress. Evidence is presented from the statements, petitions, and actions of the villagers to show that they desire change towards a higher hygienic status.
To measure such hygienic progress, a controlled experiment was attempted. The villages, whose hygienic status had been measured by a survey in 1931, were divided into an experimental sample and an equated control sample. The experimental sample was then educated in hygiene for two years by an itinerant clinic. A resurvey in 1933 measured the change. The amount of progress in the control sample was deducted from the amount of progress in the experimental sample in order to isolate the progress attributable to the clinic from that due to other agencies. The control sample of forty families showed a reliable gain of 18 %, while the experimental sample showed a reliable gain of 20 %. The excess gain of the latter is not statistically reliable. After an analysis of the detailed changes in each of the five sections of the hygienic scale, and after a discussion of the possible explanatory hypotheses, the conclusion is reached that the gain in both samples is probably due to the influence of the clinic. The control sample was not sufficiently isolated from the experimental village to prevent diffusion of the new hygienic culture patterns from the latter to the former. .
For comparison with the above samples of Arabic-speaking Alaouite villagers the progress in a demonstration sample was measured. This sample was a village of newly-settled Armenian refugees. No similar village existed to pair with it as a control so that the scientific technic was inferior in this respect. The gain in this sample was 66 %. It is attributed to the clinic and to economic factors in undetermined shares.
The annual rates of progress in the different samples and the differing rates in the more ideational and in the more material culture patterns were measured.
Abstract of Part IV. – Hygienic Forces
A system of concepts, defined by equations and based on three fundamental units, is offered as a means of measuring social forces. The three basic units are persons, years, and indicia, which measure population, time, and the amount of any social characteristic. In this study the characteristic is hygiene and the indicia are the scores of this hygiene scale.
Compound concepts are readily derived. The status of a population is its average score. Social change is the difference in status at two dates. The amount of change divided by the time required to achieve it is the rate, or velocity, of change. The rate of change of this velocity is social acceleration, the speeding up or slowing down of social change. This acceleration times the population accelerated is defined as social force. It is that which stimulates a population to speed up its social change (or the reverse) . A unit, called a "stim", is suggested, and is defined as one person changed one indicium per year per year. Further concepts of social momentum, impulsion, etc., are defined. The probable error formulae of these concepts are indicated in outline but their rigorous mathematical derivation is reserved for fuller publication elsewhere.
In a brief discussion of this theory it is pointed out that there are as many kinds of indicia as there are measurable characteristics in a society. The possibility of techniques to unify or interrelate them is indicated. The assumptions underlying this system of concepts are pointed out in the form of a paraphrasing in terms of social concepts of Newton's laws of motion. The relation of these concepts to Professor Eubank's "Concepts of Sociology" is indicated.
This theory of measurement of social forces required three surveys, but this experiment had already been carried out with but two surveys before the theory emerged from it. Consequently, it was necessary to make some assumption about the hygienic conditions prior to the first survey. Two alternative assumptions were made and followed through to their conclusions to determine the number of "stims" of hygienic force that had been generated in these villages in the two year period of this experiment.
The Clinic of the Near East Foundation and other agencies generating this force are described through a statistical analysis and through excerpts from narrative reports for deeper insight.
Dimensions of Society
A Quantitative Systematics
For The Social Sciences
by
Stuart Carter Dodd
Chairman of the Department of Sociology
American University of Beirut
New York, 1942
Copyright, 1942 by the Macmillan Company
All rights reserved—no part of this book may be reproduced in any form without permission in writing from the publisher, except by a reviewer who wishes to quote brief passages in connection with a review written for inclusion in magazine or newspaper
Published January, 1942
Printed in the United States of America
Set Up and Electrotyped by T. Morey & Son
Table of Contents
Summary of Formulae of S—Theory, S =
Frontpiece
Preface vi
Part I. The Sectors of Society, (I')
I. Science And Sociology 3
an introductory specification of the purpose of this volume as a contribution to a quantitative, systematic science of Sociology.
II. The S–Theory, S 25
a classification of societal situations by means of S-theory; its possible utility; and its verification by three techniques of testing internal consistency and by testing its correspondence to observed data; with 35 illustrative situations reduced to S-formulae.
Part II. The Characteristics Sector,
III. Indicators, I' 135
classification of the characteristics of people and their environment by completeness of observation (qualitative, quantitative, correlative) as symbolized by the exponent; by content (physical, biological, cultural, and subclasses) as symbolized by the class-script; by precision (all-or-none, ordinal, cardinal) as symbolized by the class-interval and case scripts; together with the geometric interpretation of these scripts; a sample of situation:4 used to induce the formulae of this chapter.
Part III. The Population Sector,
IV. Plurals, P 203
classification of population by qualitative characteristics as symbolized by the exponent; the class or plural script, flue class-interval script, and the case or person script, together with their geometric interpretation; a sample of situations used to induce the formulae of this chapter.
V. Distributions, I ;P 242
classification of population by quantitative characteristics and by quantitatively evaluated characteristics; "social pathology" as extreme negative valuations; a tension theory of societal action in respect to values; a sample of situations used to induce the formulae of this chapter.
VI. Correlations, I2 307
correlation as a product of characteristics (of people or of their environment); probability and causation as forms of correlation; correlations resolved into components (or factors) and their elements; 24 systematic diagrams interpreting correlation; a sample of situations used to induce the formulae of this chapter.
VII. Interrelations, P2 389
a group as interrelated people; interrelations classified by sectors, by the form of the surface of the matrix, by internal structure of the group; a sample of situations used to induce these formulae.
Part IV. The Space Sector,
VIII. Densities, L1 449
a societal classification of geographic spaces as symbolized by the four spatial scripts; societal densities as defining rural and urban sociology and ecological concepts; a sample of situations used to induce these formulae.
Part V. The Time Sector, Ill
IX. Durations, T+1 495
a temporal classification of societal situations as symbolized by the four temporal scripts; the duration of populations and of their indicators; Brief S formulae; a sample of situations used to induce the formulae of this chapter.
X. Change, T-1 528
societal processes, simple, complex, and aggregated, subclassified by sectors, exponents, class scripts, and signs; a sample of situations used to induce the formulae of this chapter.
XI Forces, T–2 736
accelerations and decelerations of change defining societal forces and societal control; a sample of situations used to induce the formulae of this chapter.
Part VI. The Synthesis of Sectors, S
XII. Appraisal of S—Theory, Sv 831
the contribution of S-theory towards understanding societal phenomena in terms of the objectivity, precision, parsimony, and comprehensiveness of the S-system of concepts with its present limitations; the contribution of S-theory towards predicting and controlling societal phenomena; with illustrative situations reduced to S-formulae.
Appendices
I. Glossary of Symbols of S-theory 899
II. Rules for Writing S-formulae 903
III. Research Suggestions 916
IV. Bibliography 923
Chapter One
Science and Sociology
I. Purpose of This Volume
Our working hypothesis is that it is possible with our present knowledge to begin constructing a quantitative systematic science of sociology. This volume presents evidence towards verifying this hypothesis. To the extent that readers judge the evidence to be adequate, the hypothesis will be positively verified; to the extent that the evidence is judged to be unsatisfactory, the hypothesis will be disproven or at least remain unproven.
The postulates underlying this hypothesis have been ably developed in Professor George A. Lundberg's companion volume, Foundations, of Sociology. We proceed to build on these scientific points of view in offering definite techniques towards further developing Sociology as a science. As the fundamental issues have been discussed by him, we shall try merely to state tersely in this chapter, the purpose of this volume, without further discussion of the philosophical and scientific issues involved. To make our purpose clear, our notion of the four terms "science," "sociology," "quantitative," and "systematic" will be briefly staled before proceeding to present a theory, or system of classificatory concepts and notation, towards accomplishing this purpose.
II. The Concept of Science (In This Volume)
A. The Function of Science
Science is seen as a means of human adjustment. It is a body of techniques, ranging from the very particular to the most generalized, by which man can make those adjustments in himself or in his environment which he considers to be more satisfactory. This continual process of adjusting can be more analytically described as composed of
a. understanding,
b. predicting, and
c. Controlling phenomena.
Insofar as we understand phenomena, can predict their future occurrence, and can control them as we desire, we are able to adjust well in our environment. This triple function of science begins with increased understanding, proceeds to predict best that which we understand most, and goes on, as far as possible, towards changing such phenomena as we desire.
In order to test the S-theory preset] fed in this volume by these three functions of science each function will be analyzed into its constituents. Thus our understanding of phenomena depends in large part upon the concepts whereby we deal with phenomena and upon the system, weaving those concepts together. The concepts must be operationally defined, as far as possible, in order that their degree of precision-to-ambiguity and their degree of mutual-exclusion-to-overlap in meaning may be experimentally measured and better concepts progressively developed. The objectivity, the parsimony, the precision, and the comprehensiveness of the sociological concepts, almost two hundred in number, which S-theory thus defines towards improving our understanding of societal phenomena will be reviewed in Chapter XII.
Here the reader is notified that these three functions of science – to understand, predict, and control – will be the standard by which this, or any other, theory should have its scientific utility appraised. Chapters II-XI inclusive present evidence. Chapter XII summarizes this evidence to enable the student to appraise the utility of the S-theory for science.
B. The Content of Science
Since the possible adjustments in the universe seem innumerable to man's limited span of attention and memory, it is expedient to classify them, dividing them up into related subfields which we call the several sciences (see Section 5, Chapter II). By specializing thus, we adjust to the vastness of the total problem of adjustment.
But science is not the only means of adjustment in our environment, for philosophy, religion, "common sense," custom, and other ways, partly overlapping, partly diverse, are also available. Science then is only one means of adjusting. What specifies this means seems to us its function, stated above, together with the process of building a science. This process is usually called the scientific method. A science is the knowledge derived by the scientific method. The method, not the content, defines the body of knowledge as a science. Any kind of knowledge – astronomical or astrological, chemical or alchemic, psychological or phrenological, sociological or proverbial – may become a science by applying the scientific method to the phenomena that are its subject matter. In these terms, "science" in this volume denotes sensory data that have undergone a process of being recorded, systematized, and verified. Each of these steps in converting sensory data into a science, requires some qualifying, lest brevity of statement breed misunderstanding.
C. The Scientific Method
The general steps of the scientific method have been variously described. In primitive form, they are Dewey's analysis of problem-solving (Reference 11) into
1. a felt difficulty,
2. its definition,
3. suggestions for solution,
4. their implications, and
5. decision and trial of one suggestion.
We shall summarize the steps in more rigorous scientific work under the four headings of Problem, Observation, Systematization, and Verification, as a generalized and somewhat over-simplified pattern from which there are many variants.
1. The Problem
Scientific work begins with some problem, be it a question to be answered, an hypothesis to test, an instrument or technic to be developed, or a field of phenomena to explore. In rigorous work the problem will be carefully formulated in words or other symbols, its limits specified, its terms defined. This we attempt to do in this volume, first in stating our working hypothesis as the possibility of constructing a quantitative systematic science of Sociology, and then in expounding each of those specifications.
2. Observation
With the problem or field of investigation defined, the next step is to observe all the relevant phenomena to get all the pertinent facts. These observations are sensory data, which are then interpreted and related to previous sensory data. Instruments, whether microscopes for bacteria, or schedule cards for societal trends, extend our senses, enabling us to observe phenomena to which our naked senses are insensitive or inadequate.
The operational procedure, what to do first, second, and third, is just as important a part of observing as are the material instruments. The instructions as to how to proceed in analyzing the unknown chemical substance, in finding a child's intelligence quotient, or a community's hygienic rating, are essential to the accuracy of the observations. Furthermore, the observations to be facts, must be objective, not subjective ones, i.e., they must be in a communicable form and agreed upon by different observers. An item of observation becomes a fact at a given time, in proportion to the percentage of competent observers who agree upon it, i.e., who respond similarly to that stimulus-situation. Recording facts tends to increase their objectivity, as well as preserving them in order that they can be tested by reobservations and communicated to others. Recording also cumulates the facts into a growing body of knowledge which is the basis of a science.
This step of collecting the relevant facts has been taken in the study reported in this volume in the form of collecting graphs, tables, maps, diagrams, and other quantitative sets of data from the literature of the social sciences during the past twenty years. Some fifteen hundred sets of such data were taken as a sample of quantitative facts which any systematic theory must fit and organize into some sort of unity. Three hundred of these sets of data, which we shall refer to as societal situations, are recorded in this volume as a sample of the facts to be systematized by some theory.
3. Systematization
The facts are then summarized. Many particular items are generalized. Principles are induced. These generalizations may take various forms, such as in
a. A concept
This is a name for a class of percepts having common characteristics. (A percept is always completely present to the senses at one time -a concept never is completely present, as it includes the possibility of other similar percepts. Thus, a visual percept can be recorded in a photograph, an auditory one on a gramophone disc.)
b. A statistical index
This is a numerical concept, a summary of many numbers in a single number defined by its computational formula.
c. A classification
This is an arrangement of concepts in a hierarchy, in which a concept is subdivided into classes and these into subclasses (see Section 5, Chapter II; Section 1, Chapter III; Section 2 and Section 10, Chapter IV; Section 1, Chapter IX). Every frequency distribution is a classification in which the classes are varying amounts of one concept, rather than qualitative subclasses.
d. A statement of a generalized relation
This may be static or functional, and either descriptive or causal. If unverified this may be an hypothesis; if ,positively verified and of utility for prediction and control of phenomena, it is called a scientific law, e.g., the law of gravity, Engel's law, that the proportion of family income spent on necessities tends to vary inversely with the size of income (see Section 47, Chapter X, for a crude presentation), the law of sampling, that accuracy of prediction varies with the square root of the size of the sample (Equation 14, Chapter. IX).
Obviously scientific laws, which are a brief, exact, and idealized statement in a mathematical formula of a relation that is true under specified conditions, vary from such highly exact statements as those about the planetary orbits, to probability statements, as in actuarial insurance. Both deserve to be called laws, provided the attendant conditions and degree of probability are specified sufficiently, so that predictions of the future can be made within known limits of error.
The generalized relation may be a causal principle such as that the typhoid bacillus under specified conditions will cause typhoid in a person, or that violation of a group's mores will cause action, i.e., stimulate a response, to preserve its mores under conditions of the violation being a maximal one and the particular mores being maximally valued in the absence of other restraints upon the group. ("Maximal" is here defined as greater than two standard deviations above the mean on the appropriate scale.) Generalized relations are of many kinds and no classification of them is attempted here beyond a remark or two intended to amplify our notion of systematizing as a step in the scientific method.
e. A theory, or system of concepts and generalized relations
This is an orderly and internally consistent organization of all the facts and their generalizations in some limited field. Thus the S-theory, which is proposed in this volume, is an internally consistent organization of about a dozen symbols for expressing in generalized form many of the concepts of Sociology, for expressing a rigorous classification of all quantitative sociological facts by means of operationally related statistical indices, and for increasing the precision of whatever quantitative generalizations can be made with our current all-too-inadequate data.
Generalizations are sought because our human capacity for responses is limited to a far smaller number than the millions of stimuli and situations in the world about us. But proportionally as we can generalize these into classes their number is reduced nearer to our capacity and we can adjust to a larger proportion of existent and potential situations. This is the psychological reason why science seeks laws. If the adjustment capacities of people were not finite, adjusting to a billion different particular situations would be as possible as to a thousand classes of situations.
But generalizations hold only under specified conditions, and these may vary. This means that generalizations are related in patterns. Hence, to use them, we need to know those patterns and this is what a theory, a system of generalizations, aims to do. It symbolizes a set of interworking generalizations so that our understanding, prediction, and control of the phenomena covered by that theory may be more adequate.
4. Verification
Generalizations related and combined into a system may be true or imaginary, and may meet the pragmatic test of enabling man to predict and control phenomena in any degree between zero and perfection. Verification is the step of determining the degree of truth, the percentage of perfect prediction and control, given by a generalization or by a theory. How adequately does this theory work? The final step of a cycle of scientific work answers this question and, of course, thereby formulates a problem for another cycle of scientific work, in trying to improve the adequacy of that theory or of a substitute one.
There are definite techniques of verification in great variety in every science. The fundamental method, to which more specific techniques contribute, is to try predicting and controlling phenomena by deductively applying the generalization (or system of them) to particular situations. If these bacilli cause typhoid, injecting them into this person will give him typhoid. Inject them and see if the predicted typhoid results. If maximal violation of maximally valued mores stimulates a group to respond by eliminating the violator with a specified probability, then, in the next hundred violations observed, elimination of the violator must follow in the percentage given by that specified probability.
Whenever the conditions, i.e., characteristics of a situation, can be varied separately by human effort, a crucial experiment to verify a theory is possible. Whenever the conditions vary but cannot be fully varied, i.e., controlled by human effort, statistical selection may be possible in order to isolate characteristics or sets of them, and so test a theory. In either type, together with intermediate variants, a deductive prediction is made from previous generalizations as to what further observations should be, whether of the manipulated or of the selected sort. The degree of agreement between these predicted observations and subsequent actual observation of the situations as they transpire is the climax of the verifying process.
The verification, however, includes more than a final testing of generalizations. It includes as corollaries, techniques which verify the previous steps and contribute to verifying the final result. Thus, the observations must be verified by repeating them with different observers and instruments, at differing times and places, and under other differing conditions. Determining the reliability and limits of the observations is part of the verifying process. It establishes the facts.
Under the systematizing step, verifying may include determining the objectivity of the concepts and of the classification in the hands of different scientists. Thus a pioneer, controlled experiment to determine the objectivity of the concepts and classification provided by a sociological theory is reported in Chapter II.
5. Operational Definitions
An important technic towards verifying is to give definitions in operational terms. Sociologists are still wedded to definitions in descriptive terms, which do not enable another party to duplicate the thing defined. A definition which tells what to do first, second, third, with specified materials, in order to get the thing defined, is for science as far superior to a mere description of its properties as a kitchen recipe for making a cake is superior, if one wants to get a cake, to a literary eulogy of its color, texture, and delicious taste. Thus, "competition" may be described and evaluated by adjectives such as "selfish," "cutthroat," "capitalistic," "individualistic struggle," etc., but we define it operationally as the process measured by calculating the standard b; deviation of percentage gains and losses of a desideratum, V, competed for among the competitors, P, in a period, T (see Equations 47a, Chapter X). This is our model for defining all the two hundred odd terms for Sociology and for the social sciences generally in this re volume. Such definitions enable another investigator to verify more exactly whether the phenomenon that is defined exists as asserted. Reliance on operational definitions has the result of so radically rearranging the traditional content of a textbook on Sociology, as to lead to dismay among many professional sociologists. Such readers will look over the graphs in this volume, for example. These graphs are classified by the operation of the exponent into the quantic classification, the degree of perfection of which will be expounded and experimentally measured in Chapter II. But as the graphs will not be grouped around familiar rubrics of "delinquency," "the family," "rural communities," "culture," and the like, they will seem to those readers to be grouped very heterogeneously. Such readers are warned that operational principles are the first demand of pure science, their application to human problems is technology and should be reserved for a later stage. We deal in this volume with human characteristics first, then with the operational combination of these with people, defining plurals and distributions, correlations of characteristics in patterns, and interrelations of people, each as defined by an operational formula. Then spatial and temporal principles in ecological and dynamic situations are studied as specified by their calculational formulae of densities, velocities, and accelerations of change. Finally any combination of these principles can be applied appropriately to any given societal problem or field.
This is what every mature science does. Physics textbooks do not study "thunderstorms" all in one place. The principles of sound waves in thunder will be studied where sound is studied with operational techniques for such phenomena whether manifested in thunder or in an oratorio of the Messiah. The principles of electric charges are elsewhere grouped and the principles of light in another section, regardless of whether it occurs in lightning or in a searchlight. These principles may be combined to explain the thunderstorm as a whole, or they may be combined by the engineer to build a subway. But the principles are best studied as grouped by similarity of operation, not by content of application. Sociologists are still too much interested in social work, in immediate application of theory to societal situation, to realize fully that a longer perspective requires putting aside applications and first building up a body of several hundred concepts and principles which are operationally defined by their statistical form (i.e., formulae) and general to any societal content. This is what this volume attempts to do. But unless the reader grasps this operational point of view, he will misjudge the societal principles and their organization throughout the rest of this volume.
Verifying of the systematizing step may further include checking statistical calculations, criticizing the formulae used, or the logic leading to the inductions or to the deductive applications. All this tends towards establishing the generalizations as truth, subject to the more final test of accurate prediction and adequate control of the phenomena generalized.
The above sketch of our notion of scientific method is an oversimplification of a typical pattern. There are many variants in scientific work on particular problems. Years of work may be concentrated on one step as in perfecting an instrument or a technic of observing. A generation of professional discussion may be largely wasted on a sterile formulation of a problem, as in the search to identify "instincts" in human Psychology. The sketch over-simplifies in that the four steps interwork and do not always follow the sequence as described above. A problem may, and c usually does, become better formulated as observations and tentative systematizing go forward. The systematizing redirects the observer to collect further facts with better techniques under other conditions, etc. Each of the steps may fertilize and influence the others in complex ways. Their intellectual isolation for discussion, however, is useful to make each step more definitely crystallized, communicable, and open to critical improvement and evaluation by fellow scientists. As an exposition of scientific method, this sketch is far too brief. However, the many implications and issues involved in its somewhat dogmatic statements have been more fully dealt with in Lundberg's companion volume, Foundations of Sociology. Therefore, a mere outline is presented here to suggest, the premises upon which the theory is based in the succeeding chapters.
III. The Concept of Sociology (In This Volume)
Sociology as the science of society, where, by "society" is meant the whole of human life, needs further restricting else Sociology would be simply the total of the social sciences. In practice, the content of Sociology is operationally defined by the studies published by sociologists, such as the members of the national a sociological associations. For a conceptual summary of this field Sorokin's definition (Reference 66, page 760) usefully draws the boundaries a around three mutually complementary fields of knowledge.
"It seems to be a study, first, of the relationship and correlations between various classes of social phenomena, (correlations between economic and religious; family and moral; juridical and economic; mobility and political phenomena, and so on); second, that between the social and the non-social (geographic, biological, etc.) phenomena; third, the study of the general characteristics common to all classes of social phenomena."
Of these three delimited fields, the third is our chief concern for a systematic Sociology. The first field of border zones and. relations between more than one, but less than all, of the social sciences, is a partial step towards the third area of studying "the general characteristics common to all classes of social phenomena." This last will be our working definition of Sociology. The general characteristics of time, place, people, and the many other characteristics of people and of their environment will be the subject matter of our sociological system, regardless of whether these general characteristics appear in economic, political, religious, educational, recreational, medical, or other traditional classes of social phenomena. We do not thereby exclude the first two subfields from the general field of Sociology.
The general theory of the science of Sociology should be the system of generalizations which are general to all societal phenomena and so to all the social sciences, and should not be limited to any one or pair, or triad of them. This does not mean that Sociology is a synthesis of all the social sciences in the sense of including the content, principles, and approaches of them all. Each has its own. Sociology is viewed here as one of the social sciences and as concerned with the greatest common denominator of the others, i.e., with what is common to them all, neglecting what is specific to each. Thus, processes of co-operation, accommodation, conflict, and competition occur in economic, political, religious, recreational, or other fields of societal phenomena, and hence belong to Sociology by our definition. Similarly, such concepts as association, differentiation, density, isolation, contact, interaction, group, plural, culture, progress, forces, and control, are common to all classes of societal phenomena.
Along with Sociology the related terms "society," "societal," and "social" need defining as to their usage in this volume. "Society" will denote human beings living together. Ordinarily as a collective noun it will connote the whole of living humanity. In the plural it denotes humanity at different periods or large relatively self-sufficient portions of it. "Group," "plural," and "community," will be used with more exact definitions (as specified in Chapters VII and VIII) for portions of humanity, while "society" will be used as a looser, broader, but more convenient term for people-in-association without specifying how many people or how much association. "Societal" will be the adjective meaning "of, or pertaining to, associated human beings." It will be consistently used in place of "social" which will be discarded (except when quoting from current literature), because "social" denotes non-human as well as human phenomena, and because it has become laden with many diverse and irrelevant connotations such as, "pleasant fellowship," being "extroverted," "fashionable," etc.'
In connection with defining Sociology the reader should be informed at the start that the theory developed in this volume may transcend Sociology as defined and be applicable to any or all of the social sciences. The twelve distinctive concepts of this theory can be developed for the specific content of any of the social sciences. The emphasis in this volume, however, is to develop these twelve distinctive concepts to fit the facts and concepts current in the sociological literature.
IV. The Concept of "Quantitative" (In This Volume)
A third specification of our working hypothesis after "Science" and "Sociology" was that our systematics of Sociology was to be quantitative. At first thought the distinction between the qualitative and the quantitative seems obvious, but there is a controversial border zone through which we propose to draw an exact boundary.
A continuous development of increasing precision in human observations can be described from observations that are qualitative, such as "white," "woman," "intelligently," and "cooperating," to their quantification in some way, as in degrees of whiteness, femininity, intelligence, or co-operation. Briefly, our theory of precision in measurement (which is elaborated with symbolic notation in Chapter III) is that human observation typically begins with distinguishing and naming a quality. Its presence or absence then makes that quality a variable, varying between two points, "all" and "none." Finer differentiation follows as in the use of the comparative and superlative degrees for any adjective or adverb as suggested by such terms as "more," "most," "less," "least." Qualities can be quantitatively compared as unequal before any units have been discerned; as when we judge that Miss X is less beautiful than Miss Y, or that W is pleasanter than Z. From ranking the degrees or relative amounts of qualitative entities, human observing proceeds to define standardized units wherever possible, so that ordinal series (ranks) can be converted into cardinal series, which are multiples of standardized units. Many phenomena occur with such obvious units as to be readily counted and expressed in cardinal units with little or no intellectual labor to define the standard unit. Cardinal -series, in turn, are further refined in precision by calibrating them in various ways as described in Chapter III. In this series of steps from the purely qualitative to precise quantities of the qualitative,' we draw the boundary line between the qualitative and the quantitative where the constant with only one value becomes a variable with either of two values – all or none. The presence or absence of a qualitative entity is conventionally assigned numerical values of unity or zero, and this becomes a primitively observed quantity. Every percentage that has ever been computed is an arithmetic mean of such an all-or-none quantity – a fact which should assure such all-or-none variables being included in the category of the quantitative.
There is no attempt in this scheme to lump ordinal scales, such as attitude tests where the step-interval between statements of attitude may be unequal or unknown, with cardinal scales, to give an impression of exactness which has not been achieved in fact. On the contrary our S-theory explicitly provides separate notation for ordinal and for cardinal series in order to clearly classify quantified series into varying degrees of precision in quantification.
This theory of measurement,' in addition to delimiting "the quantitative," serves to define a societal dimension – the theme of this volume as suggested in the title. A quality, as long as it remains unchanged, can be thought of as a point. It is just some one thing, representable geometrically as a dimensionless point in societal space. A point is not quantitative in itself but let that point be taken along with another point and immediately these two points determine a line. Let the second point, for reference, be an easily ascertained one, such as the absence of that quality (which is the same as its presence in zero amount). The presence and the absence of the qualitative entity are two points determining a linear dimension. Other points may be unobservable but can be inferred to be possible.' Thus, when the frequency of occurrence of such a quality in some specified universe is observed, the percentage of actual to possible occurrences specifies an intermediate point between complete absence and complete presence. Next, observing that quality in ordinal series, as more or less of it, specifies further points and intervals on that dimension. Finally, cardinal units specify equal and interchangeable line-sects which can be laid off end to end along that dimension and counted to measure lengths along that dimension, i.e., quantities of that quality. One can count the units, from any given point as origin, in either of two directions along the line that visualizes a dimension—hence, the name dimension. A dimension may be roughly thought of as a line, or more exactly as a linear magnitude. A societal dimension represents some societal qualitative phenomenon either in its primitive form as a point (relative to a zero point which is its absence) or as a line with distances along it specified with any degree of precision. The direction of the line can represent (as will be elaborated in later chapters) the quality, the kind of phenomenon, while its length represents the quantity, the amount of that phenomenon. These directions and lines can be combined by addition, subtraction, multiplication, etc., by the rules of geometry applied to societal dimensions as well as to physical dimensions.
A societal dimension, then, may be roughly thought of as any societal phenomenon whose amount is determinate at least to the extent of determining its presence or absence. It is any measurable societal phenomenon. In its incipient, (i.e., dimensionless) stage of observation it can be any observed societal phenomenon whatever. The fuller study of these societal dimensions, their more specific definition, their classification in an orderly system, with operational techniques for determining their number, their lengths, their directions, and their combinations, in any observed set of societal phenomena, is the central theme of the ensuing chapters. (For one example, see the general formula for the number of dimensions in a situation given by Equation 35, Chapter III)
For an adequate exposition of the underlying postulates concerning measurement, units, and quantities in science, the reader should study again these chapters in Lundberg's Foundations of Sociology. (Reference 43)
The progress of a science is marked by the increasing precision of its data (by which we mean the recorded observations of phenomena). The prestige and the utility of the exact sciences are in part, a result of increasing accuracy in observing and measuring their phenomena. The steadily increasing dependence of sociologists on instruments yielding tabulatable and quantifiable data and on appropriate statistical techniques for summarizing and for discovering relationships in their data is a healthy progress towards a more exact science of Sociology and away from its earlier status as a literary and philosophical discipline. As a contribution intended to accelerate this development, the S-theory presented in this volume deals with that part of societal data which is quantifiable. The theory covers much that has not yet been quantified and it has far-reaching implications for qualitative data. But its express field is societal phenomena that can be observed and recorded in quantitative form. It is thus limited to part of the field of Sociology – how large a part and how rapidly increasing a proportion remains for time to tell.
Because the field of this theory is quantifiable societal data, the quantitative may seem to some readers to be overemphasized. This may be as inevitable as that a textbook on Psychiatry seems to overemphasize the abnormal and neglect the normal which is outside its specified field. The reader should be assured that the qualitative data is not to be underemphasized. All data is qualitative in part. Upon the excellence of its qualitative analysis, classification, etc., the fruitfulness of all further treatment depends. Our contention is simply that to stop with qualitative data is to rest content with primitive, or incipient, observation. Qualitative phenomena need to be ever more accurately observed, yielding quantitatively observed qualities, as science progresses towards its goal of more exact prediction and control. Swift insight, inspiration, and imagination discerning new qualitative phenomena are essential at the frontier of a science, but so are plodding mathematical logic, measurement, and checking, if the ground gained is to be consolidated and permanently held. Subjective insights and objectifying techniques, in order to achieve and test those insights, must supplement each other. In this teamwork, the present volume aims to contribute more to the objective and quantitative aspects of Sociology in the belief that these have been underdeveloped hitherto. This will necessarily seem to limit this volume to a part of the entire field of Sociology.
V. The Concept of a System (In This Volume)
The fourth specification of our working hypothesis calls for clarifying our concept of a system. A system, as suggested above in our discussion of systematizing, consists of an interdependent set of generalizations, an organization, or pattern, of principles which condition each other. In the present study our system is built out of the symbols for twelve distinctive concepts (plus the four operations of arithmetic). The combinations and permutations of these sixteen basic concepts will be shown to define by formulae more than four hundred derived concepts which summarize and comprehensively classify quantitative societal phenomena.
This systematic theory is largely a methodological one. Four of its concepts deal with content, the dimensions themselves, in naming the kinds of societal phenomena. Four other concepts (the "scripts ") modify these in describing their number, precision, and classificatory relationships. The other basic concepts are "operators," stating the procedures by which the previous concepts are combined. The theory tells little of what relationships to expect between phenomena. One cannot solve for unknowns from it without further data. The equation which specifies it is a descriptive rather than a calculative one. It attempts to describe in operationally defined terms the situations as observed. It takes whatever data the observer records, good, bad, or indifferent, and describes in definite symbols the operational degree of precision of those data, tells how they may be classified, and prepares them in standardized and parsimonious form ready for further manipulation to discover deeper relationships in those data. The function of the theory is thus, largely, to improve methodology systematically, more than to immediately state a system of generalizations about the behavior of societal phenomena. It is a systematic way of expressing societal data, and not, directly, a system of the functionings of societal phenomena. It does not seem possible to deduce theorems and corollaries about phenomena from this theory alone. The theory simply converts societal phenomena into a system of data; it expresses the recorded observations of phenomena in a more orderly arrangement.
The system here lies in our organization of data, in our arrangement of concepts, and is not to be thought of as "in the phenomena." The systematizing is a form of human response to stimulus situations and, as Lundberg has clearly pointed out, is not in the stimuli themselves. For this systematizing type of response to our environment the indispensable tool is the symbol. Symbols are signs enabling people mentally to manipulate phenomena whether present or absent, real or imaginary, past or future, few or many, well or ill understood, with greater power than if people were limited to responding only to phenomena which were currently stimulating their sense organs.
The chief type of symbols is words. Language, especially when written, is the most flexible and complete symbolic system man has developed; but for scientific purposes ordinary language becomes inadequate. Symbols of greater precision and objectivity are required in building an exact science. In teaching Sociology, the unstandardized and variable meanings, the emotional connotations, the profusion of terms of overlapping meaning, are bewildering to the student.' The author's experience in teaching, in English, students brought up in a dozen different languages, has emphasized the need of a more internationally standardized and precise symbolism. After ten years of experimenting with different types of symbolism, mathematical notation stands out in his experience as the most powerful, the most universal, and the most precise."
To adapt mathematical notation to societal data the most useful branch of mathematics is statistics and an extension of matrix algebra. Since a matrix is any rectangular arrangement of numbers in rows and columns, and since almost all quantitative data in the social sciences are, or can be, so tabulated, matrix algebra is a well-adapted tool. We have extended the notation of matrix algebra with one new operational symbol (the colon) and standardized the scripts. The resulting system of sixteen symbols of an algebraic type has proved highly adequate to describe, classify, and manipulate quantitative societal data. The degree of this adequacy is evidenced by the fact that in our sample of fifteen hundred graphs, tables, maps, diagrams, formulae, and prose paragraphs, or other sets of quantitative data, like the three hundred quantitative situations reprinted in this volume culled from the postwar literature of all the social sciences, we have not been able to find one single set of data that could not be described by a formula composed of those basic symbols. We are still searching for a set of quantitative societal data which cannot be reduced to a formula in terms of those symbols. One single irreducible set of data will be more valuable in challenging the adequacy of the system which we have labeled the "S-theory" (developed in Chapter II) and in leading to its improvement, than a hundred sets that can be readily reduced to an S-formula.
Somewhat offsetting, however, the scientific advantages of a more adequate symbolism for Sociology and the social sciences is the psychological disadvantage of the difficulty of learning a new language. Algebraic symbols are difficult for many people, including a large proportion of the older sociologists, to learn and to use with facility. Over and above the inertia of mastering new associations, and the effort to form new conditioned responses, there is often a residual resistance, from early schooling perhaps, to algebraic manipulations. inferiority feelings from unsuccessful hours in mathematics classes often leads to rationalizing that mathematics is all right for the exact sciences but is inapplicable to human, or to psychic, or to social data. Resistances due to these and other causes are to be expected and the proposer of new symbols must meet it halfway, first, by making the utility or need of the algebraic symbols very convincing to the reader, and second, by making the use of the symbols very clear and simple. This we shall try to do.
To emphasize the need for better symbolism in Sociology let a few other sociologists and students of scientific methods speak. We quote from Phelps' excellent discussion of "Symbolic Sociology," in his Principles and Laws of Sociology (Reference 55)*
"It is safe to predict that sociology will be increasingly dependent upon the resources of a symbolic logic and a symbolic notation. Both of these supplements are necessary to the further study of its data as social relations or in the analysis of these data in harmony with the postulate of the interdependence of social wholes and parts. Lundberg has anticipated this need as an essential prerequisite to research in the social sciences in his statement that a standardized system of symbols is necessary if scientific facts are to be stated uniformly and intelligibly." [Reference 45, page 31]
"Symbols make two important contributions to science. They are the special instruments by which man has accumulated, stored, and transmitted knowledge. Through these human characteristics, man is distinguished from animals by the designation `the employer of signs.' The first contribution of symbolism is that human knowledge has its origins in signs, language, writing, and numerical systems. The second contribution of symbols is their aid to thought. 'Thinking,' says Eaton, `is activity, or rather the suspension of activity, through symbols.' [Reference 23] Therefore, symbols perform the double function of developing and transmitting knowledge and of representing percepts (or real objects) and concepts (or imaginary objects).
"Whenever it is possible, as in mathematics, symbolic logic, and some of the physical and natural sciences, signs or formal symbols are substituted for words and the rules of grammar. In these fields, symbols have none of the subjective colorings of words. They are objective and universal in their meaning. They have precise uses and uniform interpretations. Furthermore, they are subject to verification and proof, especially when they are employed as formulae of repeating phases of reality. For these reasons, Young has concluded that `the final arbiter of objective reality is the impersonal, non-emotional correlation of phases of reality expressed in symbols universally agreed upon.' [Reference 83.] Valid examples of such use, however, are ordinarily limited to small areas and brief periods of time. . . . Every advance in science, according to Ogden and Richards, may be described either as the substitution of some symbolic shorthand device for a previously accepted metaphysical concept or as the clarification of mystical and confusing adhesions to symbols. . . . According to Shipman's conclusions, a symbolic sociology is both possible and necessary. It is possible whenever a science is concerned with (1) limited facts and relationships and (2) the acquisition and communication of knowledge. It is necessary because no science is better than its method of symbolic representation. Moreover, literary symbols are wholly inadequate in the task of representing the multiple and simultaneous processes with which sociology deals. [Reference 62]
"Consequently, a present task of sociological methodology is the organization of a system of symbolic notation. This system must be a special sociological symbolism, adaptable to the particular nature of social phenomena, and needs not be an imitation of current developments in mathematics or symbolic logic." [Reference 55]
The argument for the necessity of a system of more exact symbols in Sociology has been further developed by Lundberg in the preceding companion volume, especially in Chapter II. This chapter could well be quoted in full at this point in our argument. The resistance to new symbolism is as old as the history of science. Such resistance can be defended as a useful sieve which refuses to pass unnecessary or ill-adapted symbols. To do this, however, the symbols must be studied before accepting or rejecting them. The example of Thomas Hobbes is a warning. Wallis submitted his work on Conic Sections for Hobbes to criticize. Instead of describing geometrical figures and proofs in Latin sentences as had been the invariable custom previously, Wallis introduced the innovation of lettered diagrams for which students since his time may well be grateful; but Hobbes sent it back with the comment that he knew riot whether the proofs were good or bad for it was so "covered o'er with the scab of symbols" that he would not even read it! (Reference 5, page 427) It is to be expected that some sociologists will react to the present volume as Hobbes did to Wallis' innovation, though opposition is more likely to be well rationalized and not so frankly based on the natural difficulty of learning something new.
History also records instances, not of apathy, but of active opposition to unproved symbols. One is the invention of the zero, which was first called a " cipher " from the Arabic word and considered so cryptic ("Why name something that doesn't exist?"), that it has left its impress after almost a millennium on the English language, where, "to write in cipher" still. means to write in an unintelligible language, i.e., to be "undecipherable." In medieval Marseilles the growing use of the ten Arabic digits which brought us the decimal system was forbidden by a law which prescribed penalties for the merchant who might be reluctant to do his arithmetic "in plain and clear Roman letters."
To be sure, the oblivion of many proposed symbolisms may be an excellent thing for science. But equally obviously, when Sociology needs a more exact symbolism, a proposed system should be studied carefully to determine whether it deserves oblivion or memorizing for use.
In addition to convincing the reader that better symbols are needed if Sociology is to progress towards becoming a more exact science, it is incumbent upon the proposer of new symbols to make them as simple and as clear as possible. Towards this aim the five following techniques have been used in this volume:
1. The symbols have been parsimoniously adopted. Several score have been eliminated as unnecessary, in the process of reducing their number to the sixteen basic concepts whose special cases and combinations yield all others in this volume. Of these sixteen only seven are new ( ::)
2. Familiar symbols have been preferred. Thus, of the sixteen basic symbols four are the familiar arithmetic operators of adding, subtracting, multiplying, and dividing (+ – X /), one is the exponent, and three are the letters T for time, L for length of geographical space, and P for the size of a population. All of our new symbols can be found on the keyboard of any typewriter. Neologisms have been almost eliminated. Even the half dozen archaic words revived here for technical use can be dispensed with by using less convenient phrases (such as "quantified indicator " for "indicant ").
3. Definitions have been explicitly given for every symbol, usually in operational form of an equation in which the right-hand member states the ingredients and the procedures to calculate the entity defined. A glossary and the topical index in the appendices enable one to turn to any definition conveniently.
4. Illustrations have been liberally included. The three hundred graphs, appended to the next eleven chapters, illustrate many times over all the symbols and most of their compounds, while evidencing the extent to which this symbolic system fits the wide range of data in the literature of the social sciences.
5. Two levels of exposition have been provided, the main text and the copious footnotes. The more technical and detailed notation has been relegated to footnotes which can be skipped by the reader who wants to understand the theory but is not trying to master it sufficiently to write formulae for quantitatively recorded situations. To use the theory fully the footnotes should be studied, but to comprehend it for a preliminary appraisal they can be ignored. Almost every algebraic formula is verbalized in a sentence following it, or in the "legend" alongside each of the three hundred situations (i.e., graphs, tables, etc.), so that any reader can, with some profit, read the entire text without having to read and interpret the notation if he so chooses. It is hoped that this device will enable the non-mathematical reader to read on and understand this systematic theory, while at the same time providing the reader who grasps the notation with a correspondingly greater power to analyze and synthesize the societal situations that are symbolized.
The specifications have now been stated which constitute our working hypothesis, that it is possible with our present knowledge to begin constructing a quantitative systematic science of Sociology. A proposal towards such a quantitative systematics is the "S-theory" outlined in the next chapter and explored sector by sector in the following nine chapters (Chapters. III – XI). The last chapter attempts a review, and an appraisal of the theory as a contribution to the science of Sociology.
* Reprinted by permission of Harold A. Phelps and John Wiley & Sons, Inc.
VI. Notes
1. For nine senses of the word social and for fuller discussion of these terms, "society" and "societal" see Eubank (Reference 25, pages 22-24 and 130-132) whose definitions we adopt here.
2. Compare Thorndike's dictum that "Whatever exists at all, exists in some amount." This basic postulate that everything qualitative is quantifiable, is more fully discussed with attendant implications in Lundberg's companion volume on postulates, Foundations of Sociology. (Reference 43, Chapter II.)
3. By measurement we denote the act of determining a quantity, i.e., the ascertaining of whether a qualitative response to a phenomenon is possible as well as the ascertaining of intermediate degrees or amounts of it. The essence of measuring is expressing in numbers. .
4. This implies non-Aristotelian logic in asserting degrees of such Aristotelian laws of logic, as that "A is A," "A is not B," in seeing that A may shade off on a continuum into B so gradually that any boundary is a purely arbitrary convenience and not inherent in the nature of A or B. For an exposition of non-Aristotelian logic – see Reference 36.
1. An impressive exhibit of the lack of agreement upon even the major concepts in current use in the sociological literature, together with a list of some 332 in occasional use, may be found in Reference 25.
5. For an analysis of the levels of increasing precision (and generality at the same time) of symbols from the verbal to the mathematical equation see Chapin's "levels of symbolic substitution" (Section 1, Chapter XII) quoted and discussed by Lundberg (Reference 43, Chapter 5).
Systematic Social Science
(A Dimensional Sociology)
Stuart Carter Dodd
Professor of Sociology
American University of Beirut
[pic]
1947
The Intention of This Offset Edition
- an invitation to colleagues.
This temporary edition is intended to get criticism with which to revise this typescript for publication.
The attempt by one person to systematize the social sciences in their present youthful state of separatist specialization and diversity has many shortcomings. Any one person's treatment of many systematizing issues will be unsatisfactory to some colleagues in other sciences. Should the symbolism here be reduced in amount and perhaps put into footnotes, or further extended with more logical rigor, and with more applications? Can this system of concepts be used fruitfully by other scientists? Should each social science be treated more and the book go beyond what is general to all of them? What improvements should be made in our attempts. to integrate method and content in social science, to present hypotheses as well as established principles, to use technical symbolic logic and statistics among undergraduates little trained in using these tools? Questions such as these make us unready for full publication at present.
Some of the shortcomings and questions can be met by incorporating the criticism of colleagues in the different social and semiotic sciences. This should make the published volume more satisfactory to all of us. But to collect criticisms this systematics must be communicated – hence this small intermediate edition of typescript. (Acknowledgment is made to a grant-in-aid from the American University of Beirut for part of the cost of preparing this typescript. Its imperfections especially in dividing words into syllables is partly due to English being a foreign language to the typists.)
June, 1947 Stuart C. Dodd
The intention of this book is to augment social science as a means. for predicting and controlling interhuman behavior towards whatever ends men value.
AUGMENTING SOCIAL SCIENCE here means increasing its basic dimensions, as follows:
Sss A More Unified Social Science.
The parts of human behavior studied by each specialized social science are interrelated. For higher indices of prediction and control in each social science other social sciences are becoming increasingly involved. This requires unifying the parts into a whole. Social Science, or general sociology, should progressively develop such a science of society as a whole. It should induce principles common to all the separate social sciences and special institutional sociologies.
Hence this systematics deals with concepts and formulas which ignore what is special to each social science and which seem equally applicable to economics or political science, to education or to anthropology.
TLPI A More Systematic Social Science.
The vast volume of facts amassed by social scientists about human relations needs better theoretical analysis. Fact finding seems currently outstripping its systematizing. Towards systematizing social science (and indeed all science),, a dimensional formula is offered here classifying every set of data (S) into temporal (T), spatial (L), populational (P), and residual (I) dimensions which are them further subclassified as needed. When expanded by their modifying scripts (s) and the operators (;) relating them, this formula, namely,
Sss = T;L;P;I
formulates the hypothesis that all data and all principles in every social science can be expressed as variant special cases of this generating master formula. All the content of this book is arranged according to this formula, affirmatively evidencing its ability to systematize. social science. The table of contents specifies this systematics, expanding the S-equation above, and relating it to the conventionally labeled fields within the social sciences.
;s A More Exact Social Science.
The social sciences traditionally have not been counted among the "exact sciences". The data of inter-human behavior have been considered too ill-defined, intangible, complex, fluctuant, or qualitative. But these defects can be reduced by research, including research in semiosis, the process of symbolizing. Such research in this book has resulted in dimensional symbols which aid in defining terms operationally and so more reliably, in making the intangible more tangible in symbols, in analyzing the complex into its more familiar parts, in measuring the fluctuations and in quantifying the qualitative. In particular, degrees of exactness are specified by the case script (s|), the class-interval script (s|), and the class script (|s). The exponent (|s) classifies all data into their qualitative (|s=0), quantitative (|s=1), and relative (|s=2) dimensions, each with their logical and mathematical operators (;) such as for sums, products, equalities, etc., for combining them with exactness. The relative dimensions include indices for correlation earlier and later events and so measuring the exactness of current social, prediction and of social control.
T A Growing Social Science.
Towards augmenting social science along the time dimension (I), this book attempts to describe its growth in terms of more known today than yesterday and still more to be known tomorrow. The scientific method whereby science grows is given much space throughout in order to teach the student how to contribute to this growth.
L A More Universal Social Science.
Towards augmenting social science in apace (L), instead of the current provincialism of many social science textbooks emphasizing one region or culture, an attempt has been made to deal with worldwide data and principles that hold in the &reties or the tropics, in the East or the West, in the desert or the city.
In trying to transcend the local, the temporary, and the limited, the generalizations, or incipient laws, about society are presented in equation form with a systematic attempt to indicate, as far as is currently known:
a. the form of the law – i.e., a hyperbolic, normal, logistic or other curve
b. its validity – its correlation, or goodness of fit, between observed data and the theoretic curve or criterion
c. its reliability – its standard error of sampling or other limits of probability
d. its degree of universality in time, space, and people
e. its other limiting conditions.
This information is currently so patchy and its full inclusion is so difficult that this volume is being put out in a small intermediate edition for critical revision from colleagues and seminar students before publishing in more finished form. Such revision seems necessary if any one volume is to portray how far social science has progressed along the road towards becoming a universal and exact science.
P A More Popular Social Science.
Towards augmenting social science by spreading it through the population (P) more fully, the technical frontier involving symbolic logic, mathematical statistics and many specialized studies in each social science has been simplified a little, it is hoped, by 170 graphs integrated with the text and by orderly exposition building up from simpler starting points. Unlike the author's Dimensions of Society (which underlies the first ten chapters of this book) all the statistical, logic, matrix algebra or geometry of n-space used herein can be learned from the text itself. By this, college upperclassmen for whom the text is designed have mastered it without previous courses in sociology, statistics, or mathematics beyond the freshman year, although any such courses would make the study of this text easier.
Towards further popularizing social science, at least in college classes, a supplementary "Teaching Manual" is in preparation which may be bound up (in part) with the 20 chapters of the text. This manual includes:
Brief annotated bibliography by chapters;
Questions by chapters, classified into:
a. "for knowledge of the text,"
b. "for provoking discussion", and
c. "for further study or research";
A reservoir of some 100 true–false test items on each chapter;
A dictionary of 300 to 400 sociological terms each operationally defined in an algebraic formula specifying its meanings as used in this systematics and as proposals towards standardized definitions.
Iv A More Useful Social Science.
To augment social science along these lines or dimensions one must believe in its value in the long run for social control and then work to realize that control. This is too large an undertaking for one person; more than one specialist in the social sciences must contribute to make the effort usefully complete and well rounded. The author is aware of many shortcoming in this intermediate edition and believes there are more of which he is unaware. Criticism by colleagues and students therefore is sought whereby to revise this edition after a year into more useful form.
In spite of defects and inadequate resources in Lebanon, however' the task was undertaken in 1940 and carried on with wartime interruptions, impelled by the thesis that: If man wants control over his social problems comparable to the control scientific research has given him over his physical problems, comparable effort must be made in scientific social research. To this end the book is dedicated.
Ainab, Lebanon. Stuart Carter Dodd
Dec. 1946
List of Plates
Chapter 1
I. Dimensions Diagrammed 24
Chapter 2
I. Diffusion of a Culture Complex 27
II. Gradients of Truancy 27
III. Gradients in Recreation 29
IV. Centralization 32
V. Density and Production Correlation 35
VI. Correlation of Georelation Surfaces 39
VII. Density and Economic Culture 48
VIII. World Population Densities 35
IX. Decentralization in Manhattan 44
X. A Georelation Surface 48
Chapter 4
I. Present and Potential Plurals---by Regions and Occupational Levels 58
II. Plurals Classified by Socio-Economic Levels 58
III. Some Plurals Classified by Nationality and Density 67
IV. A Plural Classified by Interests 66
V.-A. Plurals Classified by Sex and Age 67
V.-B. What Population Pyramids Show 73
VI. Populating Trends 73
VII. Logistic Population Growth 81
VIII. Factors of Production 85
IX. Historic Waves of Migration 88
Chapter 5
I. Model of an Interrelation Surface 94
II. Models of Interrelation Surfaces---A. Isolation, B. Contact 97
III. Interrelations---A Centralized Group Sociogrammed 103
IV. Interrelations of One Person Re-formed 103
V. Interrelations---Quantified in a "Target" Sociogram 115
VI. Interrelations in Trade of One Nation 117
VII. A Grouplet's Dimensions 119
VIII. Emotional Interrelations--Pictured Among Movie Actors and Child Spectators 121
IX. Interrelations in a village 123
X. Group Structure---An Analysis into Type Parts 125
Chapter 6
I. Isotypes 133
II. Cumulative Growth of Material Culture Patterns 135
III. Cumulative Exponential Growth of a Political Culture Complex 135
IV. Hierarchy of Culture 149
V. Inter-Pattern Relations Problems from the Analogies Type of Intelligence tests 138
VI. Joint Classes Diagrammed 145
VII. Geometrical Design of a Hierarchy of Unit Dimensions 165
VIII. Geometry of Two Dimensions 167
Chapter 7
I. Construction of An Attitude Test 171
II. Distributions Compared for Plurals and Periods 175
III. Ordinal Distributions 189
IV. Generating the Binomial Distribution 192
V. Normal distribution 197
VI. Some Properties of Distribution Curves 199
VII. Normal Diffusing of a Culture Pattern among National Plurals in a Period
(in Frequency Form) 202
VIII. Normal Diffusing of a Culture Pattern (in Ogive Form) 202
IX. A Logarithmic (or Skewed) Normal Distribution 205
Chapter 8
I.-A. Cross Classification of Three Quantitative Variables 210
I.-B. (-a. Correlation Surfaces 211
I.-C. Correlation Scattergrams 211
I.-D. Correlation of Ranks 211
II.-A. Correlation of Zones 217
II.-B. Correlation of Ratios of Sigmas 217
II.-C. Multiple Correlation 219
II.-D. Correlation Geometrically Interpreted 221
III. A Correlation Scattergram 224
IV.-A. Correlation of Time Series 229
IV.-B. Correlation of Time Series 229
IV.-C. Correlation of Time Series 230
IV.-D. Correlation of Time Series 230
V. Correlation through Common Elements 233
VII. Correlation and Its Regression Equations 236
VII.-A. Correlation as a Probability Whorl 240
VII.-B. Probability Whorls for Four Degrees of Correlation 242
VIII. A Sum of Dimensions or Vectors 242
Chapter 9
I. The Associating Process 251
II. Dissimilarizing---of Political Culture 254
III. Adpopulating 259
IV. Depopulating 260
V. Progressing 263
VI-A. Competition 265
VI-B. Competition 265
VII. Conflicting 269
VIII. Tensing 269
IX-A. Migration 275
IX-B. Migration 275
X. Diagrams of Processes Defined by Statistical Moments 281
Chapter 10
I. Celerating of the Business Equation 288
II. Economic Forces 290
III. Differential Celeration 288
IV. A Distribution of Forces 292
V. A Populating Force 294
VI-A Logistic Growth of Income 298
VI-B Logistic Growth of Empire 298
VI-C Logistic Growth of Divorcing 299
VI-D Logistic Decline of Lynching 299
VII.-A Logistic Surges in Telephoning 300
VII-B.Logistic Surges in Aeroplane Speed 300
VII-C.Logistic Surges in College Education 301
VII-D.Logistic Surges of a Fleet 301
VIII. Logistic Growth of Inventing, Wages, and Longevity Compared. 304
XI. Seasonal Cycles in Sales 313
X. Isolating a Social Force from 5 Obscuring Variables 313
Chapter 11
I. Semiotic Triangle 327
II. Vectorial Interpretation 341
III. Quantic Solid 330
IV.-A. Harmonic Distributions in Semiotics---The Rank- Frequency Distributions 354
IV.-B. The Average Number of Meanings per Word 354
IV.-C. The Number of Different Articles 354
Chapter 12
I. A Generalized Attitude Scale 359
II. Easing World Tensions 359
III. Correlation of Tension Factors 363
IV. Aesthetic Tensions Measured 365
V. Elasticity of Desire 368
VI-A Supply and Demand Curves 371
VI-B Change of Supply and Demand 372
VII-A A Harmonic Distribution of Incomes---France 386
VII-B Harmonic Distribution of Incomes---United Kingdom 387
IX. A Tension Matrix for 3 Desiderata among 21 Plurals 372
X. Tension Matrices for Various Plurals, Periods, and Desiderata 389
Chapter 13
I. Effective Social Control---of Accidents 396
II. Effective Control---of a Disease 396
III. Effective Control---of Factory Production 397
IV. Realization of Social Control---of Marriage 401
V. Realization of Control---of Women's Rights 402
VI. Diagram of Intentions, Instruments and Influence in Social Control 404
VII. Alleged Causation 409
VIII. Birth Control---Logistic Growth of Clinics 430
IX. Overlap of Factors in Social Control 433
Chapter 14
I. Physiological Origins of Institutional Behavior 438
II. Engel's Law Reconstructed 446
III. Diagram Correlating Crafts and Parts 455
IV. Intercorrelation Among Five Crafts. 459
V. Lead and Lag of Type Parts 477
VI. Cycle of Institutional Organization 477
VII. Some Spatial Correlations of Institutional and Other Indices 487
VIII. A Correlation Test of a Marxian Hypothesis 489
Chapter 15
I. Adjustees & Normality 493
II. Harmonic Distribution of Income 493
III. Conformity Distributions 501
Chapter 16
I.-A. Harmonic Distribution of Communities---United States of America 530
I.-B. United States of America over 15 Decades 531
I.-C Canada over 6 Decades 532
I.-D. India over 2 Decades 532
II.-A. Harmonic Distribution of Communities---France 536
II.-B. Germany 536
II.-C. Russia 537
II.-D. British Empire 537
III.-A. Harmonic Distribution of Communities—Europe- 540
III.-B. World 540
III.-C. "Surfeit" and "Deficit" Diagrammed 541
III.-D. Diagram of Various Harmonic Slopes or Exponents in the Generalized Harmonic
Series 541
IV-A. The PP/D Hypothesis Tested on Intercommunity Express 548
IV-B. The PP/D Hypothesis Tested on Intercommunity Telephoning 549
IV-C. The PP/D Hypothesis Tested on an Index of News 550
IV-D. The PP/D Hypothesis Tested on a Newspaper's Circulation 550
V-A.' Logistic Population Growth-- Sweden 555
V-B. Logistic Population Growth-- Algeria 555
V-C, Logistic Population Growth-- U.S.A., England, France 556
V-D. Logistic Population Growth-- Germany 556
VI. Growth of Population in Communities of Varying Sizes 559
VII-A. Logistic Prediction of Births and Deaths 559
VII-B. Logistic Prediction of Relative Age Frequencies— Solid Diagram 560
VII-C. Logistic Prediction of Relative Age Frequencies--- Plane Diagram 562
VII-D. Pyramidal Prediction of Age and Sex 562
VIII. Urbanizing 570
Chapter 17
I. Mass Time Triangle 601
Chapter 18
I. Specifications for an Electronic Statistician 605
II. Jerusalem Town Plan 608
III. Model Agenda 612
IV. Planning of Telephonic Communicating 622
V. Forecasting Business 623
VI. Economic Predicting a Quarter Century Ahead 625
VII. World Calendar 626
VIII. International Planning--- Staffs and Programs 628
XI. Planning—Finances 630
X. Popularizing 630
XI. Successive 5-years Forecasting of Total Production in Russia 631
Chapter 19
I. Dimensions of Polls 641
II. Scriptal Spiral 648
Chapter 20
I-A. S-formula & Deduction 663
I-B. S-formula & Induction 670
II.-A/B Uses of S-symbols 674
III.-A/B S-symbols -- Generalized to Linguistics 680
List of Tables
1. Institutional Analysis of the Major Life Purposes 5
Chapter 4 -- Plurals
1. The Demic Classification 55
2. Area, Population and Density of World by Continents.
59
Chapter 5 -- Groups
1. Types of Sociating 100
2. Interrelations, I, of Religious Groups 110
3. Qualitative Interrelations 113
Chapter 6 -- Attributes
1. A Logical Sum Diagrammed in a "Tree" and in Circles 147
2. A Conspectus of Logical Classes 154
3. A Conspectus of the Fundamental Relations among Logical Classes 156
4. A Conspectus of Laws of Logical Classes. 158
5. A Conspectus of the Logical Operators 162
Chapter 7 -- Distributions
1. An Example of the Intensity Function 176
2. Calculation of Distribution Moments 188
3. Binomial Coefficients, i.e., Frequencies in the Binomial Distribution (underlined) 194
4. Normal Probability 201
Chapter 8 -- Correlations
1. All-or-none Correlation, as of Rurality 1,0X, and Literacy, 1,0Y (U.S.A., 1930) 216
2. Part of a Worksheet to Compute a Correlation 223
3. Intercorrelations of Hygiene Complexes. 232
Chapter 9 -- Processes
1. Chief Social Processes 279
Chapter 11 -- Situations
1. The Scripts of the S-System 344
2. Operators as Summators 355
3. Social Quantics Outlining this System of Social Science.
Chapter 12 -- Plural Tensions
1. Limits of the Tension Factors 382
2. Occupational Tensions from Rankings by 274 College Students 384
Chapter 14 -- Organized Institutions
1. Levels of Adjustment Patterns 439
2. A Classification of the Major Institutions 442
3. A Qualitative Contingency Table between Values and Neurons 444
4. Institutional Values and Their Partial Physiological Bases 452
5. Type Parts of the Major Institutions 448
6. Agents and Clients of the Major Crafts 456
7. Voluntariness and the Senior Crafts 462
8. Sample Code for a Craft 463
9. Intercraft Patterns 466
10. A Sample Decimal Classification of Patterns (Ii) by Level of Inclusiveness 473
11. Universal Culture Patterns and Institutions 483
12. The Crafts and their Social Sciences 486
Chapter 15 -- Disorganized Institutions
1. Adjustees and their Defining Desiderata 496
2. Incidence of some Maladjustments 502
3. Correlations among Social Classes 502
4. Readjustment Classified by Moments, M 515
5. Adjustees and Readjusters 525
Chapter 16 - Communities
1. The 50 Largest Communities in the United States in 1930 534
2. Mean Epsilons 546
1. Rural Urban Differences 546
Chapter 17 -- Past Societies
1. Log Ages of History 584
Chapter 18 -- Planned Societies
1. Matrix of All-or-none Preferences 616
2. Parallel Dimensions of Social Control, Crafts and Plans 633
Chapter 20 -- Science
1. Ten Dimensional Elements of Language 683
Systematic Social Science, S = sX;s
|Section |Chapter |Quantic Formulas |Related Fields |Page |
|Part I - Analytics, X |
|A. Introduction - analyzing observable elements |X | | |
| |1. Dimensions |X |Social Sciences |1 |
|B. Environment - analyzing spatial dimensions |X = L1 | | |
| |2. Ecology |L3 |Geography, Human |25 |
|C. People - analyzing populational dimension |X = Pp | | |
| |3. Persons |P0 |Psychology, Social |51 |
| |4. Plurals |P1 |Demography |52 |
| |5. Groups |P2 |Sociometry |92 |
|D. Culture- analyzing indicatory dimensions |X = Ii | | |
| |6. Attributes |I0 |Logic, Symbolic |129 |
| |7. Distributions |I1 |Statistics, single |168 |
| |8. Correlations |I2 |Statistics, multiple |208 |
|E. Change- analyzing temporal dimensions |X=Tt | | |
| |9. Processes |T-1 |Sociology |248 |
| |10. Forces |T-2 |Dynamic |286 |
|Part II - Synthetics, S |
|F. Introduction - synthesizing recorded dimensions |S | | |
| |11. Situations |S |Semiotics |325 |
|G. Controls- synthesizing correlated forces |S = I3 | | |
| |12. Plural Control |pI3 |Economics |356 |
| |13. Group Control |ppI3 |Politics |395 |
|Crafts (institutional organizations) - synthesizing group |S = P3;I3 | | |
|controls | | | |
| |14. Organizations |+vP3 |Sociology, Institutions |425 |
| |15. Disorganization |-vP3 |Social Work |491 |
|Communities- synthesizing regional crafts |S = L3;P3;I3 | | |
| |16. Communities |m±L-3 |Anthropology, Cultural |528 |
|Societies- synthesizing coeval communities |S = T3;L3;P3;I3 | | |
| |17. Past Societies |-T-3 |History |580 |
| |18. Planned Societies |+T-3 |Teleology |602 |
|Part II - Methodics, ;s |
|Social Sciences - systematizing social situations |S = Sy;s | | |
| |19. Sociology |Sss |Research, Social |635 |
| |20. Science |Ss |Metascience |662 |
|Appendices | | |691 |
Detailed Table of Contents
Section A: Introduction - Analyzing Observable Elements
Chapter 1: Dimensions
The intention of this chapter is to interest the reader in the function of Sociology and in the implementing system or dimensions developed in this book.
I. A Sample Analysis of Society 2
A. Land 3
B. People 4
C. Characteristics of culture 4
D. Change in time 10
E. Synthesis into communities 12
II. Dimensional Analysis of Society 13
A. Analysis into sectors 13
B. The sectoral classification 15
C. Further analysis into dimensions 16
D. Synthesis into situations 17
E. Geometric formulation 18
F. Algebraic formulation 18
G. Some functions of dimensions 19
III. Scientific Analysis in General 20
A. Problem formulation 21
B. Observation 21
C. Induction 21
D. Systematization 21
E. Verification 22
F. Applications 22
Section B: Environment - Analyzing Spatial Dimensions 25
Chapter 2: Ecology
The intention of this Chapter is to analyze systematically the spatial dimensions, Lit, of social situations.
I. Space,26
II. Spatial Distribution Of Indices, Static And Dynamic 4 (I) 26
A. The Georelation Surface 28
A. Ecological concepts 30
III. People And Spatial Distribution Of Indices, P; ii; (I) 34
A. People and the distribution or space, P; r L (Demography) 34
B. People and spatially distributed physical characteristics 36
1. Earth (Land) 36
2. Water 36
3. Air 37
4. Heat 38
5. Other forms of energy 38
C. People and spatially distributed biological characteristics 40
D. People and spatially distributed anthropic characteristics 41
E. People and spatially distributed cultural characteristics 43
1. Non-human characteristics correlate imperfectly . in space with cultural characteristics 43
2. Density of population correlates with utilization of the environment 45
3. Density of population correlates with rural-urban cultures 46
4. Density of population correlates with specialization (i.e., differentiation) of culture 46
5. The size of a culture trait area correlates with transportation technology 47
6. Centralization correlates with freight transportation. Decentralization correlates with passenger transportation 47
7. The segregations of institutional extremes are correlates 47
8. Gradients of many crimes are correlates 48
IV. Summary 48
Chapter 3: PERSONS, 1P0
This projected chapter has not been written. The opinion of those reading or using this intermediate offset edition is desired as to whether such a chapter should be added to this systematics.
The plan was to describe the chief laws thus far developed in human, with the emphasis on social, psychology. This treatment of the individual person, Po, would then go on to a study of Plurals, in chapter 4 and then to groups, P', in chapter 5.
Section C --- People - analyzing Populational Dimensions 51
Chapter 4: Plurals
The intention of this chapter is to analyze systematically the primary populational dimensions, P+1, of social situations, S.
I. Larithmics – People Enumerated, P 54
II. Densities – People in Space, PL -2 59
A. The world's population – densities 59
B. Crowds – densities classified by meters and minutes 61
III. Plurals – People Characterized – I0 pp 62
A. Attributes of people identifying plurals 63
B. Dimensions of plurals 64
C. Population pyramids 65
IV. Populating – People Changing in Time, PT-1 71
A. A. Definitions and components 71
B. B, Births 75
C. Deaths 76
D. Migration 77
E. Net population growth 78
V. Optimal Population – People Evaluated, Pop 83
A. Quantitative Optima, Pop 85
B. Qualitative Optima, P. 87
Chapter 5 Groups
The intention of this chapter is to analyze systematically the secondary P2, populational dimensions, of social situations, S.
I. The Group, Defined By Interrelations 93
A. Towards defining a group 93
B. The interrelation surface 95
C. C. Isolation, contact and interaction 96
1. Spatial 98
2. Physiological 98
3. Attitudinal 98
4. Cultural 98
D. Sociation 99
II. The Grouplet, Composed, Of Persons, pP2 101
A. A. The grouplet formula 101
B. B. Intensity of a group – the amount of each dimension 102
1. Populational dimensions of grouplets 102
2. Spatial dimensions of grouplets 105
3. Temporal dimensions of grouplets 107
4. Indicatory dimensions of grouplets 108
C. Extensity of a group – the combining of dimensions 118
1. Examples of a group dimensionally specified 118
2. The number of a group's dimensions 120
3. Person-group dimensions 120
III. The Grouping, Composed of Subgroups, P2P 122
A. Definitions of grouping and groupage 122
IV. Summary of the Notation for the Population Sector 127
Section D - Characteristics - Analyzing Indicatory Dimensions 129
Chapter 6 Attributes
The intention or this chapter is to analyze systematically qualitative culture traits and other qualitative characteristics into their (unit) dimensions, iIi, in the indicatory sector, iIi, of social situations, S.
I. Characteristics 130
A. The characteristics sector 130
B. Exponential dimensions of characteristics 130
C. Class dimensions of characteristics 131
II. Culture 130
A. Definition of culture 135
B. Concepts on the ecology or culture 136
C. Concepts on the change or culture 136
D. Concepts on the organization of culture 139
1. Hierarchy of culture patterns 139
2. Analysis to culture traits 140
3. Indicators symbolizing attributes 143
III. Classes 146
A. Calculus of classes 146
1. Addition of classes 146
2. Subtraction or classes 147
3. Multiplication or classes 150
4. Inclusion of classes 151
5. Relation among the classes 152
B. Calculus of relations 157
C. Geometry of classes 164
Chapter 7 Distributions
The intention of this chapter is to analyze systematically quantitative culture traits and other quantitative characteristics into their primary indicatory dimensions, 1+1.
I. Indicants – Quantified Characteristics, I+1 169
A. Simple indicants 169
1. Degrees of precision in observing 169
2. Quantifying qualities 172
B. Compound indicants 174
II. Distribution, iI t P 182
A. The meaning of distributions 182
B. The moments of distributions 183
1. Probability – the first moment in an all-or none distribution 183
2. The arithmetic mean – the first moment generalized to any distribution 185
3. The standard deviation – the square root of the second moment 186
C. Types of distributions 190
1. Ordinal types 190
2. The binomial distribution 190
3. Normal probability distribution 195
a. Epsilon hypothesis 195
b. Characteristics of the normal distribution 200
c. Sampling theory 203
Chapter 8: Correlations
The intention of this chapter is to analyze systematically culture patterns and other correlated characteristics into their secondary indicatory dimensions, I.
I. Culture Patterns 209
A. "Traits" 209
B. "Unitary" 212
C. "Correlated" 213
D. "Dynamic Patterns" 213
II. Patterns of Paired Traits 213
A. All-or-none patterns – joint probability 213
B. Qualitative patterns – contingency 216
C. Ordinal patterns – ranks 218
D. Cardinal patterns – the correlation surface 220
E. Temporal and spatial patterns – correlations in geographic and time series 225
F. Epsilon patterns – common elements 226
III . Patterns of Multiple Traits 231
A. Intercorrelation 231
B. Partial correlation 232
C. Component analysis 234
D. Reliability correlation 234
E. Validity correlation 235
F. Multiple correlation 237
IV. Prediction from Correlation Patterns 238
A. Regression 238
B. Probability whorl 241
V. Geometry of Correlation Patterns 243
A. Correlation as a cosine 243
B. Application of cosine theory to culture patterns 244
VI. Summary of the Indicatory Sector 246
Section E – Change - Analyzing Temporal Dimensions 248
Chapter 9: Processes
The intention or this chapter is to analyze systematically the +1 primary temporal dimensions, T- or social situations, S.
I. Introduction to Processes 249
A. Classification of temporal dimensions 249
B. Definition of social processes 250
II. Qualitative Processes, Zero Moments 252
A. Extensing -- change in extensity, i.e., number of dimensions 252
B. Similarizing, assimilarizing dissimilarizing 255
C. Sociating, associating -- dissociating, I zil T 256
III. Quantitative Processes First Moment 258
A. Actional tensing EP-1 258
B. interactional tensing, EP-2 262
IV. Relative Processes Second Moments 268
A. Competing, or regrading, Rg 268
B. Revaluating, R1 . 272
C. Mobility, or repopulating, Rp 273
D. Migrating 276
E. Summary of the tensitive processes 278
F. Other second moment processes 278
V. Combinative Processes Various Moments 282
A. Hypotheses of combinative processes 282
B. Covaluating – disvaluating, Cv 282
C. Momentum, Mm 283
D. Other combined processes 284
Chapter 10: Forces
The intention of this chapter is to analyze systematically all kinds of social forces into their secondary temporal dimensions,T-2
I. Celeration -- Acceleration and Deceleration, (I) T2 287
A. Measurement of change Of speed 287
B. Classification of celeration by temporal sequence 291
C. Logistic trends 297
D. Cycles 305
II. Social Forces, P (I) T-2 306
A. Definition and examples 306
III. Units of Force, F 309
A. Time units 309
B. Index of change units 310
1. Attribute units – "P-forces" 310
2. Ordinal units – "IP-forces" 311
3. Cardinal units – "IP-forces" 312
C. Population units – "I-farces" 315
D. qualitative unit acts – attribute forces 316
IV. Systems of Forces Fr 317
A. Aggregation of forces 318
B. Summation of forces 318
C. Correlation (i.e., the scalar product) of forces 319
V. Summary of the Temporal Sector 321
VI. Summary of Part I, Chapter 1 - 10 323
Section F - Introduction Synthesizing Recorded Dimensions 325
Chapter 11: Situations
The intention of this chapter is to show how a sample of recorded dimensions may be systematically synthesized into a unit situation, S, by combining its indices (I), their scripts (s), and operators (;)
I. Semiotics 326
II. Pragmatics – Relating Symbols And Interpreters 328
A. Popularity 328
B. Completeness 328
C. Frequency of use 328
D. Duration 328
E. Familiarity 328
F. Importance 328
G. Reliability 328
H. Parsimony 328
I. Order or system 328
J. Sensory affect 328
III. Syntactics 331
IV. Semantics 332
V. Semiotic Laws 337
Vl. The S-System – A Dimensional Application of Semiosis to Social Science 340
A. The scriptal S-formula 340
B. The scripts, s rs s 342
C. The operators 345
D. Unit situations 346
E. Geometry of social situations 349
F. The quantic classification 361
Preview of Part II, Synthetics 353
Section G -- Controls Synthesizing - Correlated Factors 356
Chapter 12: Plural Tensions
The intention of this chapter is to synthesize systematically the dimensions of any tension in a plural
I. Introduction 357
Ii. Desiderata, V. 357
III. Desiring, D. 358
IV. Tensions, E s D /V 361
A. The equation defining tension 361
B. Negative tension 366
C. The tension hypothesis 366
V. Quantics of Tension, Es 367
A. The dynamic factor, Tt 369
B. The population factor, Pp 370
C. The desire and desideratum factors, Dd, 373
D. Correlation of factors, I 374
1. Tension and desideratum correlations, E.V 375
2. Tension and desire correlations, FePD 378
3. Desiring and desideratum correlations, PD.V 379
4. Population correlations, Po 381
V1. Matrices Of Tension, 1)14 383
A. Sets of desiderata and of desires, Vv, Dd 383
B. Sets of plurals, Pp 390
C. Sets of dates, tT 391
D. The complete tension theory es 392
Chapter 13 Group Tensions
The intention of this chapter is to synthesize systematically the dimensions of social control as the intentions and instruments of controllers influencing controllees.
I. Social Control (C1 = D.IeV is = 0;0;0;3) 395
A. Primary dimensions of control, Ii 395
B. Secondary dimensions of control, II • I. 399
C. Tertiary dimensions of control, I. I. I 403
D. Verification of the intertension theory 407
II. Intentions of Controllers D+T 410
A. Telic terms 410
B. Intentions of controllers 415
C. Intentions of controllees 418
D. Intentions of controllants 418
III. Instruments of Control, IN 418
A. Classification of instruments 418
B. Opinions of controllers 419
C. Behavior of controllers 420
D. Symbols used by controllers 425
E. Material used by controllers 425
F. Agents and agencies used by controllers 426
G. Instrumentation 'of control 426
IV. Influence on Controllees, Tip 427
V. People in Control 429
Section H – Crafts - Synthesizing Group Controls 435
Chapter 14 Organized Institutions
The intention of this chapter is to synthesize systematically the dimensions of the crafts or institutional organizations.
I. Patterns, Ii 426
II. An Institution, Cls 441
III. Institutional Organization – The "Crafts" – P3: I3 447
A. Group dimensions of a craft 447
B. Further dimensions of crafts 458
IV. Institutional Intercorrelation: Pattern Correlates – Cls • Cls 464
V. Institutional History: Time Correlates – Tt Cls. 474
A. Lead and lag, lunge and lapse 475
B. Institutional cycles 475
C. Institutional trends 479
VI. Institutional Geography: Space Correlates – L1 • C1 s 482
A. Universality of the institutions 482
B. Culture areas 482
C. Buildings and rooms for institutional patterns 484
VII. Institutional Social Sciences 485
A. Relations of the crafts, the social sciences and sociology 485
VII. Summary Of The Chapter 490
Chapter 15 Disorganized Institutions
The intention of this chapter is to synthesize systematically the dimensions of institutional disorganization and reorganization.
I. Definitions 492
II. Indicatory Dimensions Of Disorganization, Io 494
III. Temporal Dimensions Of Disorganization, T-2 : Io 504
IV. Spatial Dimensions Of Disorganization L2 : Io_ 509
I. Populational Dimensions Of Disorganization, Pp : Io_ 509
V. Indicatory Dimensions Of Reorganization, '+Ioi 514
VI. Temporal Dimensions Of Reorganization, Tt : Io 520
VII. Population Dimensions Of Reorganization—The Welfare Craft, P : ' 10 523
Section I -- Communities - Synthesizing Regional Crafts 528
Chapter 16 - Communities
The intention of this chapter is to synthesize systematically the dimensions of communities as all the crafts of a region.
I. Dimensions Of A Community (I) Cy 529
Ii. Populational Patterns, P • cy 533
A. Harmonic distribution of communities by size 533
B. Explanatory hypotheses 542
C. Epsilons unifying harmonic, normal and logistic distributions 544
D. Interaction of communities---the "PP over D" hypothesis 547
E. Evidence on the "PP over D" hypothesis 551
F. Interpretation of the "PP over D" hypothesis 553
III. Growth Patterns, TCy 554
IV. Density Patterns, Dn • 561
A. Correlation of density with 561
B. Correlation of density with 564
C. Correlation of density with 566
D. Correlation of density with 567
V. Culture Patterns, I •cy 569
A. Qualified culture patterns 571
B. Quantified culture patterns 574
C. Principles about culture 578
Section J – Societies Synthesizing Coeval Communities 580
Chapter 17 Past Societies
The intention of this chapter is to synthesize systematically the dimensions of the past of society.
I. Definition of "Society" 581
II. History Of Society 583
III. Age of Physical Characteristics, 1005 – "Matter 591
IV. Age of Biological Characteristics, 1004 – “Lift' 592
V. Age of Human Characteristics, 1003 – "Man Responds" 594
VI. Age Of Cultural Characteristic – "Culture Accumulates" 596
VII. Age Of Personal Characteristics – Personalities" 597
VIII. Age Of Pregnant Characteristics – "Ideals Predict" 599
Chapter 18 Planned Societies
The intention of this chapter is to synthesize systematically the dimensions of future society.
I. Social Planning 603
A. Planning as a determiner of future society 603
B. Attitudes favoring planning 604
C. Achievements in planning 606
1. Institutional plans 606
2. Communal plans 607
D. Dimensions of planning 610
II. Goals---The Ends vv V 613
A. Survey of past ideals, -T; +t(V)v 613
B. Measurement of present ideals, °T: +t(V)v 615
1. Behavior scales • 615
2. Attitude scales 616
C. Creation of future ideals, +T: +t(V)v 618
III. Implementing – The Means, o,+t(v) 619
A. Recording – the facts, (I) 619
B. Districting – the regions, Lt 620
C. Programming – the agenda, (IT-1) 621
D. Scheduling – the time table, +t T 621
E. Staffing – the personnel, PP 624
F. Financing – the budget, Mn 627
G. Popularizing – the tensions, Ee 627
IV. Fulfillment – THE Results, I 629
Section K Social Science Studying Social Situation 635
Chapter 19 Sociology
The intention of this chapter is to systematize the dimensions of scientific method in social science.
I. Sociology as Scientific Methods Applied to Society 636
A. Development of the social sciences 636
B. Development of sociology 636
C. Content of sociology 637
D. Definitions of sociology 639
II. Scientific Methods 642
A. General procedures – definable by operators--; 642
B. Special procedures – definable by scripts 647
III. Sociology and the Social Sciences 658
A. A formula defining general sociology 658
B. The social sciences 659
C. Social science ad general sociology 660
Chapter 20 Science
The intention of this chapter is to systematize the dimensions of all the sciences.
I. Sociology within Science 662
II. Deductive Sciences Systematized 664
A. The generalized S-formula 664
B. Dimensions of philosophy 664
C. Dimensions of logistics 666
D. Dimensions of mathematics 668
III. Inductive Sciences Systematized 671
A. Dimensions of music 671
B. Dimensions of physics 671
C. Dimensions of chemistry 672
D. Dimensions of biology 676
E. Dimensions of psychology 677
F. Dimensions of social science 678
G. Dimensions of. linguistics 679
IV. Science Systematized 685
A Social Distance Test in the Near East
Stuart C Dodd
American University of Beirut
Beirut, Lebannon
Abstract
Five statements of attitude ranging from friendly to hostile were so chosen from a set of thirty nine statements by the ratings of sixty judges, as to secure:
a) equidistance between statements,
b) minimal ambiguity and
c) maximal reliability
Four tests were constructed applying these five statements to fifteen national groups, eleven religious groups, five economic levels and three education levels in the Near East. On correlating the tests as given to 120 Freshmen with the results from a repetition after a month, the distance between groups showed reliability correlations varying from .50 to .96. Tables for the 124 distances between all pairs of groups and between in-group and out-groups were computed. This technic yields possibilities of quantitative definitions of various sociological concepts. Economic groups desire to ascent but they prefer to remain in familiar in-groups rather than become too intimate with a very different out group. An experimental attempt was made to modify religious distance through a college course. The greatest gains in friendliness were toward the Bahais and the Jews.
In selecting five statements for a social distance test the customary subjective judgment of the constructor of the test was replaced by the three criteria of:
a) equal attitudinal distances1,
b) minimal ambiguity and
c) maximal agreement between panels of judges.
Thirty nine statements, ranging from friendly to hostile, were rated on an eleven-point scale by sixty judges grouped in two panels. The resulting mean rated position of each statement was graphed, enabling the selecting of statements at equal intervals. The standard deviation (SD) of the sixty ratings of a statement measured its ambiguity. The difference2 between the mean position assigned to a statement by the panel of students from all the Near East and by the panel of business men of Beirut is a measure of the degree to which the judges rankings might depend on occupation, nationality, religion, age, and other factors which differed between the two panels of judges. A correlation of .99 was found between the thirty-nine positions assigned by the two panels. The five statements with the smallest difference between panels were then determined from candidates remaining after applying criteria (a) and (b) The statements were assigned scores from 0 to 100to serve as units of social distance expressed in percentages of the maximal distance, 50 being the neutrality point between friendliness and hostility.
Table I: Data on Statements Accepted
|Statement |Mean Rated |Standard Deviation |Score |
| |Position (Eleven | |Assigned |
| |point scale) | | |
|If I wanted to marry, I would marry one of them |..25 |.44 |0 |
|I would be welling to have as a guest for a meal. |2.6 |1.46 |25 |
|I prefer to have merely as an acquaintance e to whom one talks on meeting |44.9 |I.26 |50 |
|in the street | | | |
|I do not enjoy the companionship of these people |77.4 |1.00 |75 |
|E. I wish some one would kill all these individuals |99.6 |.47 |100 |
|F. I know nothing about Ibis group; I cam,* a- press an attitude |* |* |* |
*Omitted from all calculation.
The test comprised four subtests adapted to the educational, economic, national, and religious groupings in the Near East. A contribution of this study was intended to be the measurement of social distances between such a diversity of groups by means of a single scale and type of unit.
The educational grouping included three groups3: illiterates, college graduates, and people with some intermediate amount of schooling.
The economic grouping comprised five groups, defined as:
(1) A person earning less than 25 Syrian piasters daily.
(2) An unskilled worker earning between as piasters and one Syrian pound.
(3) A skilled worker earning between 1 and 4 Syrian pounds daily.
(4) A well-to-do person earning between 4 and 6 Syrian pounds daily or between approximately 1,500 and 6,000 Syrian pounds annually.
(5) A very wealthy person receiving between 16 and 64 pounds daily or between 6,000 and 24,000 Syrian pounds
The income in each level is approximately four times the income of the next lower level.
The fifteen national groups chosen were American, British, Chinese, Egyptian, French, Greek, Iraqi, Italian, Kurdish, Negro, Palestinian, Syrian, and Turkish.
These represent all the important national or racial elements in the Arabic Near East. They insure securing distances between western and imperialistic groups and eastern or dominated groups, between majority and minority stoup in each country, Arabic-speaking peoples arbitrarily divided by national boundaries, between color groups, between highly friendly groups, and all of whom excepting perhaps the Negro have been within past twenty yens expatriating massacring, or at least at war with, some others in the list.
The eleven groups in this religious grouping were Atheist, Armenian Gregorian, Bahai, Druze, Greek Orthodox, Jew, Roman Catholic or Latin, Sunnite Moslem, Shiite Moslem, and Syriac. This list represents the chief religious groups in the region.
An attempt was made to reduce the "halo” error (as in one’s attitude toward nationalistic Zionism coloring his attitude the Jewish religion) by instructions to isolate mentally the characteristics of each of the four groupings in turn and judge it irrespective of the other three characteristics. The extent to which such instructions may have reduced the "halo” is a problem for future research.
The test was taken twice without Apr signatures, with an interval of one month, by 170 Freshmen at the American University of Beirut. The student body in the university has representatives of all the different groups named In the four subtests (except Chinese). In treating the results, the means of the social distances between every group that had ten or more responders and every other group of the same grouping were calculated. There resulted 174 mean social distances within all possible pairs of groups.
The correlations between means scores of the first and second trials are given in Table II.
Table II:
|Groupings a |Number of Intergroup |Correlation |
| |Distances | |
|Religious |66 |.91 |
|National |90 |.70 |
|Economic |15 |.96 |
|Educational |3 |* |
| | | |
| | | |
|Altogether | | |
| | 174 |.81 |
* Too few to correlate.
In the following discussion persons who express their attitudes by taking the test will be called the responders, while the group toward whom they express their attitudes will be called the respondees. If the symbol xdy means the distance of group x from group y then the social distance margin is defined xdy - ydx = xmy. This may be read as the margin of x over y. The margin of y over x, ymx, is the same in amount though opposite in sign ymx - xmy. The distance of a person to an in-group xdx is the distance in the situation where the responder and the respondee belong to the same group. A distance to an out-group, xd0, is one where the respondee is a member of a different group than the responder. The average distance from all out-groups towards a particular group, x, is at 0dx.
In this paper the terms friendly and hostile, or liking and disliking, will be used to denote social distances less than and greater than 50, the midpoint of the scale. Except where specifically noted, a distance throughout the remainder of this study is the average of the two distances determined on the first and second trials as described above.
The table of 66 interreligious distances reveals that the average distances of each group towards out-groups are constant, but the reverse distances are variable. If interreligious distance may be one index of tolerance, then all of these religious groups seem to be equally tolerant, or friendly, in general. Every xd0 = 45 (where 50 is the neutrality point). The reverse distances, 0dx, of all out-group to a particular group vary from 27 to 57. There are decided differences in the degree to which a given group is liked by the others.
The Jews are the most disliked (0dJ = 57) with Atheists (0dA = 57) and Armenian Gregorian (0dG = 57) next, while the Protestants are most liked (0dG =27).
The social distance margins revealed here suggest the possibility of defining in quantitative terms such a classification of groups as that given by Professor Miller4 into horizontal and vertical groups. The definition of these groups has been largely qualitative in that horizontal groups were those to which, in every language, people refer as "lower" and "higher." such as casts and army ranks, while vertical groups were any that cut across these horizontal groups. Wherever it is possible in a social distance test such as this to arrange all the groups of one grouping in a rank order along one line (such that the social distance margins of any group to all those on the other side of it are negative) the grouping is a horizontal one To the extent that the social distance margins are not uniformly one way ones but tend to be zero in an algebraic average, the grouping is a more vertical one. According to this principle the religious groups in this situation it a vertical one, as there is no agreement on the rank of the different religious groups.
A rough test of the validity of the attitude scale exists in comparing its findings with known historical situations. Of all the national groups in the Near East, the greatest hostility might be expected to be Armenians against Turks, and of Arabs against Jews. The Armenians here have been expatriated and practically all have lost near relatives through massacres and deportations under the Turkish regime. The Arabs in Palestine believe themselves faced with an Increasing Jewish immigration which is backed by ample capital. A landless unemployed class has been created. A general psychology has been created of fear for their future, and hatred of the Jews who give rise to that fear. This has reached the point of riots and killings several times within the last five years. The test shows that the two outstandingly large distances or hostilities are of the Armenians for the Turks (AdT = 82) and of the Palestinian Arabs for the Jews (PAdJ = 75).
Again, it shows that the distance of the majority groups toward the minority groups in the Arabic-speaking Near East is large. Thus the Arabs dislike the minority Jews in Palestine the Syrians dislike the minority Armenians in Syria (SdA = 62), and the Iraqi Arabs dislike the Kurdish minority in Iraq (IdK = 55). In all cases they feel that there is an invasion of an unassimilated, foreign element speaking a different language and competing with them for political or economic power in their country.
At the other end of the scale the outstanding friendships which a historian would expect are those of Armenians for Americans, largely as a result of the aid given by Americans through the Near East Relief and other channels (ArdAm = 20). Another outstanding friendship at present is that of Iraqis for the British as the result of the peaceful relinquishing of British control in Iraq and the granting of independence to it (IdB = 12). In both the religious and national groupings, the high degree of all out-group friendships toward Protestants (OdP are 21) and Americans (OdA = 27) is easily explained as due to selection. The students in this university are those who come for an education under American auspices to an institution which, while nonsectarian, grew out of a Protestant mission college. In the popular mind it is still thought to have more of a Protestant atmosphere than the two other universities in Syria which are conducted, the one by Jesuits and the other by Moslems.5
In the vertical grouping of nationalities, just as in the case of religions, it may be observed that the average distances of any group to all its out-groups is almost constant (XdO = 40 to 42), while the reverse distances are variable (OdX = 27 to 60). Again, no group here is outstandingly friendly or hostile in general, but several groups are generally befriended or generally disliked. Thus the average distance toward Americans is 27 while toward Jews, Kurds, and Armenians it is 55 or greater. (These findings, of course, are dependent upon the composition of the total sample studied.)
The general findings from this table of nationalities is that social distances are not determined by geographic proximity nor by abundance of contacts, as much as by definite acts of a benevolent or malevolent sort between groups. Processes of cooperation or conflict seem to have determined these distances more than the mere amount of interaction. Thus distances are greater toward definite enemies than toward the completely unknown Chinaman of a different color with whom none of these students have any contacts.
Again, groups that live side by side and have many commercial and other relations together are not nearly as close as groups who have fewer contacts but are felt to be benefactors such as the British in Iraq, and the Americans among the Armenians and the university community to its students.
A further suggestion from the data is that it may become possible, on securing wider sampling, to measure in quantitative terms such frequently used but ill defined concepts as social "snobbishness' of classes, a national 'superiority complex," the degree of "ostracism" of a group, and the extent of "superordination" and "subordination" in a relationship of accommodation.
The outstanding generalization from the table of educational distances is that educational groups are horizontal ones and that all educational groups desire to rise. That is they feel more friendly toward those of greater education than toward those of Iess education than themselves; xdy > ydx where x > y in education.6
With regard to economic groups, two interesting tendencies are revealed. The first is the tendency to rise, i.e., that economic groups would prefer to go up to a higher income class rather than to sink to a lower. Economic groups are definitely a horizontal grouping. The second tendency is which works in opposition to the first. It is the desire of these groups to prefer to stay in their familiar in-groups rather than to become too intimate with a very different out-group. While these groups would like to rise in the economic scale they do not want to rise to such an extent that they would feel ill at ease, or like a fish out of water. For example, it may be the understandable feeling of a poor person that he would not be happy having to wear a dress suit and acquire the culture patterns of the wealthier classes for which he was not educated. The factual finding is that while these groups prefer to go up one level rather than down one level, and prefer to go up two levels rather than down two levels, yet they seem to prefer to go down one level rather than go up two levels.7
These data are again borne out by the second sample of students which has been analyzed. The differences are statistically significant. The numerical data might be graphically represented in the form of vectors. One vector, which may be something like "ambition," is the desire of economic groups to move up to higher levels, but is increasingly opposed by an oppositely directed vector representing a desire for familiarity, which increasingly resists moving further and further away from one's owe group and its familiar folkways.
Another hypothesis induced from the data is that it is possible to derive an index of mobility and of social stratification— from tests such as the one here discussed. It might be expected that in societies where stratification is great, as among castes in India or certain classes in European countries who expect to "keep in their place," ambition may be small, and the desire for familiarity may be dominant. But in more democratic, newer, and more rapidly growing countries, such as the United States, ambition may be more prevalent, familiarity feelings may be less strong, and social mobility as measured by this scale would be larger. Syria probably represents an intermediate example. At least the tendencies, as here measured, could be compared with similar measurements by this scale in other countries which are generally agreed to be more highly stratified economically, and in countries agreed to be less highly stratified economically. Eventually from larger collections of such data another index of stratification or mobility might be worked out and a set of international norms derived.
The average in-group distances and the average out-group distances of every group toward all other groups were computed. They were found to be in the ratio of four to one (4 = ), where the bar above the d indicates it to be an average d for the N groups. In this sample as here measured, the solidarity or liking of the members of a group for their own group ( = 11) seems to be about four times as great as the degree of liking of members of one group for members of all other groups ( = 44). A comparison of the in-group and out-group distances for the four groupings studied shows that while all these differences are not statistically significant, it is noteworthy that the greatest difference toward out-groups occurs in the religious grouping. In this region communities are largely organized by religions. For the most part, religious groups congregate in different quarters of towns, or in different villages in the country. Each religious community has its own schools, legal code, and courts for matters of marriage, inheritance, etc. Election to Parliament is proportional to the size of the religious communities. The religious cleavage is perhaps one of the deepest here. The technic at least suggests methods of measuring the deeper cleavages between groups in a social structure, as well as the relative degree of loyalty or solidarity within groups.8
One more hypothesis may be induced from the present data as an interesting generalization to explore further. It should be noted that on the average all intergroup distances are on the friendly side of the midpoint (50) of the scale. The average xdo = 40.
This suggests that in general these students are more disposed to like than to dislike people of diverse groups in their society. The processes of co-operation and assimilation, as far as they are measured by this scale, would seem to have been stronger than the processes of competition and conflict in this group of college students. It should be remembered that the out-groups represent relationships which may be expected to have a high degree of hostility among them as judged by conflicts involving bloodshed in the lifetime of the present generation.
A further study was made to explore the extent to which religious distances might be modified through education. After the first two administrations of the test to the Freshmen, there followed a month of study of the religions of the region, and then the test was administered for a third time. The study of the religions was done in a civics course in which it was naturally taken up as one section of the textbook, and as part of the course which was studying during the year the different social institutions. The study was therefore brought in naturally, and the students had little reason to feel that an artificial attempt was being made to modify their attitudes. At no time did they know that the test was going to be given to them again. The month of study comprised (a) reading the biographies of Moses, Christ, and Mohammed; (b) studying representative selections from the sacred writings of the Old Testament, the Gospels, and the Koran; (c) lectures and readings on the historical development of the institutions of the three religions of the region, the development of their sects and offshoots, and their present forms of worship and doctrines; (d) writing papers on each of the three religions; (e) visits to a mass, to a service in a mosque, and to one in a synagogue by some of the class; (f) and everyone participating in an oral speech contest on the topic, "One interesting thing I have learned about a religion other than my own. "
The taking of the test was voluntary, and no students signed their names to their papers. For this reason it was impossible to identify individuals and correlate distances of individuals before and after the month of study. The groups, however, were identified and intergroup religious distances were correlated. These distances from the third administration of the test correlated with the average distances of the first and second administrations at .84 (c.f. r = .91) between first and second administrations). There was a slight tendency to decrease distances to out-groups, i.e., to feel more friendly or tolerant toward other religions and sects. This is a mere suggestion, however, as it was not statistically significant. At the same time there was a decided decrease of distances to in-groups, i.e., 11 to 3. There was an apparently enhanced loyalty or appreciation of their own religious group. This was to be expected, as most of the students knew very little about all the material that was studied in regard even to their own religions.
In particular groups, the largest gain in friendliness was toward the Bahais and the Jews It is believed that the idealistic ten-point creed of the Bahais, which had been little known, made a strong intellectual appeal to the students. It seems also likely that there had been a "halo effect" by which the general antagonism to Jews on nationalistic grounds (which Arabs throughout the Near East share to some extent with the Palestinian Arabs) had spread over into a dislike for Jews on religious grounds. After studying the Jewish religion, however, Christians and Moslems alike may have come to appreciate that it was, in large part, the origin of their own religions and that there was a large fund of beliefs in common. This might have decreased the "halo effect" to some extent, and made the class more friendly to the religion of the Jews whatever the attitude toward their nationalistic significance might have continued to be.
In sum, the inductions from the present data have yielded hypotheses which may have wider significance in sociology. At present these hypotheses are strictly limited to the data of the group here reported. But these hypotheses may be worth checking in other groups by means of a comparable technic such as this scale provides.
Notes
1. See Thurston and Chave, The Measurement of Attitudes (University of Chicago Press, 1929
2. Expressed in units of the standard deviation of the difference.
3. Five levels would have provided finer discrimination.
4. H.A. Miller, Races, Nations and Classes (J.H. Lippencott Co., 1914), pp 14 ff
5. More precise data for validation of the scale are being sought Thus if a representative sample of intersectarian marriages can be secured, it would become possible to correlate the frequencies of such marriages with the distances between those sects, and see the extent to which a behavioristic index corroborates the findings from a verbal attitude test.
6. This is borne out from the findings of two other samples, on of 200 college students above Freshman year, and the other of 200 business people in Beirut, the data for which are not here presented.
7. To express this finding in a formula let the subscript p denote a difference in level. Thus xd(x + p) denotes the distance of group x from a group p levels higher, and xd(x - p) where p = 2, for example, denotes the distance of group x from a group two levels lower. It is convenient to refer to a level as a “similar” one or a “dissimilar” one according as p is arithmetically small of large (1 or 2 in this study). “Higher” or “lower” levels refer to p being algebraically positive or negative. “Near” and “far” refer to the size of d, the social distance, as usual.
The average distances are:
xdo = 17 xd(x - 1) = 30 xd(x - 2) = 50
xd(x +1) = 25 xd(x + 2) = 33
Thus in general, xd(x - p) > xd(x + p), but xd(x - 1) < xd(x + 2), (The latter may be read as the distance of group x form a group one level lower is less than the distance of group x from a group two levels higher.
8. A caution should be noted that while the in-group distances are dependent only on the units of this scale, the average out-group distances are also dependent upon the selection of groups in a particular study. To minimize this, all the important groups in a given region should be included. This explains why the list of groups in the groupings of the present study was made so large.
Standards for
Surveying Agencies
By Stuart C. Dodd
Dr. Dodd, Professor of Sociology and Director of the Social Science Research -Section of the American University of Beirut, outlines a set of standards by which the quality of any demoscope (poll. or survey of people) may be evaluated. The list represents a forward step in the professionalization of polling, and should provide guidance for any association of agencies seeking to set up meaningful membership requirements. The present paper has particular significance in the light of the author's previous discussion of the proposed World Association of Surveying Agencies (Winter QUARTERLY).
As surveying by sampling a population grows in use, the need for standards to distinguish between good and bad workmanship also increases. Objective indices are needed. Each should measure the degree of excellence of a survey in some respect, or along some dimension. In terms of these dimensions (which define the kinds of standards) and points on each dimension, one can specify amounts of each standard, such as a minimum point, current actual point for a particular survey, an average point, or a maximum point. Then it becomes possible, by means of these standardized dimensions, for surveying agencies to specify the quality of their service to a client or to the public, or for any professional association to set up membership requirements.
In proportion as standards may become more exacting in the future, demoscopes will become more exact instruments for observing people by sampling and the social sciences will become more exact sciences. Exactness in observing facts and inducing generalizations from them increases the power of any science to predict and control its phenomena and fulfill the function of science. Toward these ends, then, the following standards for demo-scopes are proposed.
In the column below, headed "Kinds of Standards," some forty dimensions of excellence in surveying are operationally defined by specifying some index. Wherever possible the index is an ordinal or cardinal index, but for the present it will often have to remain simply an all-or-none index asserting either the presence or absence of some attribute. This column tells how to measure each kind of standard.
The second column below, headed "Amount of Standard: Minimum," suggests an amount of each standard such that about 90 per cent of current surveying agencies would fulfill it and about 10 per cent would not be up to that standard. This minimum, stated in these relative or percentile terms for the most part, is suggested as a requirement for admission to any national or international association of surveying agencies. These proposed minimum points are collected together at the end of this paper to make a suggested set of Membership requirements. The phrase "current practice" is used to allow latitude at first while professional standards are forming. It should be interpreted to include the practices of some percentage, such as 90 per cent, of the better agencies, during the preceding year.
The third column below, headed "Amount of Standard: Maximum," specifies a higher point on each dimension and a point that is about the best yet achieved by any scientific survey with ample resources. This point on each dimension may be expected to rise in the future (as also the minimum points) due to rivalry among agencies for most accurate prediction and due to research on surveying techniques.
Whenever any Membership Committee or International Inspectorate may want to use these indices, they may have to be operationally defined greater detail. This will include constructing scales to measure those standards which are inadequately specified here. For summarizing these indices into a single net index of surveying excellence, it will be useful to devise weights for different purposes and also a standard "general purpose" set of weights. Eventually Manual of Standards currently revised to keep pace with research and growth in demoscopy will become desirable just as for any other major scientific instrument of increasing precision. Differential weighting including zero weights can adjust these standards to any purpose in hand including varying degrees of accuracy desired in a survey. Thus surveys with no respondents, as when the "interviewer" observes people's behavior directly, will omit those standards which are relevant only to the respondents.
Such periodic revision will help avoid ''freezing" standards prematurely — a fear sometimes expressed. But the chief insurance against freezing out new and better techniques — which may be opposed by vested interests in the opinion industry — is research. Scientific experiments should decide future controversies as to what are superior techniques and not majority of votes of surveyors based only on impressions from general experience.
It may be noted that these standards do not specify whether an agency is governmental or not, or operated for profit or not. A scientific instrument should be specified independently of any economic or political system as far is possible.
Standards for Surveying Agencies
|Kinds of Standards |Amount of Standards |
| |Minimum |Maximum |
| | | |
|. Agency Credence Standards | | |
|1. Responsibility | | |
|There should be on public record: |All four on public |All four should be named, with addresses, and they |
|The agency executing the survey |record |should be more accessibly published |
|The. agency (if different) controlling its policies and purposes | | |
|The body reported to | | |
|The sources of funds | | |
| | | |
|2. Integrity | | |
|Honest fact finding and the reputation for it should be the chief purpose | | |
|of every survey agency. Indications of integrity include: |Current practice |Above 90 percentile on all these indications of |
|Willingness to be inspected | |integrity |
|Frequent validity studies | | |
|Interpartisan staff and backers in preference to mono-partisan ones, and | | |
|nonpartisan ones wherever possible in preference to interpartisan ones | | |
|Sponsoring by public leaders or social scientists of high integrity | | |
|Absence of plausible charges of falsification | | |
|Pledges of honest fact finding by interviewers and by the agency | | |
|Public recording of all standards | | |
|Absence of government control of a non-government agency | | |
|Listing in Directory when ready | | |
| | | |
|3. Impartiality | | |
|Surveying agencies should not be public partisans or pressure groups on | | |
|issues surveyed | | |
| | | |
| | | |
| | | |
| | | |
| |Partisan | |
| |agencies in- |Possible partisan connections more rigorously |
| |eligible |eliminated |
|4. Certification |(e.g., either labor | |
|An international inspectorate should be available to investigate standards,|unions or man- | |
|integrity and performance of any survey or agency, when invited |factures’ agencies) | |
| | | |
|5. Validity Studies | | |
|Agencies should make studies of their survey results, correlating them | | |
|against criteria (such as elections, censuses, government statistics, |Inspected once for | |
|larger samples, friends' samples, and other independent measures of |Association membership| |
|behavior or conditions which check on assertions, etc.) whenever possible, | | |
|and publishing the correlations. Accurate predicting is the final test of | |Inspected at least annually, and getting rated |
|validity |Occasional |above 90 percentile |
| |correlations published| |
|6. Comparability | | |
|Agencies should strive for comparability between different surveys, in | | |
|different time periods, in different countries or languages by progressive | |Every possible correlation calculated and |
|research on, and standardization of, terms and techniques | |published. Correlation indices averaging above .9 |
| | |are satisfactory |
| | | |
| | | |
| | | |
|7. Duration | | |
|Age of the agency |Routine statement of | |
|Probability of continuance for at least a year ahead as rated by competent |relevant differences | |
|judges |in standards and | |
| |conditions if | |
| |comparisons are made |Keeping all standards the same except the one |
| | |respect in which a comparison is |
| | | |
| |a) 1 year old (to get | |
| |validity data) | |
| |b) probability above | |
| |50 per cent | |
| | | |
| | | |
| | | |
| | | |
|II. QUESTIONNAIRE STANDARDS | |10 years old and probability above 95 per cent |
|8. Scope | | |
|Any opinions | | |
|Information | | |
|Behavior | | |
|Conditions or records observed by sampling of people ("Demoscopy") | | |
| | | |
|9. Pretesting |Current practice | |
|Per cent of year's questions pretested on: |(mostly opinions) | |
|Experts | | |
|A sample of the eventual sample | | |
| |Current practice | |
| | | |
| | | |
| | | |
|10. Exactness | | |
|Whenever standardization of response is desired: | | |
|Standardize the wording of every question | |All five classes included |
|Begin exploring an issue with open-end questions | | |
|In proportion as responses are known, prefer closed-end questions |Current practice | |
|Prefer ordinal form of response (i.e., a series of degrees) to all-ornone | | |
|form where practicable -. | | |
|Prefer cardinal form of response (i.e., equal calibration units) to ordinal| | |
|form, where practicable | |100 per cent pretested on at least 10 experts and |
| | |on square root of eventual sample. Split ballots |
|11. Clarity | |also used to test effect of variant wording |
|The meaning of questions should be unambiguous, definite, and constant for | | |
|all as tested by the percent of respondents who when they restate it in | | |
|their own words, agree essentially on its interpretation | | |
| | |Using most exact forms currently possible |
| | | |
| | | |
| | | |
| | | |
| |Current practice | |
|12. Fairness | | |
|Questions should be so phrased as to analyze the issue fairly to all | | |
|factions as testable by per cent of agreement on phrasing by a competent | | |
|panel of persons holding all the diverse opinions on that issue. (This | | |
|"Fairness" dimension may merge into "Phrasing bias" below) | | |
| | | |
|13. Isolation of issue | | |
|Questions should be so devised as to isolate the issue free from factors of| | |
|prestige, taboo, or other extrinsic pros- sure. Until scales may be | | |
|developed to analyze and to measure this degree of isolation, the rating of| |Formal testing of paraphrases |
|experts after pretests may serve | | |
| |Current practice | |
|14. Intensity of opinion | | |
|Responses should have their intensity measured wherever appropriate | | |
| | | |
|15. Phrasing bias | | |
|To control phrasing bias: | | |
|Questions must be designed which scale together (Cornell technique) , | | |
|pr2ctici. | | |
|Their intensities must be determined | | |
|The zero at the bottom of the U-curve of the intensity function must be |Current practice | |
|found | | |
|The "debiassed" proportions of respondents on each side of the issue must | |Formal testing and 100 per |
|be determined relative to that zero point | |agreement by an adequate panel representing all |
| | |the relevant opinions |
| | | |
| | | |
| | | |
| |Current practice | |
| | | |
| | | |
|16. Thoroughness |Current practice |Rigorous phrasing refined by pretests till maximum |
|The number of questions and sub-questions probing the issue, or included in| |rating results |
|a survey on one topic, measures "Thoroughness" (assuming all the questions | | |
|to be valuable for the purpose in hand where "valuable" means contributing | | |
|to other standards, especially Nos. 5, 6, 1o, 13 and 17) | | |
| | | |
|17. Informedness | | |
|Wherever degree of information is a "biasser," i.e., correlates with | | |
|opinion on that issue, the respondent's amount of information on the issue | |Intensities routinely measured by a calibrated |
|should be determined by appropriate questions and opinions should be | |scale |
|classified by their informedness, as into "uninformed," "informed," or | | |
|"expert" categories | | |
| | |Debiassed proportions determined from well-scaled |
|III. SAMPLING STANDARDS | |questions |
|18. Universe | | |
|The universe or parent population sampled should be published especially as| | |
|to such characteristics as: territory, age, sex, economic and educational | | |
|levels,'-informedness, and other characteristics as may be relevant to a | | |
|given survey | | |
| |Current practice | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| |Current practice | |
| | | |
| | |Largest number of valuable questions exploring and |
| | |analyzing one subject of any survey recorded |
| | | |
| | | |
|19. Adequacy of sample | | |
|Size of sample (p) should be stated in absolute numbers, or as x per cent | | |
|of the square root of the population (P) sampled. Adequacy of size is | | |
|measurable by y per cent of agreement between randomly split halves. (At | | |
|the limits, a sample becomes a case, p = I, or a census, p=P) | | |
| |Universe indicated |Information measured and opinion sub-classified by |
| | |it. |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
|20. Representativeness | |Universe specified in exact detail with reasons |
|Representativeness may be analyzed into its three aspects as follows: | | |
|Population bias. Every characteristic, B, (called hereafter a “biasser”) | | |
|which correlates higher than a critical amount, r, with the question | | |
|surveyed should be determined. Common biasser are region, sex, age, | | |
|information on question-at-issue, vested interest, income, educational | | |
|level, occupation, nationality, party or sect. | | |
|Stratification. Each biasser must be stratified; i.e., proportionately | | |
|sampled. The per cent of any biasser in the universe sampled must be | | |
|matched in the sample | | |
| | | |
| |Samples to averages | |
| |over 10 per cent of | |
| |√P, (e.g., this means | |
|Randomness. Beyond the above, the sample must be randomized; i.e., selected|1200 for the U.S.A. | |
|by techniques which best eliminate further bias and assure randomness |sample). Ten per cent | |
| |of the sample is | |
|21. Method of sampling |suggested as a minimum| |
|The basis of selecting the sample should be published, whether by area or |for reporting any | |
|quota control, randomized listing, calculation, or other technique |breakdown or | |
| |sub-category. | |
|22. Panels | | |
|Wherever a constant set of informants is needed to isolate the variation of|Current practice | |
|the variable at issue through repeated surveys, a panel should be used. It |(inclusion of “check |100 per cent of √P , or p=√P (=12,000 if P = |
|may need to be checked as to its continued reprentativeness, and measured |questions” routinely |U.S.A.) y = 95 per cent (as of probability from chi|
|for biassers due to repetition (such as practice effects and annoyance from|to assure |square test of distributions from two halves of the|
|repetitions, tampering from a pressure group, etc. |representa-tiveness |sample fitting each other. |
| | | |
|23. Reliability | | |
|a) Errors of observation. Measured as differences between reobservations | | |
|at one time between: | | |
|1) Different informants, | | |
|2) Different interviewers | | |
|3) Different wording of same questions; as well as | | |
|4) Differences between reobservations at different times | | |
| | | |
|b) Errors of Sampling. Measure by the standard error of sampling | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | |Possible biasser correlated and identified by r > |
| | |.5. Biassers matched within r per cent, either by |
| | |chi square test or the proportion: |
|IV. Interviewing Standards | | |
|24. Interview conditions | | |
|Overt or covert? |Method stated | |
|Response oral, written, or by further documents? | | |
|Singly, with witnesses, or in groups? | | |
|Respondent identified or anonymous | | |
|By interview, telephone or mail? | | |
|Followed up or not; i.e., return call? |Current practice | |
|Interview invited by informant in specific appointment, or in general | | |
|required or requested by higher authority, or requested by interviewer | | |
|only? | | |
|In comfortable, unhurried situation or not? | | |
|In quiet, undistracted situation or not? | | |
|At home, work, street, or other place? | | |
|First, second or later interview (of same person on same issue)? | | |
|Number of minutes of interview? | | |
|Given prestige at first meeting? | | |
| |Current practice | |
|25. Selection | |A rigorous randomizing techniques used |
|In recruiting interviewers preference is usually given, for characteristics| | |
|practice such as: | | |
|Maturity | | |
|Education | | |
|Interest | |Most fully representative method, as determined by |
|Local languages spoken | |crucial experiments, used regardless of cost or |
|Normality in tests of: | |convenience |
|1) Intelligence | | |
|2) General opinions | |Panels rigorously used where appropriate, and |
|3) Personality | |checked. |
|Experience |Publication in | |
|Recommendations |technical journal | |
|Local residence |desirable | |
| | | |
| | | |
| | | |
| | | |
|26. Compatibility of interviewers to respondents, or social distances. | | |
|Differences between them must be uncorrelated with survey results; e.g., | | |
|differences status as to: |Current practice |Differences averaging less than 5 per cent, e.g., |
|Race | |95 per cent probability from chi square test that |
|Economic level | |the two distributions are from one universe; or r >|
|Educational level | |.97 where r applies; or |
|Sex (in Moslem areas) | | |
|Language, etc. (if biassing) | |< .025 |
|Degree of acquaintance should be administratively avoided or experimentally| | |
|measured | |where |
| | |I = 1st observing |
|27. Interviewer bias | |II = 2nd observing |
|The characteristics of the interviewer must show a correlation of zero with| | |
|the replies. | | |
| | | |
| | | |
|28. Competence of Interviewers | |Sampling errors published in full |
|Interviewers should be employees trained till they pass such efficiency | | |
|test as: | | |
|Knowledge, x percent score on objective standardized examination about | | |
|polling | | |
|Honesty. x' per cent honest interviewers as checked by supervisors on y per| | |
|cent of respondents | | |
|RAPPORT. x per cent or contacts carried through to completed interviews | | |
|RECORDING. x per cent of agreement between interviewers' records and | | |
|dictaphone records of standardized interviews | |For most purposes, 99 per cent of interviews should|
|CAREFULNESS. x per cent of errors found in editing questionnaires as per | |be: Overt, oral, single, identified, first, |
|instructions in a standardized set of situations | |invited, unhurried, undistracted, and short. |
|PRODUCTIVITY: More than x percentile in interviews per period of time |Current practice | |
| | | |
|29. Instructions to interviewers | | |
|A general Manual of Instructions in the appropriate language(s) should | | |
|guide the interviewer about selecting informants, asking questions and | | |
|recording answers, etc., supplemented by special instructions as needed for| | |
|each particular survey | | |
| | | |
|30. Supervision | | |
|The amount of supervision is measurable as the ratio of the supervisory | | |
|man-hours to the interviewer hours | | |
| | | |
|V. Reporting Standards | | |
|31. Public reporting | | |
|All surveys should be made available to the widest possible public | | |
| | | |
|32. Essentials of report | | |
|Every report of a survey should include at least: | | |
|The name of the responsible agency |Current practice | |
|The date or period and region | | |
|The universe and size and nature of the sample | | |
|Verbatim quotation of the question and the numerical findings | | |
| | |Definite standards required in such a list of |
|33. Objectivity | |characteristics |
|Surveys should be honestly and objectively reported, including verbatim | | |
|quotation of the questions and numerical findings. The "No opinion" | | |
|responses should not be lumped in with the "No" responses, etc. Subjective | | |
|interpretation or application should be distinguished from the objective | | |
|findings | | |
| | | |
| | | |
| |Current practice | |
| | | |
| | | |
|34. Limits | | |
|The relevant kinds and amounts of the standards listed in paragraphs to No.| | |
|4.0 should be stated as among the limitations of the findings in |Current practice | |
|interpreting them | | |
| | | |
|35. Recording. | | |
|All records original and secondary,- with guides to their use should be | | |
|preserved accessibly for further research or inspection | |Possible differences measured experimentally and |
| | |significant biasses avoided by appropriate |
|36. Public interest | |selection of interviewers |
|Market research and “public polling” for the public interest directly or | | |
|through reporting to a public body, such as a government or scientific | | |
|agency, should be kept administratively separate | | |
| | | |
|VI. Administrative and Other Standards | | |
|37. Preparation of respondents | | |
|Wherever unaccustomed to being surveyed the population should be prepared | | |
|by appropriate publicity, educating them to cooperate | | |
|The greater the authority backing the survey the better | | |
| | |Calculation of possible correlations and their |
|38. Tabulating | |reduction to zero by reselecting, retraining, or |
|Coding, tabulating and recording should use checks and mechanical devices | |reassigning the interviewers |
|as far as possible to reduce clerical errors | | |
| | |Above 90 percentile on all indices of competence |
|39. Speed | | |
|The speed of surveying may be specified as: "interviewing period" from | | |
|first to last, interview, or as "overall speed" or number of days from | | |
|inception to reporting |Current practice | |
| | | |
| | | |
| | | |
| | | |
|4o. Cost | | |
|The average cost of an agency's surveys including all overhead may be | | |
|reported as x dollars per interview. It will exceed the average with: | | |
|Open-ended or numerous questions |Current practice | |
|Samples difficult of access | | |
|Few interviews or surveys per year to share the overhead | | |
|Need for expert interviewers | | |
|Fullness of tabulating and reporting | | |
|Price and wage levels in other countries |Current practice | |
|Initial organizing of an agency | | |
| | | |
|41. Demoscopic development | | |
|Surveying agencies and their members should contribute to the development |Reporting | |
|of their profession through participating in their trade association and/ |a) through d) | |
|or in some of the fifteen functions below: | | |
|International surveying | | |
|Extending surveying to new regions | |Detailed and well-prepared Manuals always provided |
|International inspection | |for every interviewer in ample time |
|Research on demoscopy | | |
|A world journal | |, |
|Central Records Office(s) | | |
|International conferences on demoscopy | | |
|International. training center(s) | | |
|Maintaining specialists on call |Current practice |Supervising time |
|Encouraging regional or specialized associations of demoscopists | |----------------------- = 5 % |
|Developing professional interests | |Interviewing time |
|1) Fostering the public relations of demoscopy.. | | |
|Servicing the United Nations | | |
|Developing demoscopy in other respects | | |
| | |In addition to any popular reporting, full |
| | |technical details should be published routinely in |
| | |professional publications |
| | | |
| | |Reporting on all standards, No. x through No 40, as|
| | |realized in surveys |
| | | |
| |Current practice | |
| | | |
| | | |
| | | |
| | |Objectivity tested by zoo percent agreement as to |
| | |the statement of the findings by surveyors of |
| |Current practice |differing opinion ions on those issues |
| | | |
| | | |
| | | |
| | | |
| |Current practice | |
| | | |
| | | |
| | | |
| | | |
| | |The limitations implied by all the standards listed|
| | |in this column should be reported |
| |Current practice | |
| | | |
| | | |
| | | |
| | |Duplicate records filed an indexed in some |
| | |centralized archives |
| | | |
| | | |
| |Current practice | |
| | |Rigorous separation of civic and commercial surveys|
| | | |
| | | |
| | | |
| |Usually within a 90 | |
| |day overall period | |
| | | |
| | |Publicity having a measured and high impact |
| | |preparing for surveying; a law, a government, |
| | |civic, religious, journalistic or other influential|
| | |sponsorship |
| | | |
| |Current practice | |
| | | |
| | |Appropriate checks and machines used wherever they |
| | |exist |
| | | |
| | | |
| | | |
| | |Two days for emergencies is about the record |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| |Current practice | |
| | |x=82, norm in U.S.A. in 1946 |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | |Active member in the surveying trade organizations |
| | |and promotes the 15 funcions (eventually a 'scale |
| | |may be built to appraise such professional |
| | |participation more |
Summary of Proposed Minimum Standards
Is the following set of standards too rigorous, or not rigorous enough, to adopt at the outset?
An agency that is a candidate for membership in the world association should satisfy the Membership Committee1 that it has the following indications of integrity and competence:
• It must have operated for at least the preceding year with expectation of continuing at least another year. (No. 7)
• It must have kept- on public record its sponsorship, financial backers, and responsible officials with. addresses and other bodies to whom its reports are made or channels of publication. (No. 1)
• It must have abstained from pressure group activity, i.e., partisan activity such as to prejudice impartial fact finding. (No. 3)
• It must state its willingness to be inspected and have preserved its records of at least the preceding year so as to be available for future inspection. (No. 23, No. 35)
• It should be able to 'present endorsements by public leaders and social scientists of high integrity and variety of political views. (No. 2d)
• It should have no plausible public charges of falsification. (No. 2e)
• It should-publish a manifesto promising honest and reliable fact finding and reporting. (No. 2f)
• It should have given verbatim quotation of questions and numerical findings in reporting any survey during at least the preceding year. (No. 33)
• Its questioning should have analyzed its issues fairly to all factions on those issues. (No. 12)
• It should have published the size and nature. of the sample and its universe sufficiently to enable another agency to repeat and check the survey. (No. 2g, No. 18-21)
• It should have published some studies showing good validity for its surveys, (No. 5)
• It should undertake to continue these desirable minimum practices in the future.
• It should promise to deposit with the Association in the future the documents indicated above including a complete current file of its survey reports. (No. 3r)
Notes
1. Since the application of these standards to judge each candidate agency leaves much to the Membership Committee (until the standards can be made fully objective), the appointment of this Committee will become as important as the adoption of these standards in principle. The ways of assuring that such a Committee represents the best scientific and impartial personnel — with due regard to the public's interest as well as the profession's interest in, raising standards — and avoids freezing the status quo through vested or other interests resisting scientific progress are not discussed in this paper. They must be carefully studied, however, and embodied in the constitution of any national or world association.
L'effectif des Législatures d'après la Loi du Cube.
Une Corrélation entre Gouvernants et Gouvernés.
Dodd Stuart Carter.
In: Revue française de science politique, 3e année, no. 4, 1953. pp. 739-747.
I. La loi de la racine cubique appliquée aux Nations et aux Etats de la fédération américaine
L’examen pour 1930-1932 de importance numérique des Chambres basses des législatures par rapport aux effectifs respectifs des populations avait conduit à l'hypothèse selon laquelle cette importance numérique tend correspondre à la racine cubique de l'effectif de population. Cette hypothèse fut confirmée par le calcul de ce rapport pour 55 Nations et pour les 48 Etats des U.S.A. Pour ces deux séries de données, les effectifs de population avaient tendance à égaler le cube du nombre des membres des législatures correspondantes.
Une deuxième enquête a été menée environ deux décades plus tard sur les chiffres des années 1946-1950. Dans cet intervalle des Nations et des populations entières avaient été bouleversées par une guerre mondiale; l'Allemagne et l'Inde s'étaient l'une et l'autre scindées, et bien d'autres pays avaient subi de très profondes modifications.1 Néanmoins, les résultats de 1930-1932 ont été corroborés.
La tendance des effectifs de population à égaler le cube des effectifs des législatures, constatée empiriquement, semble une norme dans l'histoire des gouvernements des Nations et de chaque Etat de la fédération américaine. Dès lors, toutes les fois qu'il s'agira de l'organisation ou de la réorganisation d'un gouvernement, on pourra trouver dans l'expérience internationale ainsi acquise un guide pour déterminer l'effectif normal à donner aux législatures.
II.Analyse de données
A. Le rapport entre la législature et la population
On peut, au lieu de supposer au départ une relation cubique, exprimer l'hypothèse sous la forme d'une équation générale où x est un exposant inconnu:
Pl = Px Equation I
Dans cette relation, P représente l'effectif de population et Pl celui de la législature.
L’emploi des logarithmes donne:
log. Pl (natural log of Pl) = Log. P (common log of P) Equation II
d’où on tire pour valeur moyenne de x:
x = (moy. log. P1)/(moy. log. P) Equation III
Cela équivaut à désigner par x le rapport entre deux moyennes géométriques, celle des effectifs des législatures et celle des effectifs des populations. Appliquée aux diverses Nations et aux 48 Etats des Etats-Unis, la formule ci-dessus donne les résultats suivants:
|Nations |Etats Américains |
|1930 |1950 |1930 |
|2.14/6.77 = 0.315 |2.23/6.94 = 0.321 |1.97/6.26 = 0.315 |
Le rapport est toujours légèrement inférieur à celui qui conduirait à la racine cubique (x = 0.333). Mais le groupe Nations et le groupe Etats représentant des ensembles complets et non des échantillons, il ne paraît pas nécessaire de concevoir un correctif de dispersion. Les résultats peuvent être exprimés sous la forme générale suivante:
P1 = VT iv Equation IV
l'effectif de la législature approchant de la racine cubique de celui de la population.
B. Indices de corrélation
L'équation IV donne assez exactement le rapport gouvernes-gouvernants sans aucune estimation sur sondages. L'approximation des points observés par rapport aux points escomptés ou théoriques sur la courbe des racines cubiques (x = 0333) est mesurée par les coefficients de corrélation, reliant la racine cubique des populations aux effectifs des législatures. Ces corrélations sont de 0.85 en 1930 et 0.82 en 1950 pour les Nations; de 0.66 en 1930 pour 43 Etats des Etats-Unis.2 Des calculs de recoupements, relatifs à d'autres assemblées législatives ou politiques, à des conseils etc., dans divers régimes, sont nécessaires pour vérifier la valeur générale de ce « rapport gouvernés/gouvernants » comme on peut l'appeler.
C. Interprétations possibles
Les considérations qui suivent sont proposées à titre de contribution à une explication de ce rapport de racine cubique.3
1. L'effectif numérique des législatures est de l'ordre de grandeur des centaines. Ceci semble en partie résulter de considérations d'espace, de temps, et de démographie. La plupart des assemblées législatives comptent de 100 à 600 membres en partie pour une raison d'espace. Pour un débat face à face, la portée de la voix humaine – avant que l'on usât des amplificateurs électriques – semble avoir correspondu de façon la plus satisfaisante un auditoire de quelques centaines de personnes. De même, pour les débats législatifs, la nécessité en vue d'accorder un temps de parole à chacun des membres dans le cadre des séances quotidiennes et des sessions annuelles, fixait une limite normale. Si l'effectif législatif et le volume des lois s'accroît trop, la plupart des législateurs perdent leur chance d'intervenir dans les débats.
Un troisième facteur de limitation tient aux réactions psychologiques des assemblées. Un groupe de discussion fonctionne mieux quand tous ses membres se connaissent et peuvent réagir, au cours d'un débat, à la personnalité même de leurs collègues; à cet égard, toute législature remplit moins bien ses fonctions quand elle compte plus de quelques centaines de membres. Si elle devient trop nombreuse, elle cesse d'être un organisme d'individus bien intégrés et s'apparente à une foule réagissant aux moyens collectifs d'information, et non plus à un groupe de discussion.
2. L'effectif de population des Nations modernes est de l'ordre de grandeur des millions. Les quelque soixante nations souveraines ont, pour la plupart, de à 400 millions d'individus. Pour nombre de raisons, y compris la nécessité de disposer d'une population et d'une matière imposable suffisantes au maintien des fonctions militaire, diplomatique et diverses des Etats nationaux, le monde moderne s'est organisé en unités nationales qui se situent dans cet ordre de chiffres.
3. La centaine est de l'ordre de la racine cubique du million. On peut donc s'attendre à ce que l'effectif de la représentation législative corresponde environ à la racine cubique de effectif de population.
Les considérations qui précèdent peuvent tendre à expliquer la valeur moyenne du « rapport gouvernés/gouvernants », mais non le degré de corrélation. La corrélation observée peut être considérée comme due à un équilibre entre deux groupes d'influences opposées. Ces groupes d'influences sont, d'une part, celles qui contribuent à élever la corrélation, d'autre part celles qui la réduisent. Le principal facteur qui joue dans le sens de l'augmentation est la coutume d'adopter un nombre à peu près fixe d'électeurs par représentants. Le principal facteur est la variation de ce nombre au sein des Nations, pour des raisons locales, historiques fortuites ou autres.||L'importance relative de ces deux influences opposées peut être mesurée par les tests de liaison. Le coefncient de liaison est le carré de l'indice de corrélation. Les coefficients de liaison sont donc 0.72 (=0.852) pour les 55 Nations en 1930, et 0.44 (0.662) pour les Etats de la fédération américaine. || Ce qui signifie que les coefficients de non-détermination qui mesurent la tendance complémentaire de non-ajustement, sont de 0.28 pour les Nations et de 0.56 pour les Etats américains.
Cela peut être interprété en disant que les déterminants de fait de l'effectif des législatures sont dus, dans le cas des Nations étudiées ici, pour les trois quarts seulement à l'existence d'une proportion fixe électeurs par représentant et, dans le cas des Etats américains, pour moins de la moitié cette tendance à avoir une proportion unique commune.
La mesure dans laquelle la règle de la racine cubique est valable, ou les conditions pour qu'elle devienne une loi sociale, peuvent s'exprimer en termes de bon sens (qui permettent également une définition pratique): « Plus le rapport électeurs/représentant est normalisé à une valeur moyenne, plus l'effectif de la législature tend vers la racine cubique de l'effectif de population. »
D. Variations de ce rapport avec l'industrialisation
Des recherches ont été faites dans le but de savoir si le « rapport gouvernés/gouvernants » subissait des modifications avec le temps et le développement moderne des techniques de communication. Une faible corrélation de –0.2 été observée entre l'ancienneté de la constitution de l'Etat et le nombre électeurs par représentant, P/Pi.
Ce qui suggère l'existence d'une légère tendance à l'augmentation de ce nombre aux Etats-Unis pendant les 175 dernières années. L'effet de ce facteur de changement a été vérifié de façon rudimentaire en omettant les 5 « vieux » Etats dont les constitutions datent de plus d'un siècle Le groupe plus homogène des 43 autres Etats fait preuve un meilleur ajustement à la règle de la racine cubique puisque la corrélation s'élève de 0.21 pour les 48 Etats 0.66 pour les 43.
Le « rapport gouvernés/gouvernants » x, qui relie l'effectif des populations à celui des assemblées législatives, demeure inchangé à 0.315 pour les 43 « plus jeunes » Etats comme pour l'ensemble des 48. Les circonscriptions plus petites de l'époque antérieure à la voie ferrée semblent être un facteur additionnel et de perturbation suffisant pour diminuer notablement la validité de la règle de la racine cubique, mais insuffisant pour l'écarter
E. Nore -sur d'autres études
On peut signaler sur cette question des études antérieures. Le professeur L.S Penrose à développé ce principe: « Si des facteurs autres que l'importance numérique sont constants, l'élu d'un corps électoral quelconque représente une partie des électeurs pro portionnelle à la racine carrée du nombre total des électeurs »4 Il remarque que « ce résultat illustre le principe statistique bien connu que la signication de n'importe quelle observation est fonction de la racine carrée du nombre de cas sur lesquels elle repose. »
Dans notre étude, l'effectif de chaque législature tend vers la racine carrée du nombre d'électeurs par représentant. En effet, le produit de l'effectif du corps électoral moyen par représentant, Pe, le produit de Pe par le nombre d'élus. (ou la somme des corps électoraux s'ils sont inégaux) constitue à peu près le total de la population.
Pe •P1 = P
Si l'effectif de la législature tend égaler l'effectif de la population la puissance 1/3 le nombre d'électeurs par représentant aura tendance à égaler cet effectif à la puissance 2/3:
P12 = Pe3/2 = P ou Pe = P2/3 ou Pe = P12
de sorte que l'effectif de la législature tend à égaler la racine carrée du nombre d'électeurs par représentant:
P1 = Pe VII
Cette tendance se formulerait dans une équation exacte si les corps électoraux pour chaque représentant étaient tous égaux et si l'effectif de la législature était exactment la racine cubique de la population.
La question est soulevée, qu'on ne trachera pas ici d'aillerus, de savoir si la tendance pour l'effectif d'une législature à égaler la racine carrée du nombre d'électeurs d'une circonscription moyenne dérive de, ou se relie la tendance pour un élu représenter un nombre d'électeurs égal à la racine carrée du nombre total des électeurs.5
L'étude dimensionnelle de l'organisation des institutions dégage trois facteurs, ou éléments, de la population: publics, agents et clients, aux rôles entremêlés. Dans l'ordre des institutions politiques, ces éléments sont, respectivement, la population nationale, la législature, et le corps électoral de chaque représentant. || Dans le cas de 55 Nations en 1930-1932, de 57 en 1946-1950, et dans celui des Etats de la fédération américaine, la relation quantitative que l'on observe entre ces éléments est approximativement égale à la racine cubique de la population en ce qui concerne l'effectif de la législature et, par suite, au carré de celui-ci en ce qui concerne le nombre d'électeurs par élu. Cette constatation empirique constitue une norme. || Elle signifie que, bien que des déviations en deça et au delà de la norme de la racine cubique soient prévisibles pour de nombreuses raisons historiques et géographiques l'effectif de la législature, ou peut-être plus généralement de n'importe quel organe de gouvernement, peut être prévu comme suit:
|Représ lég, Pl |Population, Pl3 |
|1 |1 |
|2 |8 |
|3 |27 |
|10 |1,000 |
|50 |125,000 |
|100 |1,000,000 |
|200 |8,000,000 |
|300 |27,000,000 |
|400 |64,000,000 |
|500 |125,000,000 |
|600 |216,000,000 |
|700 |343,000,000 |
Le coefficient de corrélation des législatures nationales à cette règle de la racine cubique est de 0.85 pour 1930 et de 0.82 pour 1950. Il est de 0.66 pour les 43 Etats américains dont la Constitution date de moins de cent ans. Il est beaucoup plus bas (r = 0.22) si l'on ajoute les Etats dont la constitution fut adoptée il y a plus de cent ans, alors que des moyens inférieurs de transport et de communication, et des effectifs de population moins élevés semblent avoir eu pour conséquence des corps électoraux par représentant plus réduits.
La règle de la racine cubique s'explique par le fait que les législatures ne sont efficaces que dans certaines conditions d'espace, de temps et de dimension des groupes. Ainsi donc les législatures comptent jusqu'à des centaines de membres tandis que les Nations en comprennent des millions, les deux effectifs étant généralement liés par une relation cubique. La corrélation avec cette règle est due à l'adoption d'une proportion moyenne d'électeurs par représentant, tandis que l'écart complémentaire tient aux variations de cette proportion.
Traduit de l'anglais
Appendice
Pays utilisés dans cette étude
Effectifs de la population globale et des Chambres basses de la représentation législative
|PAYS |1930 |1950 |
| |Population (en |Représ. lég. |Population (en |Représ. lég. |
| |milliers) | |milliers) | |
|1 |Afghanistan |11,000 |106 |11.5005 |171 |
|2 |Afrique du Sud |8,129 |150 |12.112 |159 |
|3 |Allemagne | | | | |
| |occidentale | | |47,585 |402 |
| |orientale | | |18,5194 |400 |
|4 |Argentine |11,4422 |158 |16,818 |158 |
|5 |Australie |6,630 |74 |7,912 |123 |
|6 |Bolivie |2,967 |73 |3,990 |108 |
|7 |Brésil |34,683 |212 |49,340 |304 |
|8 |Bulgarie |5,884 |227 |7,160 |239 |
|9 |Canada |10,377 |245 |13,573 |262 |
|10 |Chili |4,287 |143 |5,709 |147 |
|11 |Chine | | |463,493 |702 |
|12 |Colombie |7,726 | |11,015 |132 |
|13 |Costa-Rica |534 |43 |838 |45 |
|14 |Cuba |3,962 |128 |5,199 |136 |
|15 |Danemark |3,551 |149 |4,230 |151 |
|16 |Equateur |2,527 |56 | | |
|17 |Egypte |14,218 |150 |20,045 |319 |
|18 |Espagne |23,564 |440 |28,023 |476 |
|19 |Etats-Unis |122,775 |435 |149,215 |435 |
|20 |Ethiopie |7,5002 |27 |9,0005 |606 |
|21 |Finlande |3,463 |200 |4,016 |200 |
|22 |France |41,228 |612 |41,550 |544 |
|23 |Grande-Bretagne |46,038 |431 |50,363 |625 |
|24 |Grèce |6,205 |250 |7,856 |250 |
|25 |Guatemala |2,539 |67 |2,7874 |68 |
|26 |Honduras |854 |59 |1,326 |49 |
|27 |Hongrie |8,685 |245 |9,207 |402 |
|28 |Rep. de Inde | | |346,000 |500 |
|29 |Irak |3,5002 |108 |4,800 |140 |
|30 |Irlande |2,971 |153 |2,991 |147 |
|31 |Islande |109 |33 |140 |35 |
|32 |Italie |41,177 |400 |45,996 |574 |
|33 |Japon |64,450 |466 |82,636 |466 |
|34 |Libéria |1,000 |21 |1,648 |25 |
|35 |Luxembourg |300 |54 |295 |51 |
|36 |Mexique |16,553 |170 |24,448 |147 |
|37 |Nicaragua |840 |43 |1,184 |42 |
|38 |Norvège |2,814 |110 |3,233 |150 |
|39 |Nouvelle-Zélande |1,526 |80 |1,881 |80 |
|40 |Panama |467 |46 |764 |42 |
|41 |Paraguay |864 |40 |1,304 |40 |
|42 |Pays-Bas |7,936 |100 |9,956 |100 |
|43 |Perse |10,000 |136 |17,0004 |136 |
|44 |Pérou |6,147 |110 |8,240 |156 |
|45 |Pologne |32,107 |208 |24,448 |444 |
|46 |Portugal |6,826 |164 |8,491 |120 |
|47 |Roumanie |14,282 |387 |15,873 |414 |
|48 |Saint-Domingue |1,345 |35 |2,277 |47 |
|49 |Salvador |6,142 |230 |6,956 |230 |
|50 |Suède |4,066 |187 |4,640 |194 |
|51 |Suisse |1,434 |42 |2,150 |42 |
|52 |Tchécoslovaquie |13,998 |300 |12,463 |300 |
|53 |Turquie |14,9187 |399 |19,623 |487 |
|54 |U.R.S.S |168,0002 |607 |193,0004 |657 |
|55 |Uruguay |1,941 |99 |2,353 |99 |
|56 |Venezuela |3,027 |85 |4,696 |110 |
|57 |Yougoslavie |14.438 |248 |16,040 |405 |
Sources
1. Demographic Yearbook 1948, Statistical Office of the United Nations, New York, 1949
2. Statesman’s Yearbook 1932, Macmillan Co, New York
3. Demographic Yearbook 1949-1950, Statistical Office of the United Nations, New York
4. Political Handbook of the World 1951, Macmillan Co, New York
5. Statesman’s Yearbook 1951, Macmillan Co, New York
6. American Annual 1950, American Corporation, New York
7. Ce chiffre été obtenu par interpolation entre les données de 1927 et celles de 1935
Notes
1. L’Equateur a été inclus dans la première enquête, mais nous n'avons pu disposer pour la seconde des chiffres le concernant. L'Inde et la Chine n'ont été incluses que dans la seconde enquête pour laquelle nous ne disposons pas des chiffres relatifs au nouvel Etat du Pakistan.
2. Ces trois corrélations étaient respectivement 0.65 – 0.60 – et 0.62 sur les données brutes reliant les chiffres de population absolus aux effectifs législatifs. Ce qui montre que la concordance est plus grande quand on se sert de la racine cubique de la population. Cela s'explique en partie par le fait que le diagramme de dispersion population brute – effectif législatif est curviligne, tandis que la loi de la racine cubique de la population lui donne une allure plus rectiligne en fonction des effectifs législatifs. On obtient les 43 Etats (des Etats-Unis) mentionnés au texte en faisant abstraction des 5 Etats les plus anciens de l'Union (voir ci-dessous).
3. L'auteur est redevable au professeur Paul Neurath de quelques-unes de ces suggestions.
4. L.S Penrose « The Elementary Statistics of Majority Voting » Journal of the Royal Statistical Society, vol CIX part I, 1946, pp 53-57
5. Une autre étude, qu'on pourrait rapprocher de celle-ci, a été faite par Kendall et Stuart. Les auteurs trouvent que le nombre des sièges gagnés par un des deux partis dans une élection est égal au nombre des voix obtenues par ce parti. Notre étude rattache le cube du nombre des sièges à la population totale. Nous ne voyons pas de base commune à ces exemples apparemment différents de la loi du cube. – M.G kendall et Stuart, « Cubic Proportion in Election Results », British Journal of Sociology, vol I, no 3, septembre 1950. Voir également des mêmes auteurs un exposé simpli de la loi du cube en matière électorale dans Revue Franaise de Science Politique, vol II, no 2, avril-juin 1952, pp 270-276.
The Size of Legislatures under the Cube Law
A correlation between legislatures and populations
Stuart Carter Dodd.
Review of French Political Science, Vol. 3, No. 4, 1953, pp. 739-747.
I. The law of cube root applied to Nations and to the States of the American confederation
An examination for 1930-1932 of the numerical relationship of the Lower Houses of the legislatures compared to respective size of the populations had led to an assumption according to which this numerical relationship tends to correspond the cubic root of manpower of population. This assumption was confirmed by the calculation of this ratio for 55 Nations and for the 48 States of the U.S.A. For these two sets of data, size of population tended to equal the cube of the number of the members of the corresponding legislatures.
A second survey was conducted approximately two decades later on the figures for the years 1946-1950. In this interval Nations and entire populations had been upset by a world war; Germany and India were divided, and many other countries had undergone deep changes.1 Nevertheless the 1930-1932 results were corroborated.
The tendency for population to equal the cube of the size of the legislatures, found empirically, seems a standard in history of the Governments of the Nations and each State of the American Federation. Therefore, whenever there is the organization or reorganization of a Government, there can be found in international experience a guide to determine the normal size for the legislatures.
II. Analysis of data
A. The relationship between the legislature and the population
It can, instead of assuming a cube relationship at the outset, be expressed in a hypothesis as a general equation where x is an unknown exponent:
P1= Px Equation 1
In this relationship, P, represents size of the population and Pl, the size of the legislature.
Use of logarithms gives:
log Pl = Log P Equation 2
when derived for the average value x:
x = (avg. log Pl)/ (avg. log P) Equation 3
This is equivalent to denoting by x the relationship between two geometric means, the size of the legislatures, and the size of the populations. Applied to the various Nations and to the 48 States of the United States, the above formula gives the following results:
|Nations |American States |
|1930 |1950 |1930 |
| |2.14 |
|1 |1 |
|2 |8 |
|3 |27 |
|10 |1,000 |
|50 |125,000 |
|100 |1,000,000 |
|200 |8,000,000 |
|300 |27,000,000 |
|400 |64,000,000 |
|500 |125,000,000 |
|600 |216,000,000 |
|700 |343,000,000 |
The correlation coefficient of national legislatures is the rule of the cubic root of 0.85 to 1930 and 0.82 for 1950. It is 0.66 for the 43 U.S. States whose Constitution are less than one hundred years old. It is much lower (r= 0.22) if one adds the States whose constitution was adopted more than hundred years ago then lower transportation and communication capabilities and lower population numbers seem to have had the consequence of the electoral body per representative being reduced.
The rule of the cubic root explained by the fact that legislatures are effective under certain conditions of space, time and demography. Thus, the legislatures have hundreds of members while the Nations include millions, the two sizes are generally linked by a cube relationship. The correlation with this rule is due to adoption of an average proportion of voters per representative while deviations lie in the variations of this proportion.
Appendix
Countries used in this study
Size of population and lower legislative chamber representation
|COUNTRY |1930 |1950 |
| |Population in thousand|Representative in |Population in thousand |Representative in |
| | |legislature | |legislature |
|1 |Afghanistan |11,000 |106 |11.5005 |171 |
|2 |South Africa |8,129 |150 |12.112 |159 |
|3 |Germany | | | | |
| |East Germany | | |47,585 |402 |
| |West Germany | | |18,5194 |400 |
|4 |Argentina |11,4422 |158 |16,818 |158 |
|5 |Australia |6,630 |74 |7,912 |123 |
|6 |Bolivia |2,967 |73 |3,990 |108 |
|7 |Brazil |34,683 |212 |49,340 |304 |
|8 |Bulgaria |5,884 |227 |7,160 |239 |
|9 |Canada |10,377 |245 |13,573 |262 |
|10 |Chile |4,287 |143 |5,709 |147 |
|11 |China | | |463,493 |702 |
|12 |Colombia |7,726 | |11,015 |132 |
|13 |Costa Rica |534 |43 |838 |45 |
|14 |Cuba |3,962 |128 |5,199 |136 |
|15 |Denmark |3,551 |149 |4,230 |151 |
|16 |Ecuador |2,527 |56 | | |
|17 |Egypt |14,218 |150 |20,045 |319 |
|18 |Spain |23,564 |440 |28,023 |476 |
|19 |United States |122,775 |435 |149,215 |435 |
|20 |Ethiopia |7,5002 |27 |9,0005 |606 |
|21 |Finland |3,463 |200 |4,016 |200 |
|22 |France |41,228 |612 |41,550 |544 |
|23 |Britain |46,038 |431 |50,363 |625 |
|24 |Greece |6,205 |250 |7,856 |250 |
|25 |Guatemala |2,539 |67 |2,7874 |68 |
|26 |Honduras |854 |59 |1,326 |49 |
|27 |Hungary |8,685 |245 |9,207 |402 |
|28 |India | | |346,000 |500 |
|29 |Iraq |3,5002 |108 |4,800 |140 |
|30 |Ireland |2,971 |153 |2,991 |147 |
|31 |Iceland |109 |33 |140 |35 |
|32 |Italy |41,177 |400 |45,996 |574 |
|33 |Japan |64,450 |466 |82,636 |466 |
|34 |Liberia |1,000 |21 |1,648 |25 |
|35 |Luxembourg |300 |54 |295 |51 |
|36 |Mexico |16,553 |170 |24,448 |147 |
|37 |Nicaragua |840 |43 |1,184 |42 |
|38 |Norway |2,814 |110 |3,233 |150 |
|39 |New Zealand |1,526 |80 |1,881 |80 |
|40 |Panama |467 |46 |764 |42 |
|41 |Paraguay |864 |40 |1,304 |40 |
|42 |Netherlands |7,936 |100 |9,956 |100 |
|43 |Iran |10,000 |136 |17,0004 |136 |
|44 |Peru |6,147 |110 |8,240 |156 |
|45 |Poland |32,107 |208 |24,448 |444 |
|46 |Portugal |6,826 |164 |8,491 |120 |
|47 |Romania |14,282 |387 |15,873 |414 |
|48 |Santo Domingo |1,345 |35 |2,277 |47 |
|49 |El Salvador |6,142 |230 |6,956 |230 |
|50 |Sweden |4,066 |187 |4,640 |194 |
|51 |Switzerland |1,434 |42 |2,150 |42 |
|52 |Czechoslovakia |13,998 |300 |12,463 |300 |
|53 |Turkey |14,9187 |399 |19,623 |487 |
|54 |U.S.S.R. |168,0002 |607 |193,0004 |657 |
|55 |Uruguay |1,941 |99 |2,353 |99 |
|56 |Venezuela |3,027 |85 |4,696 |110 |
|57 |Yugoslavia |14.438 |248 |16,040 |405 |
Sources:
1. Demographic Yearbook for 1948, Statistical Office of the United Nations, New York, 1949
2. Statesman's Yearbook for 1932, New York, Macmillan Co
3. Demographic Yearbook 1949 & 1950, Statistical Office of the United Nations, New York
4. Political Handbook of the World 1951, New York, Macmillan Co
5. Statesman's Yearbook 1951, New York, Macmillan Co
6. American Annual 1950, New York, American Corporation
7. This figure was obtained by interpolation between the data of 1927 and 1935
Notes
1. Ecuador was included in the first survey. In the second survey the figures for India and China have been included but we do not have figures for the new State of Pakistan
2. These three correlations were 0.65, 0.60 and 0.62 respectively on the raw data from the absolute population numbers to numbers Legislative this which shows that the concordance is greater when it is serves as the cube root of the population this explains in part by the that legislative effective gross population scatterplot is curvilinear while the cubic root of the population Act gives a more rectilinear appearance by legislative staff obtained the 43 States of the United States referred to in the text without a The old Union Member States see below
3. The author is indebted to Professor Paul Neurath for some of these suggestions.
4. Penrose, L.S., "The Elementary Statistics of Majority Voting," Journal of the Royal Statistical Society, Vol. CIX, Part 1, 1946, pp 53-57
5. Another study which shows similar results was made by Kendall and Stuart. The authors find that the number of seats won by one of the two parties in an election is equal to the number of votes obtained by that party. Our study relates the cube of the seats in the legislature to the whole population. We do not see a common basis for the apparently different in these examples from the law of the cube. - Mr. Kendall, M.G and Stuart, A., "Cube proportion in Election Results," British Journal of Sociology, Vol. I, No. 3, September 1950. See also the same authors a simplified presentation of the law of the cube on electoral matters in the French Review of Political Science, Vol. II, No. 2, April-June 1952, pp 270-276.
An Alphabet Of Meanings
For the Oncoming Revolution in Man's Thinking
by
Stuart Carter Dodd
Washington Public Opinion Laboratory
Seattle
The Argument
I. Man's Communicational Needs are Expanding
The title of this paper suggests that such a profound, widespread, and rapid change is taking place in mankind's symbolizing habits as to be called a revolution. But what is the evidence that the peoples of the world are embarked on such a revolution of their semiotic.1 habits – their customary inter-relatings of symbolizers symbols and the referents symbolized?
One evidence is implied in the enormous increase in the variety and volume of the world's communicating in this century. The new mass media of movies, radio and television have been born within our lifetime and are becoming used, singly or together, in some degree by most of mankind. The variety and volume has increased many fold for communicating in print and by mail, by telephone, telegraph and cable, by person in travel and in public speaking. In the older mass media of newsprint, magazines and books for example, the indices measuring the variety of topics covered, the frequency of issue, the pages of print, the number of writers or number of readers all have gone up mostly by multiples of the 1900 figures in most countries around the world. Whereas the larger majority of the world's adults were illiterate in 1900, a large majority bids fair to be literate by the end of this century. Whereas most children of school age were without a school in 1900, most of them at the present rate, will be in school by the close of the century.
With the world's multiplying education and mass communication more people are probably using more vocabulary, reading and hearing more of their own language, and learning more other languages than ever before since man began to talk. Translation between languages has multiplied in press and broadcast and in simultaneous telephonic multiple translation services at international conferences. The first electronic brains for multiple bulk-translating are under construction. The world seems converging towards the use of fewer languages. The League of Nations and now UNESCO have made cautious moves in studying the possibility of one auxiliary world language. Meanwhile English increase its lead in world usage, especially since World War II. It is probably known to a larger percent of the world’s population than ever spoke one before. Wherever in non-English speaking areas schools teach a second non-official language, English is rapidly becoming that chief second language in most countries even in Russia.
Along with this acceleration within the pragmatic branch of semiotics there is also profound, widespread, and rapid change going on in the other branches. Within semantics, the vocabularies of the world's languages are growing and changing with the vast development of the sciences and technologies. A few whole nations have already greatly changed their language within one generation. Thus Kemalist Turkey adopted the Latin alphabet and altered over half its dictionary vocabulary shifting from Arabic to old Turkish roots till city educated parents had trouble reading their children's textbooks. Thus Israel effectively revived and modernized ancient Hebrew which was not the childhood language for most of her citizens. Drastic change of the language of a people, within one generation is now possible, if sufficiently desired.
Dialects are receding around the world as newsprint and radio penetrate into jungle and desert and ocean islands. The growing international pressure for a common language has given birth to over three hundred artificial international languages such as Esperanto and Interlingua as proposed reforming of the language symbols for the myriad meanings man wants to express. Thus "Semantology,"2 invented by an Australian, offers a new semantic system in a typewritable set of some 200 international signs which are verbalizable in anyone's own language. Intended for universal use in highway signs at first, this analytic-synthetic visual symbolism compounded out of dots and straight and curved lines has expanded to claim potential, but still undemonstrated usefulness adequate for textbooks and other literature. The new discipline of semantics beginning to get academic recognition in courses and professorships and its two professional journal.
The syntactic branch of semiotics is also changing profoundly. Thus the logic of Aristotle reigned supreme for almost two thousand years. But since Whitehead and Russell published their Principia Mathematica as this century began; the classical logic has become but a small corner of the field of modern Symbolic Logic. Even mathematics which laymen are apt to think a highly perfected and static science has blossomed largely in this century in such branches as topology, set theory, statistics, and much more that monthly fills the mathematical journals.
The new "theory of information" seems a major "breakthrough" in all three branches of semiotics with possibly far reaching reverberations throughout the social sciences and in the biological and physical sciences also.3 It's units, (called "bits" for binary units) are built out of the logarithms of probabilities of alternative events."Bits" measure what has been variously described as the amount of information, or decrease of uncertainty, or complexity of organization, or negative entropy in a system. These bits and what they imply may reverse much of our current physical and metaphysical ideas about the universe tending to "run down" according to the principle of entropy. The principle of entropy expects all energy manifestations in the universe to trend towards a uniform dead level of temperature with no hotter stars of colder planets. Information theory suggests a clue to a counteraction that may be building up the universe continually. The information formula implies that random entities, whether fluxes of energy in any form, particles, molecules, mice, or tend to develop forms of "law and order." (Ref. 12) In short, information theory hints at profounder explanations of how the cosmos grows and works than man has hitherto been able to conceive – partly for lack of sufficiently sensitive and powerful symbolic tools for such conceiving. (For after all, "a function of logarithms of probabilities equivalent to twin alternative events" is a highly sophisticated unit to invent.)
Some of its students therefore believe that information theory alone will justify the developments of the twentieth century being called a semiotic revolution.
But perhaps the most important semiotic change that is already well developed has been the spread of the anthropologist's view of language and all speech behavior as part of culture. It is thus seen as a stable but ever developing equilibrium of usages – a symbolic tool that is often crude, sometimes sensitive, and always improvable. It is less and less seen as something divinely ordained and unchangeable with its varieties fixed ever since the Tower of Babel. Hence the peoples of the world seem more ready than ever before for the profound, widespread, and rapid changes in the development, structure, and functioning of all language which we call the semiotic revolution.
II. Man's Language Need Matching Expansion.
If one grants that man's communicating is undergoing an almost explosive expansion in this country, our next thesis is that languages in general are lagging in development for symbolic tools are not matching the growth of our need for better tools.
To gauge this lag requires a standard, for comparison. We propose a threefold standard composed of three criteria. These criteria expect a good symbolic system
1. to have the minimum number of symbols,
2. to cover the maximum of meanings, and
3. with maximum exactness, i.e., with a constant one-to-one correspondence between symbol and referent.
Fewness of symbols means a parsimonious or economical system. It means least cost or least effort in talking or writing. It means logical and mathematical elegance as when those disciplines seek to build their sub-systems using the fewest possible number of primitive terms and postulates.
By this standard of parsimony our languages are well below par. The world has hundreds of languages (or thousands if more finely sub-classified) whereas the ideal for maximizing efficiency of intercommunicating would be just one world language, Furthermore within each language there are many unnecessary symbols from useless synonyms to unpronounced spellings. Information engineers estimate in terms of "bits" that about half of languages in use is redundant. Putting all these together, it seems unlikely that mankind could express all the meanings now expressed in language with perhaps one percent of the world's present number of written and spoken symbols. This estimate implies that ninety nine percent of the world’s symbols are now unnecessary and are increasingly becoming wasteful of human energy.
Fullness of coverage of whatever man wants to communicate now or in the future is our second criterion for a good symbolic system for the world. This standard seems to us to be currently better met. For we have little difficulty in developing new words and enlarging our dictionaries as man’s experience expands especially in scientific technologies. Our alphabets can spell and our vocal apparatus can speak many times as many words as in our largest dictionaries. So there is ample room to grow in.
One lack here however, is the lack of standardization or multiplicity ways for forming new words. Standard rules for the growth of language would avoid much waste and ambiguity such as is now incurred when using old words with partially new meanings. Another lack of coverage is the part of experience (including imagination) which is not currently expressible in language. We have no words for many nuances of fragrances and other sensations. We cannot yet talk about the hitherto undistinguished abstractions of philosophy as these abstractions must be symbolized to become knowable. We cannot yet describe in words all the complexities of many organizations, systems, and societal situations, etc. Many of us in the social sciences believe we lack adequate mathematics (or any other symbolism) for dealing exactly with "patterns". Or complex configurations of all that happens in a community in one year for example.
Fixity of one-to-one relationship between symbol and referent, or constancy of meaning is our third standard for language. All current languages except perhaps logic and mathematics fall far short of this standard. (Though even in mathematical nomenclature there are relics of old ambiguities such as "one billion" meaning a thousand millions in America and a million millions in England.)
This standard is violated by every word that has multiple meanings in the dictionary as well as by every usage that is ambiguous, vague, or variable in meaning. An ideal language would have symbols each with just one constant referent which would not vary with speaker or hearer, with time or place, with purpose or context. The meaning of each symbol should be constant no matter who says, when or where, why or how. This ideal includes of course, having specific words for variables, for vague referents, and for partly unknown referents whenever one wants to communicate with variability, or vagueness, or state of the speaker’s ignorance. (a good language would include projective terms that invite and help the hearer supply his own private meaning as in Lewis Carroll’s "Jabber Wocky" whenever one wishes to communicate any such projective freedom.)
In sum, the symbols of current languages are neither few, nor full in coverage, nor fixed in meaning. Modern languages are deficient by all three criteria. And these deficiencies, we argue, will become aggravated as the explosive expansion of human communication unfolds in this century. What then can be done towards closing this gap? Can the structure of language be re-tooled to keep pace with its ever-growing social functions? Can man's symbols, largely developed in the age of ox carts, be streamlined somehow for the jet propelled thinking of the future?
III. An Alphabet of Meanings Can Expand Language Well.
To streamline language by making its symbols "few, full, and fixed" and so able to cope with man's rising flood of communicating, we re-propose developing an alphabet of meanings. This alphabet would seek to analyze into a small number all the meanings, all the referents or objects of thought that man may think or say, now or ever in time to come. Syntheses of those few elements would then restore any o the original set of meanings and also yield almost unlimited numbers of new combinations', for expressing meanings man has never thought of as yet. Thus our present phonetic alphabet by analyzing the sounds of a given oral language into a small number of elements is able not only to synthesize the millions of words in current languages, but could synthesize further sounds in hundreds of millions of further new words if wanted in the future.
This analytic-synthetic plan of attack has proven very practical, and highly creative throughout semiotics and other sciences. Chemists have analyzed all substances into some hundred atomic elements which recombine to yield all those substances plus all the newly created marvels of chemistry. Military administrators analyzed all the diverse role-actions of men in the armed forces so fully that they were able to synthesize tens of millions of men in powerful military organizations in World War II. Mathematicians and logicians develop their sciences by analyzing out ever more basic primitive terms, postulates, and operating rules by combining which they creatively develop their growing semiotic sciences. Linguists less thoroughly, have analyzed language, as in Basic English, down to a thousand most useful words (out of the million or more available in standard English) and then combined them to publish the Bible, for example, as a synthesis of those thousand linguistic elements.
But our proposal to develop an alphabet of meanings goes far beyond any current analysis in any semiotic science. It goes beyond the analyses of logic and mathematics in extending to all of a language and to all languages. It goes beyond the analysis of artificial languages like Esperanto, Ido, Romanol, Lingua franca, Interglossa, Interlingua, etc., and of simplified national languages like Basic Englis4 and Model English (Ref. 6, 7) in that our proposal starts where they leave off. They analyze words down to current roots and affixes in some existing languages and regularize their recombinations. Our proposal starts from those roots and affixes and analyzes them into a much smaller number of elements. At present just ten elements of meaning seems adequate to do at least half the job. In achieving such a small alphabet, we believe our proposal goes beyond Leibnitz project for such an algebra in his New Essays Concerning Human Understanding though our proposal is essentially his dream.
Before exhibiting in the next section our proposal and the "alphabet of connective meanings" we have invented as a starter, let us look at a bit of arithmetic which explores the possibility of carrying the task through before we begin it.
At about this point some readers have remarked "This is a fantastic proposal – as if one could count in advance all the thinkable thoughts of mankind!" To this we note that men similarly thought it fantastic to number the stars in sky, or the hairs of one's head, or the sands of the sea. But as science developed techniques such unenumerable mysteries proved reducible to finite counts within operationally specified limits.
So we specify here an upper-limit and ask: What is the maximum number of meanings mankind can deal with in the next thousand years? This is reducible to: "How many thoughts are thinkable by a person in one year?" Let us start by defining "a thought" as an instance-of "a meaning" occurring in one person's nervous system. In other words we translate "a meaning" into behavior, a finite flux of neural activity. However unobservable this may be at present; we can compute certain consequences from this definition such as the upper limit we seek. For one property of every neural flux or pulse of activity is that it requires a finite interval of time in which to flow. For nerve currents travel at finite speeds ranging up to 125 meters per second in men. This time interval furnishes an operational unit for measuring the number of thinkable thoughts per second. So we next define, as a start, the unit of duration of "a thought" arbitrarily as a hundredth of a second. This seems a plausible top speed since no person can typewrite, speak, or read one hundred words a second, for example. Information engineers estimate that the ablest minds tested to date cannot handle over fifty "bits" of information per second. (If this definition of a unit thought's time interval as one hundredth of a second is objected to by anyone, he is welcome to choose either a full second or a ten thousandth of a second instead. This factor of a 100 either way will make no appreciable difference in our argument.)
At a rate of 100 "thoughts" per second a person could think at most 3,153,600,000 "thoughts" a year (. 100 x 60 x 60 x 214 x 365). Multiply this by the world's population of around two and a half billion people – and quadruple it to allow for population growth a while ahead (. 1010 people).
This annual number of "thoughts" in the next thousand years would yield 3.1536 x 109 x 1010 x 103 or under 32 sextillion (. 1021) thinkable thoughts at most. This is certainly a maximal number or "ceiling" as it expects everyone to think 100 different thoughts a second, day and night all his lifetime. This also assumes that every "thought" of every person is unique. But, of course, most of a person's neural activity is highly repetitive and most people's thinking has enormous numbers of thoughts common to other people. Thus every cultural act, every word one speaks is a repetition of other persons' previous acts and speakings. So we may take this largish number of thirty two sextillion as a safe upper limit for the number of meanings- mankind could symbolize at most in a thousand years. It may well be a million million (=x 1012) times larger than the actual number of humanity's thoughts in the next millennium. So sextillions give us an ample margin.
Now a little algebra of combinations shows how small a set of elements can yield that number of combinations. Let us figure that the elements-of meaning could be twice combined like letters into words and words into sentences. Let us limit these "words" to not more than ten elements of meaning and their "sentences" to not more than 'seven such "words. " This seems equivalent to easy diction in English such as most twelve year olds would understand. Then just ten elements of meaning is enough-to yield sextillions of such "sentence" combinations. For ten elements yields 210 or 1024 different combinations or "words" without even considering any permutations or repetitions of elements. Then the 1024 "words" taken up to seven at a time in sentences yield more than 32 sextillions of different combinations. This exceeds the maximum number of thinkable "thoughts" estimated above as of the order of sextillions. And by merely allowing up to ten "words" in a "sentence", ten elements of meaning can yield billions of those sextillions of combinations.
Thus we are assured that an alphabet of meanings with as few as ten elements is theoretically ample to express all the meanings mankind can possibly express in a thousand years.
Now that we know that ten elements are theoretically enough to do the job, the question becomes can a particular set of ten elements be discovered or invented that can actually do it? Can a ten element alphabet of meanings be exhibited? Can advance specifying of a desirable semiotic invention lead to deliberate inventing of it? That is what the next Section IV starts out to do.
IV. An Embryo Alphabet Can Now be Exhibited.
Our argument that all mankind's thinkable thoughts can be expressed as compounds of a few elements of meaning, or unit symbols for all possible referents, needs clinching by exhibiting a set of elements that promise to do the :job. Let us therefore look at our embryo interlanguage called "Tilp" which was invented to explore this reductive possibility. (Ref. 2, 3, 5)
Tilp is a set of ten elementary meanings. They were chosen from extensive trials to be a "most useful" set in that their combinations yield about half of human speech as described below. They are a small enough set to be symbolized by ten letters, or by the ten digits, or by the ten fingers, or by any other decad of visible, audible, or otherwise sensible sign vehicles. We use ten carefully chosen letters in order to make Tilp both a writable and speakable language. For by including a meaningless letter, "e", for euphonic purposes the ten lettered-meanings of Table 1 can be strung together in pronounceable compounds. These compounds are built up as the equivalent of current words in any language (as illustrated in the sample dictionary Table 2). The rules for combining these ten elements and comprising the grammar of Tilp are also exhibited (Table 3) followed by an example of a prose paragraph paraphrasing the Lord's Prayer (Table 4).
The ten "dimensional" elements of meaning offered in Tilp are tabulated in Table 1. Each letter-element has two submeanings denoted by its position as a prefix or post-fix (attached to e or to any letter-element). The ten are subclassified in three "levels" as qualitative elements, quantitative elements and relational elements. These levels are Kant's "categories" stemming from Aristotle's set of ten categories. They are symbolized by exponents of 0, 1, and 2 respectively (Ref. 1, 2) in our dimensional analysis of human behavior. They subclassify the universe of all concepts or meanings on the basis of form into kinds, degrees, or relations of whatever content maybe denoted.
Within the qualitative level, Tilp has standardized four subcategories namely "space" and "time" as universal to all the sciences and to all existence; "people" as the most distinctive and observable subcategory for the human behavioral sciences; and "all else" as the complementary class (i.e., all not time, space or people). This residual category of content is to be further subclassified, of course, as far as possible or needed to describe anything in any situation at issue. Meanwhile it serves a guarantee that our symbolic system is a perfectly closed system since by definition there can be no concept, no symbol whatever, that is outside our system. It also serves as a convenient handle to deal with everything whether currently known or unknown,5 symbolized or symbolizable.
4 qualitative elements
t = time
t– = an instant
–t = a period
l = space
l– = a point
–l = a length
p = people
p– = a person
–p = a plural
i = "all else" (not t, l, p)
i– = a factor, an adjective
–i = a product, a noun
4 quantitative elements
o = negation
o– = "not," denial, zero
–o = "anti-," minus
n = 1
n– = first
–n = one
d = 2
d– = second
–d = two
k = 3
k– = third
–k = three
2 relational elements
z = relative amount
z– = less (than)
–z = more (than)
r = relative direction
r– = to
–r = from
e = euphonic base letter without meaning
The quantitative level in Tilp uses four elements (although only two, 0 and 1 are essential and although, on the other hand, all ten digits are often convenient). These four elements are "negation" (= 0), "one" (= n), "two" (= d), and "three" (= k) each with twin sub-elements as noted in Table 1. The relational level in Tilp uses two elements dealing with relative amounts (= z) as "less" and "more" and with relative direction (= r) as "to" and "from." For fuller details on these ten elements and on the dimensional analysis of behavior from which they are derived as dimensions of speech behavior the reader can consult References 1, 2, 5 and 10.
With this (or any other) set of elements of meaning, rules are needed for combining them. The algebra of sets provides such rules. Chiefly the rule for forming a product of two sets (called their "intersect") is involved. By it, any element of meaning may be "multiplied by" any element, forming what may be variously, described as an "association" or "joint occurrence" or "combined meaning" or "logical product" or "pair relation" or "conjunction." This is symbolized by juxtaposing the letters as in any algebraic product. Thus the English word "some" is "oz" in Tilp which is formed as a product or joint meaning of its two elements "o–" for "none" an z" for "more than" yielding "some" as "more than none."
For other examples of word building by forming logical products of pair relations of elements note how the personal pronouns may be derived. Since "n–" stands for "first" and "–p" stands for people or persons their product including the euphonic factor, e, is "nep" meaning "first persons" or "we." By including the factor "–n" for "one" the singular of the first personal pronoun "I," is the product "nepen."'
Similarly for the second persons, "dep" meaning "you" is built as the product or joint occurrence of "d–" for "second" and "–p" for "persons." In the same way, "kep" and "kepen" from "k–" standing for "third" mean "they" and "he (or she)."
The singular and the plural numbers simply call for the factor "–n" (or "en" if alone) for "one" and "–nz"(or "enz" if alone) for "more than one." The comparative of any adjective or adverb is denoted by "iz" which combines "i–" (= "something") and "–z" (= more than). The superlative reduplicates the comparative by saying "iziz" for "most" meaning "more than more" as children and some languages say it.
The three tenses of any verb illustrates well how compound meanings can be synthesized from analytic elements. Thus "–t" stands for a time interval (see Table 1) and this identifies verbs since every action requires a time factor. Then since "o–" denotes "zero," the product "ot" means literally "zero time" or a mathematical point in time. This can assert existence of anything at any instant of time as does one meaning of the copula "is" in English. So "ot" = "is," the copula in the present tense. Then the past tense "was," can be derived by prefixing "z-" standing for "less (of time) than the present instant." So "zot" ='"was." Then the future tense asserts "more (of time) than the present instant" by inserting "–z" to get "ozt" as meaning "will be." Finally, any verb can add "zot," "ot" or "ort" to denote respectively its past, present, or future tense.
These glimpses of word buildings in Tilp may open into a larger vista in the little "dictionary" that is presented in Table 2. (Ref. 2)
Table 2 gives the Tilp word (with a brief suggestion of its derivation) for each of the hundred "connective" words of Basic English. "Connective" words here include all verbs or "operators," pronouns, prepositions, conjunctions and adverbs and adjectives of time, place, person and number. "Connective" words contrast with "substantive" words such as most nouns naming things predominately and most adjectives naming qualities. Table 2 evidences that all the connective words of Basic English can be expressed in terms of the ten Tilp elements of meaning.
The details of Table 2 are not important here. Many of them will be refined with linguistic research. As the author is a sociologist and not a linguist, no claim is made that Table 2 is a finished product. It is rather a suggestive beginning – which strongly argues that such analytic speech is possible. Further research should continually discover better elements of meaning which subsume more meanings. Tilp should evolve. A step in this semantic evolution4 is the evidence from Table 2 – 4 that many meanings – the connective half or so current language – can now be reduced to a few elements of meaning
Table 2. The 16 Operators and 84 connectives of Basic English
|Basic English Operator|Dimensional formula |Phrase Explaining the Formula |
| |(i.e. word in “Tilp” | |
| |dimensional language | |
|be |ot |zero (0-) duration (-t), existing at any instant |
|do |et* |pure verb or act (-t), made pronounceable by the euphonic vowel, (e) |
|let |oet* |zero (o-) acting (et), do nothing, not act |
|make |it |a thing (i) acts (-t); also, acting on a thing |
|come |relet |moves (let) towards (re) La point (l-) acts (et), i.e. moves], e.g. 'A comes to B' means 'A moves |
| | |toward. B' |
|go |erlet |moves (let) from (er), i.e. 'A goes; means 'A moves from (wherever A was)' |
|get |relit |moves a thing (lit) to (re), i.e. 'A gets X' means. 'A moves X to (himself)' or 'to his own |
| | |possession' |
|give |erlit |moves a thing (lit) from (er), i.e. 'A gives B to C' means 'A moves B from A's possession to C's |
| | |possession' |
|take |replit |moves a thing (lit) to (re) a person (p), i.e.,('A takes X' means 'A moves X to A' |
|put |erplit |moves a thing (lit) from (er) a person (p), i.e. 'A puts X away' |
|send |preplit* |a person (p-), to (re) a person (p-)1 moves a thing (lit) |
|have |irot |of, from a thing, part of, possessed by (ir), opposite (-0) of verb (-t), i.e. passive reversed to |
| | |active voice; possesses. Thus English may phrase it: 'I have ears' |
|keep |irort |will (-ort) have (ir-); since: ot = be, or present tense, i.e. zero duration; z-ot = was, or past |
| | |tense, i.e. less than the present date; ozt = will be, future tense, i.e. more than the present date |
|see |oculot |eye (oculo) acting (-t), i.e. 'eyeing' in colloquial English |
|Seem |oculotro |be seen (ro) i.e. to the opposite; makes active verbs passive |
|Say |linget |tongue (ling) acting (-t) |
|Basic English | | |
|Connective | | |
|a |en |one |
|about |ool |all (oo) lines (-l1) sides |
|after |toz |a date (t-) and some [time] (oz) |
|again |det |second (de) time (-t) |
|against |ro |opposite (-o) relation (r) |
|all |oo |opposite (-o) of none (o-) |
|almost |zoo |less than (z) all (oo) |
|among |rer |to (re) and from (-r) relations |
|and |tir |of (ir) one date (t-), at the same time, simultaneous |
|any |en |one |
|as |zizo |all not (-o) less (z-) nor more than (-z) a thing |
|at |le |A spot |
|be |ot |zero (o-) time (-t), present moment, present acting |
|because |irt |acting from a thing |
|before |neloz |first (ne) line (-l) [i.e. front-and-back] positive (oz) [i.e. front] |
|before |zote |less than (zo) a date (te) |
|between |rer |among, to and from |
|but |zet |lessening, making (it) less (z-), excepting |
|by |per |from a person |
|down |kelzo |third (ke) line (-l) [i.e. up-and-dawn] negative (zo) |
| | |[i.e. less than (z-) zero (o)] |
|east |geodeloz |second (de) line (-l) [i.e. right-and-left] positive or more than (-z) zero (o) [i.e. right] earth |
| | |(geo); |
| | |geodeloz = rightward on earth (facing north as conventional norm) |
|enough |proiz |not (o-) more (-iz) to (r-) a person (p-) |
|even |irikel |third line (kel) [i.e. up-and-down] and to and from a thing (iri) [i.e. even up to] |
|ever |oot |all (on) times (-t) |
|every |oon |all (oo) ones (-n) |
|far |ezl |more (-z) distance (-l) |
|for |ri |to (r-) something (i) |
|forward |reneloz |towards (re) first dimension (nel) positive (oz) |
|from |er |(-r); primitive element |
|he, they |kep |third (ke) persons (-p) [kepen = he; kepez = they] |
|here |lezil |this (zil) spot (le) |
|how |inet |the (in) doing (et) |
|I, we |nep |first (ne) persons (-p) [nepen = I; nepz = we] |
|if |ez |is greater than, includes, is implied by, i.e. 'A ez B' means 'A if B' or 'if B then A' or 'A includes|
| | |B' or 'B is part of A', etc. |
|in |lele |at (le) a spot (le) a spot (le) |
|little |zi |less (z-) thing (-1) |
|may |oit |not (o-) making (it) [i.e. permitting, letting] |
|much |iz |more (-z) things (i) |
|near |zel |less (z-) distance (-l) |
|no |oe |denial (o-), zero; primitive element |
|north |geoneloz |positive (oz) on first dimension (nel),on earth (geo) |
|not |o |by definition as primitive element |
|now |toil |this(zil) moment (t-) |
|of |it |from (-r) a thing (i) |
|off |ler |from (-r) a spot (le) |
|on |oler |not (o-) off (ler) |
|only |enen |one, one |
|or |ezt |greaten, make more than, add |
|other |ozil |not (o-) this (zil) |
|out |lelo |opposite to (-o) in (lel) |
|over |kelozle |up (keloz) spot (le) |
|please |otiprit |be (ot) desiring (iprit); person (p-) acts (-t) towards (r-) something (-i) [i.e. strives towards, or|
| | |desires]; the adjective (i-) or participle of the verb ‘prit’ = desire |
|so |izlel |that (izl) distance (el) |
|some |oz |more than (-z) zero (o-) |
|south |geonelzo |earth's (geo) first dimension (nel) negative,(zo) [i.e. longitude line ] |
|still |izotot |was (zot) and is (ot property (l-), existed and existent now |
|such |irizli |or (ir) that (izl) class (-1), that kind of |
|than |ere |relation, from (er) and to (re) |
|that |izl |more (-z) distant (-l) thing (i) |
|the |in |one (-n) thing (i) |
|then |tizl |that (izl) date (t-) |
|there |lizl |that (izl) spot (l-) |
|this |zil |less (z-) distant (-l) thing (i) |
|though |zet (or ez) |but (zit); if (ez) |
|through |releler |to (re) in (lel-) from (er) |
|till |reten |to (re) one (-n) time (t) |
|to |re |(r-); primitive element |
|together |len |with (len) |
|tomorrow |eniz eten |one (en) more (iz) day (eten) [i.e. one (en) period (et)] |
|under |kelzole |a point (le) below (kelzo) [i.e. the third dimension, (kel) negative (zo)] |
|up |keloz |third (ke) dimension (-l), positive (oz) |
|very |iziz |most [i.e. more (iz) more than (iz)] |
|well |priz |a more (-z) thing (i) to (r-) a person (p-); approval |
|west |geodelzo |earth's (geo) second (de) dimension (l), negative part (zo) [i.e. left when facing north] |
|what |ini |the (in) thing (i) |
|when |tin |the (in) moment of time (t-) |
|where |lin |the (in) place (l-) |
|while |int |the (in) duration of time (-t) |
|who |pin |the (in) person (p-) |
|why |erin |from (er) the (in); rizorti for past or future thing, cause or purpose |
|will |ozt |acting (-t) in a future period (ort) [i.e. positive (oz) period (-t)] |
|with |len |one (en)-spot (le), at (le) one (en) |
|yes |ci |a thing (-i) [asserted] |
|yesterday |zen eten |less (z-) one (en)day (eten) |
|you |dep |second (de) persons (-p) Depen = you, singular; depz = you, plural] |
* These dimensions are a minimum set which has been found necessary as far as explored to date by the writer, for the data, concepts, and principles of the social sciences, the physical sciences, and the semantic sciences (logic, mathematics, linguistics) for the top level of classes. Further subclassifying splits up 'things' into as many subdimensions as needed (such as 'oculo' and 'ling' above). How much further subclassifying will be needed is still unknown, of course.
The examples of word compounding in Table 2 have also illustrated the next level of combining whereby grammatical rules tell how to combine words into sentences. These rules relate words to words. They fix the word order in sentences such as the sequence used in Tilp, namely: "subject - verb - object - indirect object – with all further modifiers following the modified term." The rules of grammar dealing with case and number, person and comparison, tense and voice and mood are all summed up for Tilp in the ten exceptionless rules stated in Table 3.
Table 3. Grammar Complete in Ten Rules
1. Inflections are eliminated. Sentences comprise invariant roots and particles only, i.e., entities and relators.
2. Modifiers follow the modified.
3. Case is expressed by prepositions and the word order: Subject, verb object; e.g., possessive, of, = ir; indirect object, to, = re.
4. Number is indefinite, although en = one = singular, -z = more.. plural.
5. Person occurs in three persons only: nep = first person,[I, we]; dep = second person [you (singular), you (plural)] kep = third person [he, she, they].
6. Comparison is by more: more = iz; most = iziz.
|7. |Tense |Past |Present |Future |Verb sign |Perfect a zo: e.g., 'I will |
| | | | | | |have gone' nep erlet zo otz |
| | |z- |o |-z |-t | |
| |Copula |zot |ot |otz |et | |
| | |= was |= is |= will be | | |
8. Voice Inverter, ro, makes a verb passive. E.g., 'He was hit' = kep hit zotro.
9. Mood,, etc.
a. Imperatives use et. 'Stop!' = Stop et.
b. Conditionals (subjunctive) use ez. ez = 'if'.
c. Participles use adjective sign i-. it = 'being'.
d. Infinitives use noun sign -i. ti = 'to be'.
10. Interrogation inverts subject and verb. E.g., 'Did he come?' Come kep sot?
Table 4. The Lord’s Prayer in Tilp in 13 Elements of Meaning
A sample dimensional analysis of a paragraph of speech behavior, B
|Dimensional categories |The Acts |
| |(verbs) |
| | |
|Algebraic version B= | |
| |Aa; |
|Inter-rogative Version | |
| |What is done? |
|Tilp letter | |
| |-T |
|SYNTACTIC CRITERIA |PRAGUATIC CRITERIA |SEMANTIC CRITERIA (relating | |
|(relating symbols to |(relating symbols to |symbols to referents) | |
|symbols) |symbolizers) | | |
|Consistency Acquaintancy |Reliability Popularity |Verifiability | Acts |
|Recurrency Parsimony |Durativity Density |Standardizability |by People |
|Importancy Accuracy |Predictivity Generality |Stability Translatability |in Time |
|Efficacy |Simplicity Utility |Controllability Univocability|and Space |
|Synergy | |Durability Multipliability |for Values |
| | | |with Symbols |
| | | |and Materials |
| | | |and Other Context |
|These 24 criteria (= " M ") for the excellence of scientific theories or symbolic systems are derivable as a |
|Cartesian product of the set of 3 semiotic relations (= the columns = ) times the set of 8 type-parts of any |
|transact (= the rows = )Thus M = the matrix of 24 "system-metrics." Each system-metric is an |
|operationally-defined percentage-like index measuring one criterion. (See Ref. 11) |
Consequently we believe that an alphabet of meaning, perhaps developed from Tilp, could revolutionize human speech and thinking and thereby significantly accelerate man's cultural evolving.
References
1. Dimensions of Society, MacMillan, 1942, 944 pp.
2. Systematic Social Science (offset) University Bookstore, Seattle, 1947, 788 pp.
3. "A Logical Interlanguage," Forum, Palestine, September 1943.
4. On Measuring Languages," Journal of the American Statistical Association, Vol. 44, March 1949.
5. "Tilp - a ten letter alphabet of meanings," General Semantics Bulletin, Spring-Summer 1951, No. 6-7.
6. "Model English," General Semantics Bulletin, No. 10, Summer 1952•
7. "Model English" Ch. 10 in W. N. Looke and A: D. Booth Machine of Languages, Wiley, 195, 243 pp.
8. "The Transact Model - a predictive and testable theory of social action," Sociometry, Vol. XVIII, No. 4, December 1955.
9. "The Transact Model Evaluated - by 24 'system-metrics' or measured criteria" (to appear).
10. "The Reiteration Rule for Organizing Semiotics" (to appear).
11. "Introducing 'System-metrics' for Evaluating Symbolic Systems: 24 criteria for the excellence of scientific theories" (to appear).
12. "The Probable Acts of Men - the 'momental growth' models for predicting mass behavior" (to appear).
Notes
1. Semiotics, the study of symbols (and hence most of communicating) includes three branches, namely:
Semantics as the relations of symbols to their referents, or meaning thus including most of language;
Syntactics as the relations of symbols to symbols thus including grammar, logic and mathematics;
Pragmatics as the relations of symbols and the symbolizers thus including all speech behavior and much of social psychology and aspects of sociology.
2. Bliss, C. K., International Semantology: Vols. I-III, Semantography Publishing Co., Smarcubra Bay Road, Pagewood, Sydney, Australia, l948.
3. Questler, Henry, Information Theory in Psychology, Free Press, 1955, 436 pp.
4. Ogden, C.K. The General Basic English Dictionary, W. W. Norton & Co, 1942
5. For standard subclasses of this first residual class we have used (Ref. 8,9
action (= a) to include all human behavior especially speech behavior,
values (= v) or "desiderata" to include whatever men want, i e., try ,to get or keep
"circumstances" (= c) to include as second residuals all not t, l, p, a or v
These six qualitative elements of meaning (which we prove a more versatile and powerful set than hitherto used in Tilp.
6. To prevent collision of letter symbols some convention will be needed. Letter groups such as "geo" and "stom" here might be enclosed in quotes to flag them as a Tilp unit of meaning: and not their ordinary English meaning if any.
7. Of course a mechanical device involving no analysis is possible whereby every meaning of every word in the largest dictionary is given a number. Less than ten million numbers would probably be used, we guess. This requires only the permutations and combinations allowing repetitions of the ten digits taken not more than eight at time Thus with Tilp-like "words" which never exceeded eight letter-elements, one could, name, without any economizing analysis such as this paper proposes; all the meanings man now expresses in words. But this requires the arbitrary learning of millions of bonds or names-for-referents whereas our alphabet cuts down this learning to a few dozen bonds.
8. Medieval Marseilles passed a decree forbidding merchants to ciphers (a name for something non-existent!) and enjoining them thereafter to figure "in clear Roman letters." Imagine today figuring one’s change from CXVIII dollars for purchasing LXXXVIII items at XVC cents each!
9. A communications engineer reading this manuscript noted that while speech is about 50 percent redundant, this redundancy is useful in assuring perception in spite of auditory and semantic "noise." We agree that redundancy can be useful but the present somewhat random redundancy could be replaced by more efficiently controlled redundancy which Tilp could produce by simply repeating statements – as Air Controller Towers now do.
10. These 24 criteria supersede the less comprehensive set of ten indices in Reference 1. The latter were designed more restrictedly to measure the familiarity and the regularity of candidate international languages. In the case of Tilp its three indices of familiarity to the world's population would be zero and its seven indices of regularity would all be 100%. The 24 more discriminating system-metric indices would range, we believe, from 0 to 100% for Tilp.
A "Doubling and Halving" Technique for Measuring Causation Among Opinions
Stuart C. Dodd and Sung Chick Hong
Washington Public Opinion Laboratory, University of Washington
This paper describes exploratory testing of an operationally defined model whereby causal analysis is possible among two opinions in one poll. The model is used in a controlled experiment on some hypotheses of intergroup attitudes.
Reprinted from Language & Speech, Vol. 6, Part 2, April-June 1963, pp. 68 – 87
The Model
The fractioning model that is developed and partially tested here is intended to go beyond the usual correlational analysis of a population's opinions as asserted on one poll. It is intended to open the way for causal analysis more generally. But the illustration of the model, which is reported here as a controlled experiment on some hypotheses of intergroup attitudes, is limited strictly and entirely to the realm of polled opinions, i.e., to speech behavior as observed in the standardizing and checkable context of a poll. The model is intended to help in analyzing1 eventually any basic opinion at issue as a speech transaction that is specifiable as a speech-action-in-its-full-context of the causes, costs, and consequences of that basic opinion. Towards this distant goal of opinion research, this paper reports a first step in specifying some model operations for analyzing within one poll the alleged causal relations between just two opinions which may be named " A " and " B ".
This simple two-variable model is an operationally defined theory of "pair-causation." It comprises five hypotheses or submodels according as
1. Opinion A influences Opinion B (i.e., " A is a cause of B "), or
2. B influences A, or
3. they are mutual influences, or
4. neither influences the other, or
5. both are influenced by a common set of causal circumstances, C.2
Here "influence" is defined as: A is called an " influence on " B (or " a cause " of B) to the extent that changes in the alleged effect, B :
1. occur in time sequence, A earlier and B later ;
2. occur with non-zero correlations ; and
3. occur when all other contextual circumstances, C, are constant or correlated with changes in B. A "pre-correlation" index of percentage causation is then the part correlation of changes (or partial correlation less rigorously) between the alleged cause and its alleged later effect when other possible causes, C, are controlled.
100 r21A (2B • 1C) ≠ 0 = the " per cent influence "
(of A on later B with context, C, controlled) Equation 4
This defines causation operationally as an algebraic product (representing a behavioral joint occurrence) of three factor-events, namely:
a sequence (of positive and negative changes in two variables),
a correlation (between those changes, in a specified population and period),
an isolating (of this two-variable system)3
Testing a One-Way Influence (Between Two Opinions)
As a pilot exploration with this causal model, one-way influence as by Eq. 1 was tested between two concurrent opinions, dealing with enforcement of some national mores, in an evening class population of 557 students at the University of Washington in Seattle. The correlated opinions selected for causal analysis were:
Opinion A was the belief of the group polled that in their community imaginary subgroup who were dubbed " Ethnians " to be free of contaminating historic associations, behaves provocatively, violating some norm in one of five ways (specified below), each way being measured in three degrees (called " fully " " half way ", and " none ").
Opinion B was the group's later expressed opinion of that subgroup in terms of a ten-point rating from extreme disliking to extreme liking.
The first hypothesis is stated as:
"If only Opinion A (imputing provocative behaviour to a subgroup) is changed, then Opinion B (disliking that subgroup) will be changed correspondingly."
Both opinions exist concurrently in the respondent group of students polled and are correlated. But are they also causally related? In which directions and in what ratios? This inquiry, as an offshoot of a larger study, asked in a controlled experiment which effectively isolated the two opinions at issue: Will changing Opinion A then change Opinion B? This is the issue in Hypothesis A above. As belief that a subgroup behaves fully "provocatively" changes to belief that it behaves half as provocatively and then changes further to the belief that it behaves not at all provocatively – all while nothing else whatever is known about the subgroup – will the initial disliking felt towards the subgroup show two jumps in the rating towards lilting that subgroup more ?
In formal terms Hypothesis A is here : If (as a cause) a group's opinion (= A) imputing an undue degree of specified provocative behaviour to a subgroup is changed by fractioning it (as from full imputation to half-imputation to no-imputation), then (as the effect) that group's opinion (= B) will change significantly, provided all else is constant or uncorrelated with Opinion B, from disliking the subgroup towards lilting it about as well as the respondent's own in-groups.4
A. Procedure: Change stimulus opinion A by fractions in isolation
Five forms of "provocative" behavior were chosen as hypothesized examples of deviance from some U.S. social norms to such a degree that they were considered a threat by U.S. citizens (as asserted in this poll). These behaviors had been hypothesized by G. A. Lundberg (1954) to be cases of at least mild violation (which are often attributed to Jewish subgroups) of some mores to the extent that citizens would react by disliking the subgroups who were perceived as violating those mores most. The mores concerned:
1. undivided loyalty to the nation, and
2. striving towards equal opportunities and away from monopolistic or unfair practices in constitutionally guaranteed respects for its citizens.
The five so-called "provocative" behaviors may be listed as:
1. undue ethnocentric behavior,
2. disproportionate political influence,
3. allegiance to a foreign state,
4. monopolistic holding of high-status jobs,
5. undue special privilege.
Each was specified in a paragraph of some 150 words as the characteristics of an imaginary "Ethnian" group in an hypothetic community into which the reader has moved, as follows:
Form 1:
Specification of "full" undue ethnocentric behavior: "Suppose you moved into a new community and found that a group of Ethnians lived there too. Suppose that these Ethnians were very strongly convinced that they were superior to all other people. They constantly tried to prove their superiority by citing examples of members of their group who were famous in the arts, sciences, business, and politics. Most of them clung to holidays, habits, and public practices that were very different from the rest of the community. They regarded it as a great loss of purity if one of their children married a non-Ethnian. They formed separate charity organizations solely to help their own members. They showed great loyalty to their own group in all matters and tended to regard almost all criticisms as a sign of hatred on the part of a larger community. This led them to react to non-Ethnians with very strong dislike and antagonism."
Form 2:
Specification of a "full" disproportionate political influence: "Suppose you moved into a new community and found that a small group of Ethnians lived there too. Suppose that these Ethnians were especially active in politics. An unusually large number of them, considering the small size of their group, were office holders, and the Ethnians were very strongly inclined to vote for only those candidates who were Ethnians. Any politician running for office had to get his picture into the papers with a group of Ethnians if he was to have a chance of being elected. Likewise, appointments to offices controlled by Ethnians would go to Ethnians in more than proportionate numbers. The rest of the community usually felt that Ethnian office holders were much more concerned about the welfare of Ethnians than of the whole community. Certain types of legislation hardly came up for consideration because it was well known to the legislators that such bills would be offensive to the Ethnian group."
Form 3:
Specifications of "full" allegiance to a foreign state: "Suppose you moved into a new community and found that a small group of Ethnians lived there too. Suppose that these Ethnians retained a very strong attachment to Ethnia, their foreign homeland. The local Ethnians always celebrated the national holidays of Ethnia and paid little attention to holidays of the larger community. The picture of a prominent leader of Ethnia was proudly displayed in many Ethnian homes in this community. They were also very active in helping Ethnia. For example, they carried on extensive and highly publicized money-raising "drives" for the benefit of Ethnia not only among their own members but in the community as a whole. The Ethnians maintained a very skilful and efficient organization to gain favorable publicity for Ethnia, and brought strong pressure to bear on public policy regarding Ethnia. In other words, the Ethnians seemed to pay their primary loyalty to Ethnia, rather than to their local community."
Form 4:
Specification of "full" monopolistic holding of high status jobs: "Suppose you moved into a new community and found that a small group of Ethnians lived there too. Suppose it is found that Ethnians had much better jobs and more money than the average. The Ethnians lived in the larger homes in the more desirable neighborhood in the community. The Ethnians claim that this is merely the result of the Ethnian's superior abilities. Other people in the community, however, feel that the prominence of the Ethnians is due to the favoritism of Ethnians for their own group, through appointments and other influences they exert. In any event, it is a fact that Ethnians are over-represented in high-status occupations like medicine, law, music, and politics. It is also assumed throughout the community that one's chances of getting a top job, or important business contracts, or important managerial positions are higher if one is an Ethnian than if not."
Form 5:
Specification of "full" undue special privileges: "Suppose you moved into a new community and found that a small group of Ethnians lived there too. Suppose that these Ethnians were always trying to gain certain special privileges. This made them, for example, exempt from public criticisms of the type that other groups had to put up with. That is, the Ethnians had special organizations to protect Ethnians against criticism or attacks which other members of the community had to put up with as a matter of course and without any special protection. Thus, the Ethnians would make organized protest about books, movies, and speeches which unfavourably criticized them. These Ethnians would try to suppress such criticism through censorship, boycott, or influence in the publishing or communication industries. Even the police would sometimes hesitate to arrest an Ethnian for disorderly public behavior because of fear that Ethnians would complain about it as persecution and would bring pressure to bear on higher officials to reprove the offending policeman."
After reading the paragraph about Ethnians, who might be called "Group 1," the respondent was asked, "How would you feel toward the Ethnians?" and was provided with a ten-point Stapel scale for rating his opinion ( = B) of them. The end points of the scale were anchored by the phrases "Like very much" at the top and "Dislike very much" at the bottom. Each of the five subsamples of respondents (averaging 111 persons) was asked about just one type of provocative behavior so that there would be no multiple associations with "Ethnians". (This implies techniques for getting and for testing comparability among the subsamples.)
Then each sample was introduced on the next page to a different community of Ethnians, who might be called "Group 2", and who were specified in a 50-word paragraph as showing about half the former behavior of Group 1.5 Again respondents rated their degree of liking-to-disliking of the Group 2 Ethnians.
Then a third paragraph, labeled "Group 3", presented the Ethnians in yet a different community with none of the behavior previously specified, but "just like everyone else in the community."6 A third rating of lie-to-disliking was recorded.
Thus for each of three degrees of provocative behavior (i.e., "full", "half", "none") the mean rating of liking for such a subgroup was subsequently observed and computed. The differences, A, between successive mean ratings measured the change in feeling consequent upon a change in the provocative behavior of the Ethnians as perceived and upon nothing else since the respondent could have no other knowledge or association whatever with those Ethnians This polling technique of fractioning the isolated stimulus in a sequence of steps while holding all else constant and measuring the consequent change in each later response achieves our operational definition of causation as a time sequence of correlated events which can always be reiterated under fully reiterated conditions.
B. Findings : One-way causation was isolated and measured
[pic]
Figure 1: Correlation between the mean ratings (= B) of "Like-Dislike" expressed toward Ethnians by all respondents and the degree of provocative acts (= A) previously imputed to Ethnians.
As evidenced in Figure 1 (and Table 1), this controlled experiment clearly proves that decrements in these five provocative behaviors imputed to Ethnians correlated with and caused significant increments in liking those Ethnians. There were ten tests or Ws where the alleged cause, the degree of each of the five Ethnian behaviors, was changed from full to half or from half to none. In every one of the ten testings of the hypothesis the consequent mean rating of liking changed significantly and showed a trend to be in direct proportion to the prior change in the perceived provocative behavior. Without exception among the ten half-steps of change (and also among their combined five whole steps of change from full provocative behavior to none), every change in the antecedent causal opinion produced a definite change in the subsequent effect opinion. Decreasing the provocative actions of a subgroup as seen by a group did cause significantly increased liking for that subgroup, when all other influences were rigorously held constant. This establishes the fact that perceiving provocative actions engendered antagonism to the actors.
This basic finding of clear-cut one-way causation between two opinions held by a majority group was checked and confirmed in several ways, viz
1. This cause-effect relation of opinion A to Opinion B is shown alike by all five different samples, each of some 111 respondents, as exhibited in Figure 1 and Table 1 (and also by the further sample from the pretesting).
2. It is shown in all five different examples of provocative behaviors from social, economic, domestic, and international political fields.
3. This causal relation between a majority's opinions of a minority's provocative behaviour and of consequent disliking for that minority is further evidenced by differing statistical analyses and indices as follows :
a. The fifteen deltas (Δ) measuring changes in ratings of liking were all statistically significant increments (at the 5 per cent level) as the imputed provocative behaviour decreased. The second-order differences between half-step deltas and the whole-step deltas were also significant at the 1 per cent level. (This checked on whether the midpoint of responses represented statistically significant differences from each extreme point and were not just semantically tripped off by the word " one-half " occurring in the stimulus paragraphs.)
b. The slope of the regression lines in Figure 1 is always downward. It shows a linear relation of two of the provocative behaviors (to the lilting rating) and a slightly but significantly curvilinear relation of the other three provocative behaviors. (As far as three points can show this.)
c. The fifteen mean ratings in Figure 1 show a correlation coefficient of – 0.924 (N = 15) between the amounts of perceived provocative behaviors and of mean liking for the behavers. This means crudely (from Eq. 1 in cumulative form) that Opinion A was 85 per cent (= r2) of the total causation of the disliking Opinion B as here measured.' (The remaining 15 per cent is accounted for as described in footnotes below.)
It should be stressed that the controlling or partialling out of other possible causal influences, C, called for in the part correlation formula, Equation 4, was here far more perfectly achieved by the Ethnian device. Instead of an average statistical control of just one possible third cause, C, the experimental control by asking for reactions to unknown Ethnians held all other possible causes constant among the respondents.
d. The last correlation (r = — 0.924) between two sets of fifteen means was recomputed among individuals where individual differences (as in generosity of assigning rating of liking, for example) attenuate the correlation. These five correlations in the five subsample populations, averaging 111 students each, ran from r = — 0.40 (for " Foreign allegiance ") to r = – 0.70 (for "Special privileges"). These correlations between changed opinions towards Ethnians among individual voters, though attenuated (and so below the correlation of means of changed opinions) were yet in every case negative in sign, substantial in amount, and statistically significant. Both sorts of correlation indicate highly causal relation between the two opinions. A, the seeing of provocative actions; and B, the later expressing of feelings of dislike to the actors in consequence. These correlations also indicate-that the more the irrelevant individual differences are controlled (as correlating means), the higher the causal relation is shown to be.8
4. This causal relation of perceived-provocative-action to disliking-the-actors here transcends the opinions of the particular respondents who rated the Ethnians. This was tested by recomputing Fig. 1 and the correlation of – 0.924 in (3c) above from a "purified subset of those 557 respondents". From previous data we hypothesized that the characteristics of the raters which would be most likely to bias their ratings of minority groups would be the raters'
a. color, or
b. religion, or
c. anti-Semitic attitudes, or
d. stereotyping (of Jews).
Accordingly, these four possible biases in rating were controlled or reduced by selecting the 131 white, Protestant, non-anti-Semitic, and non-stereotyping raters. (The last two traits were defined by selecting raters below the median on scores from questions in the poll which were taken as identifying anti-Semites and stereotypes.)
This purified, or less biased, subset of 131 raters reproduced Figure 1 data and curves and correlation almost unchanged. The r of – 0.924 (between the three degrees of the five provocative behaviors vs. the fifteen mean liking-to-disliking ratings) among the 557 original raters with all their biases became a correlation of – 0.926 among the purified raters with all those biases reduced. As in Thurstone scales, the measurement of causation among opinions here by our "fractioning technique" was highly independent of the potentially biasing opinions of the judges (on the substantive issues whose correlated and sequenced ratings measured that causation).
5. This causal relation was further checked by using the Washington Public Opinion Laboratory's calibrated equal-interval, social distance scale as an alternative to the liking-disliking rating. The average intercorrelation of these two indices in the five subsamples was r = – 0.86. There were thus two somewhat different, though highly correlated criterion indices of the alleged effect, antagonism. (Of course in other situations this effect might have many other contributing causes which were not studied in this study seeking to isolate the provocativeness-upon-disliking opinions with all else constant.) The liking rating was interpreted as reflecting the respondent's personal opinions more, while the social distance scale was interpreted as reflecting his perception of his in-group's public opinion more.
The two criterion indices were expected to yield similar results and largely did so. The correlations of the imputed provocative behavior with each criterion were in general so similar that the detailed causal analysis with deltas (A) of change was undertaken here only with the disliking index. The generally similar results from the two indices support our belief that this technique of causal analysis by fractioning the antecedent opinion in controlled experiments transcends the particular indices and opinions reported here. But this belief should be checked by further fractioning experiments, of course.
Testing a Reinforcing Influence (Among Three Opinions)
A second body of findings concerned a second causal factor which seemed to multiply the disliking effect of a main group towards a subgroup when caused by the main group's imputing provocative behaviors to the subgroup. This second reinforcing factor was the explicit perception of threat. Each subsample of the main group was asked to rate on a ten-point scale their perception of the threat posed by just one of the five provocative behaviors, as follows:
In general, how threatening do you think:
1. extreme ethnocentric behavior by minority groups would be to our national unity ?
2. it would be to our system of democratic government to have minority groups enjoying disproportionate political influence ?
3. extreme foreign allegiance by minority groups would be to our national defense under present circumstances ?
4. extreme monopolistic holding of high-status jobs by minority groups would be to our free economic system ?
5. extreme special privileges enjoyed by minority groups would be to the feelings of solidarity among community members ?
Each of the five subsamples were divided at the median rating into a "threatened" half and an "unthreatened" (or less threatened) half of the respondents. Table 1 and Figure 1 were then recomputed for the threatened compared with the unthreatened raters.
In every case the threatened respondents increased their disliking for the minority when seen as highly provocative, more than did the less threatened respondents. The perception of threat (= Opinion A) acted as a reinforcing cause to increase the effect (= Opinion B) of disliking the threatener. For every one of the five provocative behaviors in the five subsamples of respondents, the regression lines of Fig. 1 became steeper as the perception of threat increased.
To measure the correlation of the threat factor here, one crude index is the per cent of increased disliking associated with increased perception of threat. To compute this, let delta, A, denote the maximum change in the ordinate or dislike rating in Figure 1 as the provocative behavior goes from " none " to " full ". For the unthreatened respondents this increment of disliking, Δu, when averaged over the five provocative behaviors, was 33 rating units, or 39 per cent, of the whole range of the ten-point (i.e., nine-interval) rating scale. For the threatened group, the increment of disliking, or Δt, was 4.5 rating units, or 50 per cent, of the possible range of the dialing scale. Thus the delta or effect in increased dialing in the threatened group was 28 per cent
(i.e = ) larger than in the unthreatened group. Perception of threat here increased the disliking rating 28 per cent on the average ; the perceiver group who felt more threatened disliked the perceived subgroup 28 per cent more than did the perceivers who felt less threatened.9,10
Table 1
Mean ratings (= B) on "like-dislike" expressed towards Ethnians by the white Protestants with high threat score (= AD compared with those by the white Protestants with low threat scores (= - AD, when Ethnians changed their degrees of imputed "provocative acts," for all five hypotheses
|Name of Lundberg's five hypotheses as |Degrees Of |Degrees Of " Provocative Actions " |
|to provocative actions |Threat Perceived | |
| | |Extreme |Half |None |
|Ethnocentrism, Al |Low |3.19 |4.58 |6.97 |
| |High |2.70. |4.70 |7.86 |
| |Difference |-0.49 |-0.12 |0.88 |
|Political influence, A. |Low |2.73 |5.00 |7.03 |
| |High |138 |4.33 |7.18 |
| |Difference |-1.15 |- 0.67 |0.15 |
|Foreign allegiance, A. |Low |3.49 |5.18 |6.68 |
| |High |3.19 |536 |7.33 |
| |Difference |- 0.30 |0.38 |0.35 |
|Economic monopoly, A. |Low |4.65 |5.73 |6.27 |
| |High |3.88 |5.33 |6.30 |
| |Difference |- 0.77 |- 0.40 |0.03 |
|Special privileges, As |Low |1.94 |3.61 |6.48 |
| |High |1.52 |3.08 |6.81 |
| |Difference |- 0.42 |- 0.53 |0.33 |
|All 5 hypotheses |Mean |- 0.62 |- 0.22 |0.34 |
|averaged |difference | | | |
| |between low | | | |
| |vs. high threat | | | |
| |perceivers | | | |
| | |Average |Average |Average |
| | |greater |greater |greater |
| | |disliking |disliking |liking |
| | |by the |by the |by the |
| | |highly |highly |highly |
| | |threatened |threatened |threatened |
Note : A rating of "1" denotes maximum disliking; a rating of "10" denotes maximum liking. (A minus difference indicates that the mean rating given by those with high threat scores is smaller, i.e., expressing more dislike than that given by the low threat group. A plus difference indicates the reverse.)
Discussion
Further exploration of Lundberg's five hypotheses was made11 of which the testis of this fractioning model for analysis of causation among opinions was a part. From that larger analysis of factors correlated with the majority's indices of dishing and social distance towards each of eighteen minority groups in the United States, it became clear that the factor of perceived provocative behavior (with five subtypes) was of secondary importance in determining the majority's antagonism. That antagonism was predominantly correlated with American experience in socio-economic and political ways with these groups who were mostly associated with waves of immigration from specific countries of Europe, Asia, Africa, and Central America. This mass experience seems deeply stereotyped and very observably indexed a priori by ranking the minorities according to averaged ratings of skin color from darkest (Negro) to lightest (Swedes).
The high correlation observed in a statistical population of eighteen minorities between this color ranking (with its massive historical associations) and the antagonism indices (of rated dislike and of social distance) masked and partly reversed the correlations ( = rAB) expected by the Lundbergian hypotheses between antagonism indices (B) and imputed provocative behaviors (= A1, 2, 3, 4, 5 ). Thus, for example, in the case of imputed foreign allegiance (Hypothesis 3) among the seven colored minorities of non-European origin12 the Lundbergian hypothesis was clearly disconfirmed by a positive correlation of rAB = 0.75 (where the hypothesis expected negative correlation, i.e., increases in imputed provocative behavior vary with decreases in liking). But among the nine white minorities of European origin the Lundbergian hypothesis was clearly confirmed by an (expected) negative correlation of r = – 0.74. Thus only when other causes of disliking a minority are controlled does the partial cause studied here (the imputing a provocative behavior, Opinion A) become clearly visible and demonstrated.
To demonstrate causal relationship over and above any observed correlations requires rigorous experimental control in:
1. changing the alleged cause upwards and downwards (only downwards changes were made in this exploratory testing) and observing later correlated changes in the alleged effect ; and
2. isolating those changes in a closed system at first free from all other influences (as in reacting to Ethnians who are totally unknown except for their changing as Opinion A specifies).
These considerations have led to more explicit specifying of the implicit preconditions in Lundberg's hypotheses. Accordingly, our revised phrasing below states explicitly and measurably:
1. that these five types of provocative behavior cause antagonism when perceived as pursued to abnormal and threatening degrees, and
2. that this causation must be observed free of other obscuring influences.
Lundberg's hypotheses as refined by Hong and Dodd may be stated in their revised 1961 version as:
Indices of antagonism that a majority group feels toward a minority group tend to change directly (with whatever time lag):
A. With large changes in the degree of each of the following five behaviors which the majority group imputes to the minority group:
1. Undue ethnocentric behavior
2. Disproportionate political influence
3. Allegiance to foreign states
4. Monopolistic holding of high-status jobs
5. Undue special privilege
and A'. With the degree of threat which the majority group perceives in each type of such behavior, provided that all other antecedent circumstances, C, that may be correlated with the indices of antagonism are held constant.
Most simply: If men see a group's acts as a threat to their well-liked norms, and if all else is constant, then they will tend to dislike all members of that group."
Our critique of the findings above in the partial exploration and partial testing here of our fractioning technique of causal analysis of opinions should note first of all what this testing has not tested.
In testing whether Opinion A causes Opinion B, the converse Hypothesis b has not been tested here, namely, that Opinion B is a reverse one-way cause of Opinion A. Only Hypothesis a was studied, namely that : If a group sees a subgroup behaving with sufficiently undue provocativeness to the extent of seeming a threat to the main group, then that main group will tend to dislike that subgroup and show large social distance test scores towards it. The converse Hypothesis b says that: If one dislikes a subgroup, among his antagonistic actions may be either
a. initiating perceptions of the provocative behavior (done by subgroup members) and/or
b. developing exaggerations of any pre-existing preconceptions of its provocative behavior.
Informal pretests made (a) seem unlikely, but (b) seems to us to be plausible and well worth testing in future research. There well may be feedback effects whereby citizens who dislike any minority group for whatever reasons will become prone to impute more provocative behavior to that minority than is objectively warranted. This secondary or feedback reaction seems likely to us, in view of the large literature in this area, to be of lesser amount than the primary or initiating force of perceiving the provocative behavior. But the fractioning technique may help to measure their current shares in the total variance which reflects how either variable may have grown or decayed or lagged as a stereotype.
Since this experiment did not test the "converse" Hypothesis b, it did not test the "interaction" Hypothesis c which says that imputing provocative actions and disliking the actors are jointly present, each causing the other to some percentage extent as reciprocal influencings. In the future Hypotheses a, b, and c could be tested in one poll, by doubling and halving in turn Opinion A and Opinion B while measuring the effect on the other opinion (when sequencings were properly rotated among matched groups).
A second deficit of the present study not dealt with in this paper is to extend this study of polled opinions imputing provocative behaviors to checking "validity" studies of actual such behaviors in community settings. How close is the one-to-one correspondence here between speech and action, the imputed and the actual behavior, the majority's belief that the minority acts "provocatively" and the extent of such actions by the minority? Under what measureable conditions does what degree of misperception occur? Ordinarily for a widespread, strong, and long-held perception of one group by another to exist sociologists assume a high degree of correspondence between that perception and the facts to hold currently or in the past. If group X dislikes subgroup Y, there must be a definite set of causes to each of which the fractioning technique can with sufficient research on the opinions involved assign shares in the total causation of a measured effect. This validity problem of correlation between a measured opinion and the objective fact was here highly controlled by the "Ethnian" device. The paragraphs describing the Ethnians are assumed to have a one-to-one correspondence to the facts specified therein and to be so reacted to in this imaginary situation because the respondents have no contrary knowledge of, or experience with, Ethnians. But in general, the validity correlations will have to be further studied between speech-action-in-a-poll context and life-action-in-a-community context. How closely does the poll transaction correspond to its referent transaction?
A third deficit, urgently calling for further research, is to study comprehensively all the other influences, the total causal setting other than the imputed provocative behaviour, which increase liking for the minority. How much does raising the minority's status in income, education, and social class generally, and their conforming to the majority's norms, result in their becoming more liked? These other influences or causes of disliking should be inventoried, measured, intercorrelated, and weighted by their feedbacks and interactions into a single system or comprehensive transaction model as fast as research can be organized, staffed and financed.
A fourth deficit is the unrepresentativeness of the population. The five subsamples of Seattle evening students cannot represent the nation's opinions nor experience with minorities. When this and further exploratory studies will have developed a more comprehensive, reliable, and valid questionnaire, well-scaled for fractional analyses, then it should be applied, and periodically re-applied, to a cross-section of the American public. Then these national generalizations might be compared internationally to discover just how local and dated and culture-bound they may be.
A fifth deficit here was incomplete control of the timing. The sequence of changes in the imputing Opinion A went from full to half, to none. This should be also tried in reverse and in irregular sequence and any differences due to sequencing should be averaged out.14
Turning from what this pilot study did not test to what it did test, it measured in one poll and within the limited universe of speech behavior the extent of one-way causation by one opinion on another, when all else was highly constant. It thus develops a generally useful technique for causal analysis in human speech behavior.
The particular opinions studied were those specified by Lundberg's hypotheses about ethnocentric and other minority behavior (when perceived as unduly extreme and threatening) producing antagonism in the majority towards that minority. The Lundberg hypotheses were fully and consistently confirmed when their revised preconditions were rigorously specified and experimentally achieved in isolation. (This, of course, is the implicit requirement in the testing of any scientific hypothesis.) The five hypotheses expected and clearly and reliably found that : Insofar as a group sees a subgroup's behavior:
1. as provocative, i.e., with
a. undue ethnocentrism or
b. disproportionate political influence or
c. allegiance to a foreign state or
d. monopolistic holding of high-status jobs or
e. undue political privilege ; and
2. sees that behavior as a threat to specified interests or norms (mores) or the main group, and insofar as all else is constant (so that there are no modifying further factors complicating the observed situation), in just so far the perceived provocative actions will correlate with and cause general disliking for the actors.
Among the many further modifying factors which may cause varying degrees of disliking for eighteen minority groups in the United States, the color factor (as a convenient summarizing index of all the historic differences of social status, or dominance, etc. that are associated with it in U.S. experience) proved a much more important factor than any of the provocative behaviors studied. In the statistical population of these eighteen minority groups, the high negative correlation (r = – 0.74) between disliking ratings and provocative behaviour ratings within the nine white or European groups tended to be reversed to the high positive correlation (r = 0.75) within the set of seven colored (or mixed) groups of non-European origin. Americans have different expectations apparently from white peer groups than from colored lower status groups. These differentials need to be more explicitly allowed for in stating hypotheses, such as Lundberg's, which deal with a group's defensive behaviour under perceived threats to its mores.
Conclusions
This study reports exploratory testing of an operationally defined model whereby causal analysis is possible among two opinions in one poll. In summary, the data reported here, illustrating the "fractioning technique" for causal analysis among opinions show that:
1. Opinion A was the full cause of Opinion B (as specified herein) when all else was highly constant. This was confirmed without exception, in five population samples, in five different types of behavior, by respondents with four biases and without these biases, by three variant statistical techniques and by two alternative indices of the criterion variable.
2. Whether Opinion B also caused Opinion A (by feedback or retroaction) was not studied here.
3. Both opinions are so strongly correlated with a complex but measurable background of Opinion C as to obscure and even reverse the high correlations between A and B where:
Opinion A (= "provocativeness") was a group's belief that a subgroup behaved provocatively to a threatening degree in five specified ways;
Opinion B (= "disliking") was that group's dislike for the subgroup;
Opinion C (= "color-rank") was whatever a ranking of the subgroups by skin color (i.e., white, yellow, brown, red, black, in rated degrees) meant to the main Seattle group in U.S. experience.
4. Rigorous experimental control in testing causation among at least two opinions is now increasingly possible (given adequate resources for researching) such that percentage shares of causation seem assignable : to a primary or initiating cause; to a reinforcing cause ; to a counteracting cause ; to a feedback or regulating cause; to a common background of causation ; and to dimensional observation errors such as
a. differences among individual respondents and differences among population samples of respondents,
b. differences in particular opinions or speech actions,
c. differences in the timing (appearing as differing sequences, frequencies, speeds, or durations), and
d. differences in the "facets" which state
i. origin points,
ii. units, and
iii. indices in observing all the foregoing.
5. The sociological generalization was here confirmed (when rigorously specified and observed free of other factors) that: People tend to enforce their mores.
Appendix
Is the "half" response a semantic artifact?
For the more critical psychologists and semanticists reading the foregoing exploratory study, the refinements below may be of interest.15
During the analysis of these data the Washington Public Opinion Laboratory staff debated the extent to which the two intervals in the total range of the liking-disliking attitude were a semantic artifact tripped off by the respondent's seeing the word " half " in the stimulus paragraph which specified the behaviour of the " Group 2 " Ethnians. That this semantic factor had negligible influence is indicated here by two sorts of evidence:
a. the statistical tests reported below, and
b. a psychological consideration as to the " whole " which was cut in " half ".
1. A variance analysis
If A causes B (or influences the occurrence of B), then a change in A should be associated with a change in B. Within certain limits there should be an association between the magnitude of a change in A and the magnitude of a change in B.
It is this relationship that has been tested in Hong's data on evening class student responses to the manipulated image of Ethnian behavior. With five different kinds of "provocative" behavior depicted in five different questionnaire versions, the fictitious Ethnians were first presented in Group 1 as extremely provocative. Then they were presented in Group 2 as being only half as provocative. Finally, they were presented in Group 3 as being just like the majority—not at all provocative. Thus there are three comparisons that can be made:
a. from 1 to I,
b. from / to 0, and
c. from 1 to 0.
Some critics of this study had held grave doubts of the validity of Hong's Ethnian data on the ground that it seemed that the respondents might merely be imposing an artificial consistency on their responses. When told that the Ethnians – whom they have just said they dislike to a certain extent for being extremely provocative – are now only half as provocative, it would be "expected " that they might say they now dislike the Ethnians only half as much. Or, at any rate, the decline in dislike associated with the decline in provocativeness from " extreme " to " not at all " might be " expected " to be about twice the decline associated with the drop from " extreme " to " half ". What bothered the critics was this thought: If disliking Ethnians drops twice as much when Ethnians change from 1(X) to 0(X) as when they change from 1(X) to 1/2(X), this may mean only that the respondents recognize that 0 is twice as far from 1 as is I, and that X itself never entered into the response at all.
To be sure that Hong's data meant anything at all, then, it was necessary to confirm that X has something to do with the response patterns. Analysis of variance accomplished this. Three analyses of variance were performed. When provocativeness was reduced from "extreme" to "half", the change in disliking varied significantly between the five forms. Similarly, when provocativeness was reduced from "extreme to "none", the change in disliking varied significantly between the five forms. And again, when provocativeness was reduced from "half" to "none", the change in disliking varied significantly between the five forms. This supports the conclusion that respondents were not merely giving responses that were superficially "consistent" with the numbers presented in the questionnaire. The kind of provocative behaviour described made a significant difference in their responses. In other words, X did enter into the equation.
Having established that the data mean something, then, the next question is: What relation is there between X (the five provocative behaviors depicted in the questionnaire) and the liking-disliking response? With respect to all five types of provocative behaviour, it can be said that a reduction in the amount of the described provocation is associated with a reduction in the amount of disliking. Moreover, the reduction from "extreme" to "none" is, in all five cases, associated with a significantly larger drop in disliking than is the reduction from "extreme" to "half as much", or the reduction from "half as much" to "none". In broad terms, then, it might be said that the reduction in disliking tends to be roughly proportional to the reduction in provocativeness of the fictitious Ethnian minority.
However, this statement would, literally, imply a linear relationship in which the reduction from "extreme" to "half as much" should produce the same amount of change in dialing as does the reduction from "half as much" to "none". This is not the case in three of the five instances. With respect to ethnocentrism, economic monopoly, and special privilege, there is a significant difference between the amount of reduction in disliking associated with the first half-step drop in provocativeness and the amount associated with the second half-step drop in provocativeness. For both ethnocentrism and special privilege the larger reduction of disliking occurs when provocativeness is changed from "half as much" to "none". But for economic monopoly, the larger reduction in disliking occurs when provocativeness is changed from "extreme" to "half as much". For political influence, and for foreign allegiance, the two half-steps do not differ significantly. For the five hypothetical provocative behaviors taken together, then, there is not a clear and simple pattern. The data neither support a (generally) linear nor a (generally) curvilinear relation between amount of change in provocativeness and amount of change in disliking. All that we can conclude is that in this simulated experiment there is an association (sometimes roughly linear, sometimes curvilinear) between the two types of change.
It must be stressed that this analysis pertains to the relation between two kinds of change, or between the amount of change in two opinions, and is not just a study of the relation between two variables, as such.
These exploratory data thus support the weak form of the hypothesis that expects some kind of positive correlation but are inadequate to test a stronger form of the hypothesis that expects a specified curve of correlation (e.g., as a linear, or logarithmic, or exponential, or other specific curve). The doubling and halving technique, recommended above for future testing, provides more adequate data with four (instead of three) abscissa points observed for X and with unequal spacing (i.e., 2, 1, I, 0) to test better linear and logarithmic subhypotheses.
2. "Half" of what?
A further consideration arises from inspecting Figure 1. There, in the right-hand column, the respondents first reported an average liking for Group 1 Ethnians of around a score of 3. They next reported (in response to a stimulus paragraph containing the word " half " several times in describing Group 2 Ethnians) an average liking score on a ten-point rating scale of around 5, or the " neutral " position. Third, they finally reported (in response to the paragraph on Group 3 Ethnians who were without any imputed provocative behavior) an average liking score of around 6.5. These two changes in averaged liking scores from 3 to 5 (= 2 score points) and from 5 to 6.5 (= 1.5 score points) were not equal changes (indicating curvilinearity in some of the data). Nor were they half of any semantically apparent whole. They were not half of the gap from the initial disliking score to neutral liking, nor half the gap from initial disliking score to neutral liking, nor half the gap from initial disliking to the maximal liking score of 10. They only approximate half of the semantically unexpressed and psychologically still implicit range from the lowest disliking score of 3 to the highest liking score of 6.5. Even this emerging psychological range of 3.5 score points only became explicit when the respondents made their third judgment. This argues that the respondents expressed their affective attitudes towards Ethnians on the basis of the cognitive meaning to them of the three descriptive paragraphs of Ethnian behavior including the word "half", and not just as a semantic response to whatever the word "half" meant to them.
Of course, in future use of this "doubling and halving technique", this possible semantic factor can be more cleanly controlled by a Thurstone scaling technique. This would select, as stimuli, paragraphs without quantity cueing words such as one of the attitude continuum bounded by the two paragraphs defining the end-points of "half", but paragraphs which previous ratings by judges had placed as at the midpoint the range. Such an equal-interval scale of imputed Ethnian behavior could extend the present crude 2-intervals to n-intervals. Then the hypotheses of causal relation between Opinion A and Opinion B can be more strongly stated as expecting a specified curve of relationship, and the empirical testing of it can become definitive.
References
1. Dodd, S. C., Systematic Social Science, Social Science Series No. 16 (offset edition) (American University of Beirut), 410, 1947
2. Lundberg, G. A., Schrag, C. and Larsen, 0. N., Sociology (New York), 1954
Notes
1. Our implicit further assumption is always that all these causal analyses must be testable eventually by their resynthesis such that the scientist can predict: "If A (the full set of causes).is reiterated, then B (the speech behavior or opinion at issue) will be reiterated (i.e., repeated identically). Insofar as such "If A, then B" statements are produced by public opinion; researchers with high generality among diverse opinions, high probability among respondents, and high reliability among repolls, in just so far they help make behavioral science an exact science with predictive generalizations which can approach the status of scientific laws
2. The equations specifying these five chief hypotheses constituting the model are (where A denotes a change in time and pre-subscripts "1" and "2" denote sequence in time):
|Hyp. a: |Is A a cause of B? |Answer: r1ΔA 2ΔB • 1ΔC ≠ 0. |Equation 1 |
| |i.e., Is A a prerequisite of B? | | |
|Hyp. b: |Is B a cause of A? |Answer: r1ΔB 2ΔA • 1ΔC ≠ 0. |Equation 2 |
|Hyp. c: |Is C a cause of both? |Answer: rAB ≠ 0. |Equation 3 |
| | |when Eqs. 1 and 2 vanish. | |
|Hyp. ab: |Are A and B mutual causes, |Answer: Neither Equation 1 nor 2 vanish, i.e., both | |
| |i.e, interacting? |equations hold. | |
|Hyp. ab': |Is neither A nor B a cause of the other? |Answer: Both Equation 1 and 2 vanish. | |
The experiment reported here tested Hypothesis a and started exploring Hypothesis c, both in one poll of a population's attitudes towards minority groups.
3. For fuller operational defining of causation by this model, see S.C. Dodd, Systematic Social Science, Social Science Series No. 16 (offset edition) (American University of Beirut, 1947), 410.
4. This "weak" form of the hypothesis expects only correlated changes of A and B which includes both rectilinear and curvilinear correlation. The "strong" form of the hypothesis (not studied here) would specify the form of the curve relating A and B as linear, logarithmic, harmonic, etc.
5. A sample paragraph specifying "half of full" undue ethnocentric behavior is: Group 2 –"Now, suppose that the Ethnians in this community changed to become somewhat more like the rest of the community. Suppose they came to feel only half as superior as before. They only felt half as strongly about inter-marriage, and only half as ready to dislike non-Ethnians."
6. The stimulus paragraph specifying "no undue ethnocentric behavior" was : Group 3 – " And now suppose that the Ethnians changed completely, so they became just like everyone else in the community. They no longer felt superior, nor did they object to intermarriage. They took criticism just like everyone else."
7. In computing this correlation in future research, several technical flaws revealed by this exploratory study should be remedied by improving the frequency and the control of the sequencing of the changes of Opinion A the alleged cause). One flaw is that the r of – 0.92 in Figure 1 is computed from cumulative data and not from the uncumulated deltas or changes called for in Equation 1. Cumulating always builds up the r as it lengthens the range. But in Figure 1 since the two increments in the X variable (imputed provocative behaviour) were equal and so without dispersion, the X variable became a constant and incapable of any correlation. Hence the statistical significance of the deltas was tested as the chief criterion. In the future the deltas should be varied. Extending the present "halving technique" to a "doubling and halving technique" can do this. This fuller technique presents:
a. the " standard" stimulus paragraph specifying the " full" Opinion A (score = 1) (which should be sufficiently short of the extreme so as to allow it to be " doubled " later in the respondent's estimation),
b. half the standard stimulus (score = 0.5),
c. none of the standard stimulus (score = 0), and also presents:
d. twice the standard stimulus (score = 2).
Respondents were found generally able to react to the stimulus Opinion A when "doubled" in amount even though they were unaware of any units of A. Eventually the four statements of four degrees of Opinion A should be scaled by Thurstone's technique to make their intervals exactly correspond to their scores above.
This doubling and halving technique yields four points instead of three for each causal opinion changed and produces three instead of two first-order differences and four instead of one higher-order difference. By thus increasing the frequencies, the N, the reliability of the observations can be improved.
This technique also yields more ways of varying the sequencing of positive and negative changes. Such properly varied sequences can obviate spurious correlation that is correlated with time and so with all continuous growing.
This doubling and halving technique has several further advantages. It can test better whether the cause and effect variables, Opinions A and B, are:
a. graphically related linearly or curvilinearly,
b. algebraically related additively or multiplicatively,
c. behaviorally related as simple separate actions or as mutually modifying interactions,
d. psychologically related by the Weber-Fechner function of " response equals log stimulus" as generalized in Guildford's power series (B = k Aa),
e. best measured in binary scales or powers of two and bit units, (H = Σ p log2p), thus viewing causation as an organizing process resulting from random interaction of structured sets of elements.
(See Stuart C. Dodd, " How Random Interaction Organizes a Population," Synthese, 12: 1 (March, 1960).)
8. Three further controls of irrelevant variables here could bring the observed correlation of –0.92 up to unity.
a. If we assume that differences among the five subsample POPULATIONS (averaging 111 students) were measured by their differences in rating the third "Group 3" (which has no provocative behavior and should thus be the same for all five subsamples), we could therefore add or subtract a constant to their mean liking which characterizes each sub-sample. This would raise or lower each whole curve in Figure 1 by an amount that is constant for each curve so as to bring their right-hand end points to coincide. This reduces the variance within one column in Figure 1 while leaving the variance between (the means in) the columns unchanged so that the correlation in Figure 1 would be increased. This control of population - differences would raise the r = – 0.92 near to unity.
b. If, further, the differences among the five provocative behaviors were controlled so that their effectiveness in causing disliking were equalized, the correlation would become almost unity. This means measuring each subsample's change in its liking rating in terms of its own range or variance or total change of ordinate in Figure 1. It would make all curves in Figure 1 have their left-hand end points also coincide. Again this reduction of variance to zero within one of the three columns in Figure 1 must result in a higher correlation.
c. A third control would be to equalize the SCALING UNITS of the five provocative behaviors continuum. If this were done (as by a Thurstone scaling technique), the paragraph intended to describe a half-way point could be either rephrased or relocated on the continuum. This, in effect, tends to make the midpoints of the five curves in Figure 1 coincide and, when combined with (a) and (b) above, would compel the correlation to become unity.
Thus the total variance of the disliking criterion variable is here accounted for by the provocation variable (85 per cent) and by measurable differences among the five populations, the five behaviors, and the scale units (15 per cent) for the combination of these three errors of observation as measured here. All this controlling of errors of observation and of sampling illustrates the possibility of complete causal analysis of two opinions by combining appropriately controlled experimentation of the speech behavior that is called an opinion by the rules of logic and statistics in dimensional variance analysis.
9. In terms of the social forces operative the percentage result is the same. Our operational definition of a social dyne or unit of social force (F) is "one person (P) changed from scratch one unit of any specified index (I) in a unit period (T), F = PI/Ta." The social force (in per capita terms) of perceiving a threat (for the ordinal unit period definable as " from before to after the event of perceiving a threat") was here just one social dyne, i.e., one rating unit of increased disliking (4.5 – 3.5 = 1). This is a 28 percent increase (= 1/35) of the social force of 3.5 social dynes from perceiving provocative behavior. (See S. C. Dodd, "A theory for the Measurement of Some Social Forces," The Scientific Monthly, 43: 1 (July, 1936).)
10. This shows that perceiving a threat and disliking the threatener were here correlated but may or may not be a causal factor since we did not here measure the change in the perception of threat in one set of people in "before and after" polls.
11. Since this further analysis of Lundberg's hypotheses was tested with respect to eighteen minority groups in the United States in 1958 only to the correlational level of analysis and not to the deeper causal level with which the present paper is concerned, it is to be reported more fully separately.
12. The averaged rankings by three judges by skin color "as popularly perceived" in the U.S.A. for the ethnic minorities was: FOR 7 COLOURED GROUPS OF NON-EUROPEAN ORIGIN – Rank 1, Negroes ; 2, Amerinds ; 3, Mexicans ; 4, Porto Ricans ; 5, Filipinos ; 6, Japanese ; 7, Chinese. FOR 9 " WHITE " GROUPS OF EUROPEAN ORIGIN – Rank 8, Yews ; 9, Italians ; 10, Greeks ; 11, Poles ; 12, Russians ; 13, Frenchmen ; 14, Germans ; 15, Irish ; 16, Swedes.
13. This leaves for further research all questions of still earlier or other causes, of course. Why do people see these five acts as threats? To what interests (norms or mores)? To what degree? How justifiably? How intensely are these imputing opinions held? How were they in turn built up from personal experience and the community's culture? This study reported here only tried to analyze and test in isolation, as specified by Equation 1 (but with deficits cited below), the one causal link in society's vast network, namely that: "SEEING THESE FIVE ACTS AS THREATS CAUSES DISLIKE, IF ALL ELSE IS CONSTANT, FOR ALL SUCH ACTORS."
14. A difference suspected as due to sequencing was the last item in Table I. Here the more threatened respondents gave a mean rating of disliking the subgroup that was 0.34 rating points greater than the rating given by the less threatened half of the respondents. This observed and significant difference of 0.34 should have been a zero difference since no threat was here perceived by either half of the respondents. It might be due
a. to "overcompensating" or "relief " by those feeling threatened when the threat was removed, or
b. to some difference in friendliness in the two halves of the respondent population. Experimentally varying the sequence of stimuli in further testing could settle whether (a) or (b) or some other factor,
c. accounts for this at ()resent unexplained difference of 0.34 points in Opinion B (disliking).
15. Section (a) here is drawn from notes by William R. Canon, Jr., who supervised the relevant computations.
The Likability Models for Predicting Probable Acts of Men
A theory of values
By
Stuart C. Dodd
Institute for Sociological Research
University of Washington
Seattle
July 1963
I. The “Likes” Model
A. The Problem
Towards solving the problem of forecasting a man's acts, let us observe first that there are many historic theories of human nature that have used a host of synonyms for "As a man feels, knows, and has done, so he is likely to do again." This trinity of causal factors, which we shall call the three "modes" of action, weaves through human experience in myriad versions, blends and organized contexts. Consider the sample table of these versions for a quick survey. (See Refs. 15, 21)
Table 1: Correlates of the Three Modes of Speech Behavior
|Discipline | Feeling “Likings" | Knowing “Likenesses" | Doing “Likelihoods" |
|Psychology |Affective |Cognitive |Conative |
|Sociology |Appreciative |Cognitive |Evaluative |
|General |Emotions |Intellect |Conduct |
|Philosophy |The Beautiful |The True |The Good |
|Kant's "Ultimate |Feeling |Knowing |Willing |
|modes" | | | |
|Chief Institutional Field |Art |Science |Ethics |
|Somatology |Endomorphs |Ectomorphs |Mesamorphs |
|Neurology |Afferent and Autonomic nervous |Central nervous system |Efferent nervous system |
| |system | | |
|Physiology |The senses (and glands) |The brain |The muscles |
This table shows that a large part of the problem has been a semantic one – the lack of standardized terms, like C.G.S. units in Physics, by which to identify and measure these modes in any context however masked, diluted or compounded.
B. The Observing
To solve this semantic subproblem the Likes Model (Ref. 21) offers the three standardizing indices of a Liking felt, a Likeness known, and a Likelihood of an act done.
These modes indices in "Likes" phrasing answer the questions: How much do you like V1 (the valued-object at issue); How alike is it to Vs (any standard V for comparison); and How likely is it that you will do By (which may be the (valued) behavior of voting, buying, listening, etc., that is to be predicted. These three questions seek to specify operationally in polls and so to isolate and measure the looser terms "affective, cognitive and conative behavior." Each mode is thus itemized in observing it as perceived by one man with respect to one item object, V1, at one time when considered alone and rated ,on a standardizing self-anchored scale such as shown in Exhibit A. This vertical rating scale for feelings is substantially the Stapel scale of Gallup polls and the Cantril ladder with self-anchoring endpoints. (Cantril, Hadley, "A Study of Aspirations," Scientific American, Vol 208, )
The feelings are reported in terms of the words "I like," "I dislike" because of impressive evidence that these words work more generally and better than any other synonyms in English. Thus Edwards Personal Preference Scale, the most thorough and exact study yet made of personal values with social desirability controlled, states its 450 items in terms of "I like." (Edwards, Allan, "Edwards Personal Preference Schedule." Psych. Corp., New York, 1954, pp. 17) Cantril's recent resume of transaction studies finds strong and growing neurological evidence for reticular systems influencing total response in rats and men that seem best named as "I like" and "I dislike" systems. (Cantril, Ridley & Livingston, Wm. K., "The Concept of Transaction in Psychology and Neurology." Journal of Individual Psychology, Vol. 19, 3-16. May 1963.)
Our Washington Public Opinion Laboratory found from much experimenting that "I like" and "I dislike" revealed feelings more purely and widely and easily than any alternative English words. The appropriate grammatical variants3 of "liking-disliking" seem usable from infants to elders, from feeble-minded to savants, towards most any object: a thing or person, act or state, word or idea, in almost any context. Therefore we propose to standardize it as the polling unit of feelings (or emotional intensity, or approach-withdrawal attitudes) for summarizing, as much as just one index can do so, the total reaction of a person towards an object.
For standardizing, among many English synonyms and possible systematizing neologisms, we recommend the terms in Table 2.
Table 2. Likes Terms Suggested
|Mode |Noun Form |Adjective Form |Verb Form |Chief Technical Meanings |
|Feeling |The liking |liked |to like |to prefer |
|Knowing |The likeness |alike to |to liken |to see as similar |
|Doing |The likelihood |likely |to likelify |to make likely, to intend |
|Any mode |The "likes" |"likes" |to "enlike" |to like and/or liken and/or |
| | | | |likelify |
un- = not dis- = the opposite of...
Stilted or novel forms need not be used by people preferring traditional usages; but may prove useful for purposes of scientific regularity and system.
Next, the Likeness indices can measure a person's knowledge about the behavior to be predicted in degrees and as comprehensively as desired. Thus one might ask: Do you think Person X will be most alike to or unlike the best of our presidents (checked by later using "worst" instead of "best")? Which of the income classes listed here is most alike to yours? Which of these answers is most alike to your idea of the best answer? These "likeness scales" are often similar to a semantic differential scale.
To construct a likeness index simply take the n multiple choice questions of any poll and let each choice define an "itemized" likeness index. Let each choice define a "standard of comparison," Vs, at the right end of a rating scale with either its denial or its opposite as defining the left end. Then ask the respondent to check his opinion at some point on that scale to show how alike his opinion is to one end of it or the other. More coarsely, every closed end answer checked in a poll implicitly says: This answer, out of the set here, is most alike to my opinion.
The general form of a Likeness rating then is shown in Exhibit A. The chief difference between a Liking rating, LkF, and a Likeness rating, LkK, is that the former subjectively relates oneself to an object in a total approaching or withdrawing response (e.g., "I like V1") while the latter objectively relates one object to another (e.g., "V1 is alike to VS'').
Next consider how the doing mode can be more precisely yet generally measured by a Likelihood rating. To estimate the probability of the future behavior, B, from the respondent's present answer, two questions may be asked. One question is: "How many times could you have done this (B) and how many times did you do it during the past twelve months" (or other suitable period)? But his recent past may differ from his perceived future which is the more predictive index. So the better question is: How many times could you do this (B) during the next twelve months and how many times do you intend to do it, all things considered?" These last three words tend to include among the predictive factors (in addition to the respondent's internal likings felt and cognitive and relevant likenesses seen) all the external pressures and constraints he may see as operating upon his eventual behavior. Is the housewife who firmly intends to vote likely to find that a. sick child, lack of a baby-sitter, or other external circumstances thwarts her intentions?
An alternative and more easily answered question asks: "How likely or unlikely is it, all things considered, that you will do BV in the next twelve months? A format for this Likelihood Rating is shown in Exhibit A.
The three ratings of liking, likenesses and likelihood may now be averaged (with equal weights until better regression weights are known) to get an arithmetic mean "likes" rating, LkFF+K+D. Alternatively a geometric mean rating
[pic]
is recommended for study. For the three modes may well combine in one's nervous system as mutually influencing factors in a product and more than just separate addends in a sum so that
[pic]
This sub-hypothesis that the "likes product" predicts better than the "likes 'sum" can be tested by comparing their correlations (from suitably built indices) with the later behavior as it transpires.
C. The Likes Hypotheses
Now that a sharp tool for dealing with the three modes has been described, let us use it in formulating hypotheses to start predicting some probable acts of men such as in buying or voting, in listening or reading, in working or playing, in eating or sleeping, in planning or aspiring, etc. Of course, the more the behavior studied is a repetitive mass action coming soon in time, the better the predicting is likely to be.
What men like most they are most likely to try to get, in so far as other factors are constant? But this obvious rule is often complicated by inadequate knowledge and experience of the best means to get what is wanted most. Therefore knowing the full content of the desired experiencing in the future together with its still later consequences and its earlier causes or means to it will be further factors guiding men's acts and so will be factors in predicting. In addition to these desires and knowledge, man's customary practices will channel and predict much of their current and future behavior. All these factors furthermore will combine and. feedback into various complexes such as their beliefs (built from past experience, current knowledge, and hopes and fears for the future) as to what acts seem most likely to get them what they want with least effort, waste, or unwanted mixtures.
From all such common sense considerations, let us formulate our three mutually supportive likes hypotheses. But instead of using the cruder all-or-none "If .A, then B" form, we shall use the more exactly graded form which says "Insofar as A, the antecedent conditions, exist fully and solely, the consequent behavior B is expected." Our "likes model" or system of six interworking subhypotheses each a composite of many more specific ones, expects:
A poll tends to predict a probable act of each man insofar as:
1. it measures well (= the methodological sub-hypothesis)
2. what that man sees (= the transactional sub-hypothesis)
3. as most liked, (= the "likings" sub-hypothesis)
4. most alike to it, and most likely (= the "likeness" sub-hypothesis)
5. and most likely (= the "likelihood" sub-hypothesis)
6. while unseen factors
are neglectable. (= the usual sub-hypothesis of constant residuals)
Note that this statement of the transact that is called "predicting probable acts of men" includes explicitly the methodology along with substantive content. It specifies the antecedent behavior of both the observer and the observee since either alone is less than the "full context" needed for fullest prediction. If the poll is poorly constructed or conducted, the predicting may likewise be poor.
Note also the transactional emphasis on haw the observee perceives or defines the situation. Will Mr. X accept this invitation to be chairman of Committee Y? To predict his answer it is not enough to catalogue his relevant likes and dislikes, nor his similar practices and interests. One also asks how does he see it all. How does he perceive its consequences to his image of himself, his wife's image of him, the image that others he refers to may have, the time it will take in competition with other interests, his ability to do the task, the external situation that may help or hinder him, and many other factors. All these factors are not static but may be partly controlled by a persuasive person who modifies X's perceptions and so influences the outcome of the predicting. Thus the poll must not merely inventory the predictees' total set of relevant likings, likenings, and perceived likelihoods but also how he weights and organizes them into a total decision. How does X see it, all in all?
For the "core sub-hypotheses" on the three modes in terms of likes phrasing a simpler formula may help in daily use. It is to remind oneself that: Acts seen as most alike to the acts that are most liked and recently likely are likeliest. This is our "likes rule." It is the core of our "likability transaction" for which all the other hypotheses in this paper are the context.
These perceptions of the predictee can be systematically probed in fuller context, by extending this modes model into a "modes-and-tenses model." (Ref. 15, 21) Ask each of the questions about his likes, in the past, in the present, and in the future tense. Thus in addition to asking Mr. X what he now likes, ask also what he did like--what his relevant interests have been--and what he will like--what he anticipates enjoying under relevant future conditions.
For a specified example of this modes and tenses model study the appended questionnaire on "Ratings of an Organization." (Exhibit B) It shows how questions in the three tenses can be adapted in this case where the behavior of supporting or opposing a national voluntary organization was to be predicted. Thus for the future tense it asks about the perceived goals of the organization, for the present tense it asks about activities, and for the past tense it asks about achievements. This questionnaire also shows how to ask questions probing and predicting cybernetic factors in the three modes. These social feed backs help to enforce the norms of any group and so influence the probable acts of a man.
From such social feedback which helps to form reputations, much of the mores of a culture or norms of a society may grow up. For what are such norms if not the three likes factors among people with a common history? People living together develop morals and ethical codes such as the Golden Rule. This rule says when spelled out in "likes" language, "Do to others what seems most alike to what you would like most to have others likely to do to you in like circumstances." This last sentence might be called a prescriptive and social form of the likes hypotheses. For it extends the likes modeling from predicting one man's behavior to predicting the normally probable acts of men. These "normally probable acts" mean jointly, of course, a normally distributed action, a normatively produced action, and the normal results of such an action. The word "normally" if extended in the three tenses then implies here a descriptive content of an action, explanative causes of it, and predictive consequences.
D. The Testing (Ref. 25)
These likes hypotheses are being loosely tested, informally and implicitly, in daily life by people everywhere--who are mostly unaware of it. Anyone may test them more explicitly insofar as he observes-and records the following four steps of scientific methods about predicting the probable decision (=B) of any Mr. X.
1. Inventory X's relevant "likes," i.e.:
a. X's likings pertinent to B,
b. his recent acts likely to forecast B,
c. and his knowledge and skills most alike to the means of his attaining B;
2. Measure the degree of each of these three "likes";
3. Correlate them in all pairs to get regression weights;
4. Systematize them in a causal hierarchy.
Thus may seem an impossible counsel of perfection. But in the near future, with office computers as available as typewriters, it may prove more efficiently predictive for a Board of Directors in making appointments, instead of spending an hour of a dozen men’s time roughly evoking n facts about the candidates, to have each of them in ten minutes fill out a questionnaire probing these n facts exactly (and let the m candidates do likewise to combine self-perceptions with peer-perceptions) and feed these n (12/m) facts to the computer to marshal and come up with a comparative prediction about each candidate's relative fitness for the job as seen both by himself and. by the set of responsible judges.
A simpler test might be for the reader to rank a set of n objects of one specified sort, first by his liking for each and again by the likeness of each to his most liked object in the set. The likes hypothesis would be confirmed (as to its sub-hypothesis expecting correlations of likings and relevant likenesses) insofar as another person (who knew only the reader's likeness rankings) predicted that, if the first object were unavailable, the reader would choose (i.e., rank next most liked) the object he had ranked as most alike to the unavailable one. Such tests, of course, need to have their preconditions mapped more fully and exactly. Within what range of objects, of what sorts, under what further specified conditions, is a given prediction most likely to agree with the outcomes?
The most rigorous test of the likes hypothesis is the power of the likes indices in predicting specified behaviors better than alternative and comparable indices. A case of such testing is reported below in Table 5. Here the indicies representing the three modes of speech out-correlated 162 psychological questions and scales showing a multiple correlation of .78 (R2=.61) with the criterion behavior to be predicted. This correlation means that the modal indices (See Exhibit 2) here accounted for 61% of the variance of the predictand. The remaining 39% is presumably due to lack of fulfillment of sub-hypothesis #6 that all variables other than the likes indices shall be constant.
Of course, all market research and other polls that are checked against some predicted outcomes are testing synonyms of the likes hypotheses, though usually in less full or efficient form.
E. The Applications
(Ref's 3, 4, 5, 7, 10, 11, 13, 16, 21, 22, 23)
The core likes models and their larger context, the likability models, we expect will have wider potential applicability than other formal models yet proposed for predicting human behavior. To test this claim, comparison to any other model should be made in terms of the following three features:
1. Testability in controlled experiments in the future ranging (within specifiable probability limits) from the most minute predicting for an act of one man an hour ahead up to the most comprehensive predicting of the activity of a nation a century ahead. This testability will be limited by time and all other costs and currently imperfect techniques and much more, of course, but the model seems to us to be in principle testable insofar as society may desire to do so.
2. Precision, in offering operationally specified scales for its variables (in generalized dimensional form to be made more specific to each act as needed so that calibration indices of its reliability, validity, etc., can be determined).
3. Universality to most behaviors of most men at most times in most situations.
With this large claim goes the usual condition of .a "hard" science. Submit it to the test of experiment. Will it predict? How much of our expectations from the likability models will be confirmed, rejected, modified, or restated accordingly as controlled experiments may test and refine its sub-hypotheses more and more fully?
F. The Systematizing
Our sixth step in modeling calls for integrating any new model into our over-all transact model. (This predictive transact model for all human behavior (especially speech behavior) grew out of our earlier descriptive dimensional system. (Refs. 1, 2, 13, 14, 17)
This first step analyzes any observable and recorded human behavior-in-a situation into standardizing sets of variables called dimensional "factors" that must always be present in non-zero amounts if there is to be any behavior at all. As the most useful set of variables we take the six categories called the acts and wants (or things-liked) of people in time, space and other circumstances. The circumstances category always needs further subclassifying in particular studies. We often usefully subdivide it into symbolic, material and still further circumstances. Each variable is analyzed as a potential factor in a product (called the "transaction"), not as an addend in a sum. (Thus, applying suitably all the usual statistical formulas (which are mostly for linear or additive variables) to the logarithms of the raw data when they record some interaction may often prove more predictive.)
Transact modeling, secondly, analyzes each variable into its four "facets." These four facets are four aspects or basic specifications for every variable that has been or can be symbolized. For they specify the four reiteration cycles (see Ref. 17) which describe and explain generally how every symbol whatever arises, exists and combines with other symbols. These comprehensive hypotheses about the reiteration rule undergird and explain the genesis and nature of the four facets.
These four facets (and other compounds) of each variable can be viewed in many terms or versions or symbolic systems. In geometric terms, a variable if seen as a line has two internal facets of line sects and points and an external facet of some space around that line, as listed in Table 2. In system terms, a variable, if seen as a system, may have subsystems and parts and supersystem as its four facets. In general mathematical terms, a variable may be a set of elements, a sum of units, a product of properties, a power of the foregoing, or any of their inverses or combinations. In layman's language a variable may be a case or a, class, a zero or larger degree, a relation, or system of the foregoing. These versions (and some others) of the four facets of a variable are compared in Table 2 and should be studied and mastered for building transact models. For every variable one can name possesses these four facets or aspects and can be defined and specified by them. Table 2 glimpses the four facets in various versions but fuller understanding of them require studying textbooks of logic, mathematics and statistics or shortcuts as in Refs. 1, 2, 14, 17, 24, 29.
Table 3. Eight Versions Of The 4 Facets Of Every Variable
|Facet |Lay |Geometric |Arithmetic |Logical |Systemic |Corner Position Scripts, s |
|#0 |Cases |Points |Sets |Cases |Part |Presuperscript, SX |
|#I |Degrees |Line-sections |Sums |Subclasses |Subsystem |Presubscripts, SX |
|#II |Relations |Vectors |Products |Classes |System-in-hand |Postsubscripts, XS |
|#III |Systems |Spaces |Powers |Classifications |Supersystem |Postsuperscripts, XS |
The "corner script" (=s) standing for the four facets of every variable (=X) may then be spelled out as fully as needed in a notation standardizable throughout all the sciences as special cases of .
Our transact analysis, thirdly, analyzes the operators (called "functors"-- such as the familiar signs +, -, x, /, Xe,=, , etc.) Refs. 14, 17. These functors tell exactly how to combine and relate together both the symbols of the factors and facets and their empirical referents in the behavior of the observees and observers. Thus a sum of two acts like saying "either act this way or don't act" (=1A+0A = "1+0A") is an example of adding both the words (in a "union of sets") and their referents (in stating alternative acts); while the product of two acts like "Asking and Answering a question in a poll" (=A1 x A2 = "A12") is an example of multiplying both the words (in an "intersect of sets") and their referents (in stating a joint act or interact between poller and pollee.)
The essence of our transact analysis then is to analyze any recorded behavior-in-context into its "features" so fully and exactly that their synthesis restores that transaction at the current moment and predicts its recurrence in recurring context in the future. The "features" of a transaction here mean its factors, facets and functors both explicitly as symbols and as their behavioral referents, both as names and as things-named, both as the words with which we operate and as their meanings about which we operate. Thus a transact analysis of a likability situation involves observing and recording as fully as relevant for predicting (or for other specified purposes) at least all of its standard six factors, their four facets, and four chief functors combining them.
So far in describing our likability model we have analyzed only a speech part of the Acts factor--as polled responses of liking,(LKF) likening, (=LkK) and "making likely" (=LkD). ("Likelifying" for those who prefer a uniform system of names even at the cost of neologism.)
The skeletal descriptive formula, or algebraic shorthand, for the likes model is:
LKFKD LKF; LKK; LKD Eq. 2
and its predictive determination formula is:
+K2 R2 > .5 ? Eq. 3
Here R2 denotes the appropriate form of multiple determination (or correlation squared) of the predictor likes index, Lk, with the predictand behavior, B, and here K2 is the complementary index of non-determination. This questioned inequality states the hypothesis for testing that likes indices LkFKD, will correlate (i.e., R) with the behavior-to-be-predicted (=B) more highly than will another residual set X of predictor indices.. (Furthermore as le falls, the superiority of the likes index in predicting, measured by R2, will rise.) This hypothesis expects in general that the likes index will account for over half of the "variance of the predictand behavior, B*". (See.Table 5 for a confirmation of this expectation.) This likes subhypothesis then expects, as stimulus to research, that the remaining half of the variance of B will be found due to the things-liked and the setting (i.e. to the other factors in the fuller likability models) together with all errors.
II. The "Likables" Model
A. The Problem
In order, next, to increase the predictive power (=RLkB) of our likes model, let us enlarge its scope to include the major factor--the things-liked in the context of its core acts of the three likes. These things-liked appear as garbed or garbled under many names—as values, valued objects or valueds; as commodities or goods and services; as purposes, objectives, or goals; as wants, needs, or desiderata; etc. Most of such names may be standardized by the term "things-liked" (with V, for valued-object, as its dimensional symbol). This makes its meaning clearest as it implies the operational definition: a thing-liked (by people, P, in context, C) is whatever P pollees assert a measured degree of liking for. Then any combination of things-liked and the acts of liking, likening, or making likely (i.e., "likelifying") will be here called "a likable."
The problem in our likables models can now be stated as: How well do likes and things-liked predict an action of men?
B. The Observing
To gather relevant observations, note first that combinations of man's likes and things-liked involve much of the fields of psychology and physiology, economics and political science, sociology and anthropology, religion and philosophy, education and recreation, in short the social sciences and the humanities:
What the terms "likes" and "things-liked" connote in diverse products may be further suggested in the set of paired terms below.
Table 4: Products with Acts and Objects or "Likes" And "Things-Liked" As Factors
Likes Acts: Things-liked Likes Acts: Things-liked
Wantings Wanteds Give------------------------- Get (ratio)
Desirings --Desiderata Pursuing------------------- Goals
Valuings --Valued-objects Costs----------------------- Goods
Demand --Supply Payments----------------- Purchases
Effort. --Achievement Striving--------------------- Results
Means----------------Ends Sacrifice------------------- Reward
Stimuli --Responses Worth----------------------- Amount available
Quid------------------pro quo (of the object)
From such motley collections of terms the need for standardizing factors should be evident. This is what the transact model offers with its formal analysis of behavior into functors, facets and factors which can be operationally defined (as proposed in the rating scale indices for the likes above, for one example).
In observing the variables or factors in modeling likables, just how are they combined? Whatever these empirical combinings of behaviors and the mathematical functions representing them may be we have called them both together "functors" as noted above.
The most important case of these functors, or combined algebraic-behavioral operators, is to represent interaction of people, or group behavior, always as an algebraic product and never as a sum. This distinction spells the difference between chance prediction from r = 0 and near perfect prediction' from r 0 as in Exhibit C. For here combining the knowers (=p), of an item who interact with the non-knowers (=q) as their product (=pq) yields almost perfect correlation and maximal prediction of the observed increments (=A) in knowers, i.e.,
rpq.B = .997 Eq. 4a
Predictive efficiency of "Interaction as a product"
But combining the interactors inappropriately as a sum, p + q, must always yield a zero correlation of a constant, p+q = 1, with a variable, A,
R (p+q).B = 0 Eq. 4b
Predictive inefficiency of "Interaction as a sum"
In observing further the features of a likables model note that along with products three other classes of functors are useful, namely, appropriate forms of ratios, bits, and matrices.
A ratio, such as A/V, can compare acts of liking in units of things-liked. It answers the question: How much is V worth to you? or What will you give or do (= A) to get or keep what you want (= V)? Let us call this the “give-get” ratio or “tension ratio.” In economics, this give/get ratio is the demand/supply ratio determining the price (if all else is equal). In daily life, this ratio may often be seen as the Effort/Achievement ratio or Sacrifice/ Reward ratio which, when at unity, represents the threshold of Satisfaction.
Likables modeling will increasingly make use of bits since these are par excellence the units of decision or of choice between just two equal alternatives. The bit here is a unit of becoming not of being; it measures the change from two possibilities to one actuality, the answering of a yes-or-no question, for any content whatsoever. The number of bits, H, among N equal and independent things is their log to the base 2 (i.e., H = lg N or N = 2H). Thus the binary system in which computers operate is often the most suitable in expressing choices, preferences, likings and the likability structures built on them.
Matrices are usually needed in observing data for likables modeling to arrange sets of elements in rectangular arrays along n axes. Matrix algebra with its rules for operating in orderly and predictable ways with large sets of data, solving systems of equations, etc., seems to us an essential tool in studying likables models. Matrices will be more and more needed to cope with these likability models as their factors and facets and functors increase in number and in systemed complexity.
C. The Hypotheses
Let the likes hypothesis be enlarged now to include things-liked in a system of eight sub-hypotheses which may be christened the "likables" hypothesis, viz:
"A poll tends to predict a probable action of men in context (called a "transaction") insofar as:
1. it measures well
2. what those men see
3. as most liked,
4. most alike to it, and
5. most likely
6. all relative to some "thing-liked"
7. as fittingly recorded (in matrices of give/get ratios, etc.)
8. while nothing else changes."
Note that the proviso, if "it measures well, is a highly compounded set of methodological sub-hypotheses which imply that the factors, facets, and functors are explicitly observed, measured, correlated and systemed. (Ref 7, 8, 9, 10', 16)
D. The Testing
Out of the myriad, implicit and mostly unconscious testings of these likables hypotheses that go on daily among people everywhere, we report here (see Exhibit C) two relevant experiments on diffusing knowledge. (Ref. 19, 26, 27, 28) ('To specify what operations define "measuring well" we have developed a composite 1000-point index called "scient-scales." A report on this unfinished research is available as Ref. 24.)
The objective of Experiment was to spread news items (= the thing-liked, V) reliably from man to man through a set of P persons. The desired behavior, Bv, to be predicted and controlled was here "knowing the item" (well enough to retell it).
The chief causal preconditions for the "base-line" interacting were three--pairing off, steadily, and randomly. These were stated in the "If clause" of the,- the "logistic hypothesis" as: "If P people communicate items of knowledge, in pairs., steadily in t unit periods, with equal opportunity for each person and item, then the S-shaped logistic growth curve, defined by ∆ = kpq, will pre-diet the diffusing of those items (∆ = an increment of knowers, p; where q = l-p = non-knowers, and k = the "potency parameter" (held at unity, at first) is defined as "new hearers per teller and period."
This compound hypothesis can be paraphrased as a case of our eight likables hypotheses as follows:
"A poll tends to predict the probable acts of men in specified setting which are here their learning of item messages through t periods in a growth curve specified by the logistic equation, A = kpq, insofar as:
|(Specific sub-hypotheses here: ) |(Exemplifying the general sub-hypotheses of: ) |
|1. it measures well |1. Methodology or specified standardizing of observer behavior. |
|2. what each of the P men heard |2. Perceptions of observee interactors-- |
|3. and told to a new man |3. Pairing interactions |
|4. with liking enough to retell it |4. Liking of specified degree |
|5. with likeness to the first telling in wording and in speed enough |5. Likenesses to standard of specified degrees |
|to neglect unlikenesses | |
|6. and with likelihood that is the same for each man, item and period |6. Likelihoods in specified degrees |
|7. all relative to some "thing-liked" which is here the P news items |7. Things-liked in specified amounts |
|8. as in the give/get ratio: give a telling/get a hearing = means/ends|8. Tension ration |
|= quid/pro quo, etc. | |
|9. while nothing else relevant changes |9. Constant context otherwise |
The experimental procedure started with selecting a class room of 58 college freshmen boys so as to control 52 variables, i.e., so as to hold within normal ranges 52 physical, physiological, psychological, or social influences affecting communicating. Each boy was a starter of a different message item. All persons paired off at will in t successive minute-long periods, telling their partners all the items heard up to that unit-period of interacting. Their matrix of recordings of items was compiled from each interactor’s record sheet of partners and items for each period. This matrix specified the path of each of the P items which thus replicated the diffusing of one item P times. This P-fold replication greatly increased the reliability by reducing the variance due to sampling fluctuation from 13 per cent of the total variance to .3 per cent. One matrix array for each period showed its number of new hearers as A, the observed fact, to compare with the predictive expectation of the logistic curve (see Exhibit C).
The findings (Ref. 26, 19) clearly and reliably confirmed the whole set of the hypotheses of this base-line likables model. For the observed increments in diffusing, ∆, correlated with the expected increments almost perfectly at r = .997. Clearly the hypothesized preconditions had been fully met in this situation--as in several other like experiments from our Washington Public Opinion Laboratory. Any error or lack of fit in any one of the sub-hypotheses could reduce the nearly perfect prediction of the outcome from the "if clauses." Every sub-hypothesis (and its implied component sub-sub-hypotheses) must jointly be well fulfilled for the resulting intra-class correlation-of-fact and theory to be unity within three decimal places as it was here. This unusually excellent prediction of the later diffusing behavior from its immediate causal antecedents of communicating under base-line conditions of pairing off with like-chance and like-speed, is strong evidence for this case of the likables hypothesis. We confidently predict that checking tests by others, if done with equal rigor, will further confirm this base-line diffusion case of the likables model.
We furthermore expect that this logistic formula:
∆= kpq, ΣA2/P = A2 = σA2 if k = 1 EQ.6
for the variance of an interact will become increasingly tested and recognized as one of the most basic and exact lave of elementary social behavior (in communicating and acculturating) that the social scientists can point to.
In this base-line Experiment0, the three modes were held constant in order to establish a model for controlled experiments on human groups which contains a firm zero point. Departures from this zero point (specified by equalizing all opportunities to act) can then be measured according as each likables factor is varied in turn in specified degrees in isolation or in combination as desired. This permits transcending the usual quagmire of ill-known or ill-measured cultural and situational relativities which keep much of sociological research in the state of a "soft science."
Our next step, built upon the Logistic Experiment0 was "Experiment1 on Clique Size," (see Exhibit C) This simulated varying of the three modes together over two points in the simplest of the many social structures which Logistic Experiment, now opens to exact controlled experimentation. (Ref. 27, 28) Here communicating the item of news was limited for each person to a clique of persons of constant size P. Each person's clique was defined as the P1 persons he liked most out of the total diffusible population of persons. In life each person's clique could be observed, much as Moreno did, by asking him:
a. Whom he liked most to interact with in some way such as "telling news to," here, or
b. Who was most alike to his ideal listeners, or
c. Whom he was most likely to tell news to.
In the simulated experiment here each person's n clique members were drawn by lot from the P population with replacement, yielding P randomly overlapped cliques, each of size P1. This simulates the fact that in any large population every one has a subset of daily contacts however variable and vaguely bounded which we call "his clique." This structuring of human interaction into cliques is a necessary and universal form of organizing society whatever its cultural subforms.
This "cliques model" in logistic diffusing expects that limiting the retelling to one's own clique would slow down or decelerate the diffusing and so reduce the steepness of the logistic diffusion curve. (See Exhibit C) This diffusion curve's average slope should by the cliquing hypothesis be flattest for the smallest clique of two persons and steepest (coinciding with the simple logistic curve, Eq. 6) when clique size became maximal at P persons --"the clique of the whole."
This model for randomly overlapped cliques of size P varies the four likables factors as follows:
1. The thing-liked, the desideratum, V which is here the number of a person's potential hearers (i.e., his clique) varies from 1 up to P-1.
2. The liking, varies over two degrees (i.e. LkF, = 1 or 0) appearing either as "liking-well-enough-to-choose-as-a-clique-member" or as "unliked and so not a clique member."
3. The likeness similarly, being locked to the liking with perfect correlation, varies as an all-or-none alikeness vs. unlikeness (i.e. LkK, = 1 or 0) to whatever defines clique membership. k
4. The likelihood, similarly modeled here for simplicity at the start as perfectly correlated to the liking and likeness, varies as a two-point variab1e(i.e. LkD = 1, 0) from unitary to zero probability of any person, BP, being told the news item by person, AP, according as B either was or was not in A's clique.
Note that assuming perfect correlation among the three modes is a second structuring base-line of relationship among the model's dimensions. Then departures from this base-line of r = 1.0 (i.e., the distribution of lower correlation indices observed in life) can help to specify, measure and standardize the modal organization within any person or group with respect to some thing-liked or some behavior to be predicted. Thus our "correlating of modes" hypothesis expects that low correlations between feelings, knowings and doings may flag schizophrenic personalities or societies in respect to the predict-and behavior; while high correlations may indicate integrated or normal personalities and a society fulfilling its roles and living up to its norms in the measured respects.
In testing this specific cliques submodel (Ref. 27, 28) of the highly general likables model, we repeated a simulated logistic experiment m times, in a population of dice instead of people, for a series of clique sizes ranging over n = 2,3,4,5,6,P (i.e. m = 6 here). The m resulting curves are shown in Exhibit C. The diffusion curves did become steeper with increasing clique size exactly as the hypothesis predicted. As the cliques enlarged from 2 to P-1, from a dyad to the whole population, (Ref. 28) the growth of the news item accelerated and the curves steepened from flattest .up to the logistic curve at the upper limit.
This glean confirming of the clique-size model grew out of an earlier series of experiments (Ref. 27) which revealed a confounding factor. This factor, .which we named "seeking out non-knowers," accelerated diffusion. It exactly offset the decelerating from smaller and smaller cliques at cliques of four persons (n = 4). This observation of a balance point of two opposed forces led to measuring the social force named "seeking out non-knowers" as that force which equaled the accelerating effect of doubling the dyad clique and so might be called equal to a second bit 2H = 1g2X = 1g21 = 1 in this standardizing context (where the dyad (=2X = 21) gives the first bit so that together 1H + 2H = lg2 + lg2 = lg22 = = 21g2 = 2.1 = 2 bits. Thus the two series of experiments tend to verify the two principles or causal factors that if the thing-liked is "celerated diffusing" in pairs, then the tension ratio relating means to ends is statable in prose as: Enlarging cliques (= PK/T2) and seeking out non-knowers (= P2/T) are likely to accelerate diffusing (= Pw/T2 ). Restated algebraically in terms of the population and time dimensions, this tension equation is: [P1P2/T2= PK/T2]. In colloquial "likables" language this may be paraphrased as: The more people everyone likes and is likely to speak to (i.e., the cliques) who are unlike himself in respect to knowing the news item), the faster the "probable acting of men" that is called "spreading knowledge."
This controlled experiment simulating human interacting or elementary group behavior of communicating m simplest all-or-none forms as specified here seems to us to confirm crucially this case of the likables model. But more importantly it opens a vista whereby controlled social experiments under.-more.; and more complete and life-like conditions can help develop a harder behavioral science. We see scientific experimentation as highly promising for solving ever a little more fully the ancient problem: How may the probable acts of men be predicted better and in more situations.
E. The Applying
The potential applications of this likables model seem so numerous and pervasive that one illustration will be grossly inadequate. Yet we shall choose just one example because it seems likely to help men answer the ancient question: Whither mankind? by the scientific answer:
a) Towards whatever man likes most;
b) towards whatever seems most alike to what men prefer most; and
c) towards whatever seems to men most likely of achievement (within the limits of what they like most).
This answer might be condensed to: "Whither man likes or seems alike to it and likely.
For we see the future evolution of man as increasingly coming under his own control--though often still this seems a distant goal. The goals, the value systems, that men choose and research on and work for (the three modes again!) will progressively set the trend of man's development. But for this humane evolving, men need a more explicit and unified value system for each person and for all people than we see now among the world's diverse cultures, competing religions and clashing ideologies.
Towards building a unifying value system, we here offer (see Exhibit D) a beginning entitled (Ref. 11): "What is Liked Best? –a Scaling of “Likables”.
This 400-word statement of a general human value-system (or “scaling of likables”) was built to the size specifications stated in Exhibit D. It is undergoing controlled experimental testing and development as research funds become available. It is intended eventually to show how scientific methods can hold humanity to formulate and test a more unifying value-system of full philosophy of life. Though its language may seem naïve or childish at first blush to sophisticated philosophers, a little experience with polling on illiterate peoples’ value systems in many cultures will show the need for using the shortest and most familiar words in the language, The trite language of this statement, its style in rhymed verse for listing and inventory of 200 items, in easily remembered form, the choice of items and the technique of ranking them and other specifications are matters not to be superficially judged by the reader’s opinion. These are matters for controlled experiments to test against comparable alternative statements of value-systems by indices of relative reliability, relative predictivity of peoples’ later goal-pursuing behavior, relative asserted satisfyingness, relative time-use, etc. Deeper study of Exhibit D will reveal profounder implications for axilogy and a more powerful instrument here for developing a science of values than the casual reader may at first suspect.
The author is often asked: Have the procedures above been used in practice and with what results? We have found college classes most available for study and for experiments on scaling human value-systems. Other interested participants have been value-oriented organizations such as Unitarian Fellowships, discussion groups among church youth, and the conferences and meetings of the American Humanist Association (AHA). In our Project Consensus we have conducted “consensus-forming experiments” on national and state conference groups of the AHA, AAPOR (the American Association for Public Opinion Research) and AAUW (the American Association of University Women). Here we regularly have the exciting experience of announcing to the group: You people here as a group are likely to change your opinion or value-judgment of this issue (e.g., desegregation, population control, etc.) about 50% of the maximal possible change within pairs in the next half hour on one condition—namely, that each of you will talk with the person sitting next to you seeking a joint vote or decision on this issue. Initial and terminal polls measure each person’s “shift” of opinion, if any, during the hour’s sitting. Each A or B halves of the group discuss either issue A or B (and serves as control group for the other issues) with a partner ending in a joint vote or “forced agreement.” This intervening experience of seeking common ground in order to make a group decision for action that seems most nearly satisfying to all, operates to shift people’s opinions if at all, in the direction of their mean. This is but the age-old experience of cavemen to UN Assemblymen that, if N people want a group decision for action more than they want their own opinions adopted, then they will shift their opinions as publicly asserted toward a consensus.
It the twelve groups (N = 150, 60, 110, etc.) thus far put through this controlled experiment on forming consensus, the average reduction in the initial dispersion of their scaled opinions was 50%. These groups in half an hour shifted their value-assertions or opinions half of the maximum possible amount within pairs. They achieved 50% of perfect consensus with their partners when stimulated by discussion-with-intent-to-agree.
Of course, the durability of such shifts of asserted values, the correlation of polled vs. practiced opinion, and fifty other dimensional factors in this situation remain for M.A. and Ph.D. thesis exploration. These still unpublished experiments call for much checking while exploring their promise. Such inquiries can isolate and measure likes and likables, manipulate and test them, better and better with research, on opinions first and on overt behavior eventually.
These value-system scales and consensus experiments seem to promise techniques whereby the value sciences can progressively chart the field of valuing forces more and more fully. Insofar as men can learn hoe to predict and control their likes and likables, in just so far, it seems to us, men can with increasing efficiency become more self-directing and self-governing.
This statement of exactly 200 values or things-liked best in Exhibit D is intended to help each reader to decide explicitly on his own value, system. He can adapt it, item by item, either to himself or to any in-group, through the five steps below.
1. Read the statement (responsively for group discussions).
2. Rate each of the 200 things-liked by the 10-point Likings scale above. This begins to make it fit the rater. The rater, using a "Rater’s Abacus," distributes exactly 1000 points of score among the 200 items so every rater will average five points per item. This equates "opportunities" among all raters and among all items.
3. Revise it--adding, deleting, changing ideas or wordings, to make it state your own values better, but keep the total number of words at exactly 400. This requirement of constant wordage forces responsible weighing of relative values as in making a budget. If every change (=A) that the rater makes is preferred by him to what it displaces (=B), and if he, thus weighs every item, then the whole statement will come to state what he likes most in life.
4. Compute the raters "Individuality Score" as a per cent of the 200 items changed (1/2 a per cent point for each replacement and 1/4 point for each rewording in synonyms.) The complementary per cent, averaged in any group, measures roughly how fully the statement above represents a common value system or set of norms for that group and might be called a "Commonality Score." This score could tell from polls how far any set of persons, groups or nations profess a common value system and with what differences in the inventory or the weighting of their top 200 items.
5. Let people pair off, note differences in their ratings and items changed; and discuss them, requiring the pair to make a joint vote on each diverging item before some deadline.
6. Berate all items to see how much each rater changed his initial values due to the social pressures and discussions in Step 5.
7. Compute the mean per cent of change in a population's values system due to discussion and interaction, with all else highly constant. We hypothesize that as discussion increases, with many pairings, and more adequate discussion, the variance of the Individuality scores will decrease towards a consensus (at the mean rating in that set of people). This hypothesis implies the common sense notion that discussion with intent to agree promotes agreement. Its formal confirmation might facilitate the engineering of group agreement from family councils up to the UN Council.
This consensus experiment, though crude at first, yet can develop a technic for helping to form and measure people's value systems as explicitly, comprehensively) and sincerely as people may like enough to pay its social costs (i.e. its give/get ratios!). It further offers a technique for reforming any person's or group's value-systems as far as mutual persuadings can do so when all else is kept unchanged in tightly controlled experiments.
F. The Systematizing
Without going into technical detail, note that this likables model enlarges the transaction (called the "likables" transaction) to predict some later behavior (=B) using as predictors both the three likes acts and the things liked (=V). The-dimensional formula for the likables model explicitly develops the facets of the two factors--Acts (=A) and Values (=V) in give/get ratios (AV-1)--so as to try to maximize the correlation (=R) between the earlier predictor behavior, the tension (=B1) and that behavior as it later is observed to occur (=B2). Thus:
B1 AFKD V-1 Eq. 7 defining the likables predictor, B1 and
→ ? 1.0 Eq. 8
measuring the predicter-predictand correlation to the outcome. The size (end the reliability) of this correlation between the likables predicter and the outcome may be said to validate the model or verify its hypotheses within the limits of the situation studied.
III. The Full Likability Models
A. The Problem
In order finally to increase the predictive power (=RAB) of our likables model, let us enlarge its scope still further to include all of its relevant context or circumstances (as far as the observer's resources and current technology permits). Our problem then becomes: How to define and measure the most fully predictive transaction? What further factors beyond the Acts and Values (the three likes and things-liked), with what facets modifying them and functors combining them, seem likely to raise the predicting of B2, the outcome behavior at issue (such as diffusion of knowledge in the logistic transaction above).
B. The Observing
For predicting, the most complete observing of the context of the predictor action should start with the categories of variables that are always present in every human behavior and that appear as necessary factors in a product, and then go on to study further subcategories of variables that are only sometimes present and that appear as alternatives or addends in a sum. We therefore observe first the standard transactional context factors as follows:
1. Acts-in-the setting of the core acts—especially speech acts and all-- acts relating other variables together in systems relevant to B.- Thus in the logistic situation above, the core acts were simplified to the tellings and hearings and recordings of items between persons who were almost unaware of the contextual acts by the experimenter. These acts of the observer included previously selecting a homogeneous set of core actors, controlling irrelevant influences, instructing the diffusers, etc. This set up interacting equably (i.e., with equal opportunity) and steadily as specified by the hypotheses at issue. These acts in the context or setting, generally set up the governing influences, constraints, limits, directives, expectations, hopes and fears, and climate for the core acting. They may define and organize and purposefully guide the core acts. Many of the operative context acts were set up in the distant past such as our forefathers' act of setting up the U.S. Constitution with its expected human rights operating as normative factors in current desegregation situations;
2. Actors of the core acts (whether dependent predicted acts or independent predictor acts) and also, as far as necessary to predict B, the actors of the conditioning acts;
3. Timings of any and all acts as far as they appear relevantly in their many facets as dates, frequencies and probabilities, sequences and causal chains, durations or ages, speeds and celerations, trends, cycles and fluctuations, Change and evolution, the active vs. the static, verbs vs. nouns, adverbs vs. adjectives, etc., as systematized in our dimensional analysis (Refs. 1, 2, 5, 6, 13, 14) along with the rest of the features of human transactions.
4. Spacings of the foregoing in the points, lines, areas, and volumes that condition human and all other action, as far as these may correlate with B;
5. Values or things-liked that in the three tenses (as past memories, current likings and future purposes) may operate, consciously or not, on most if not all of the human behavior that may be involved in the prediction of B;
6. Material or relevant material equipment, energies, and resulting states, the earth, the sea and the air, in short whatever correlates of B are in any part non-human and weighable (i.e., expressible in C.G.S. units);
7. Symbols or all language, numbers, records or any forms of speech acts ranging from a grunt to a library, from music to mathematics, from idle talk to a University's curriculum--all only insofar as it may correlate observably with the behavior to be predicted;
8. Residual circumstances or whatever else, including combinations, that correlate with the predictand transaction, B, well and surely enough to raise the net correlation which tests the likability model in a given study.
C. The Hypotheses
With the above sketch of context factors in mind, let us finally synthesize our modeling in the "likability theory." Its ten systemed hypotheses expect:
"A poll tends to predict a probable transaction, B (i.e., behavior-in context) insofar as:
1. it measures well,
2. what those men see,
3. as most liked,
4. as most alike to that, and
5. as likely,
6. all relative to specified things-liked,
7. as fittingly recorded in matrices of give/get ratios, etc.,
8. and all in relevant setting of the eight factors called the antecedent Transaction A, and
9. insofar as the features of the Transactions A and B are well matched,
10. while nothing else changes."
This likability theory or system of hypotheses rounds out the more limited _likes and likables hypotheses by explicit inclusion of all the relevant context (sub-hypothesis #8) and by the matching technic predictively applied (sub-hypothesis #9). This matching means that an earlier transaction A is expected to predict a later transaction B only to the extent that their features are relevantly matched, factor for factor, facet for facet, and functor for functor. Transaction A is likely to predict Transaction B better and better according as they have more and more nearly the same factors (i.e., the same acts and actors, timings and spacings, materiel and symbols, values and residual circumstances) with the same facets (e.g., comparable means, comparable variances, comparable correlations, etc.) and combined by the same functors listing, in A, for listing in B; adding in A for adding in B; multiplying in A for multiplying in B; powering in A for powering in B; and so on for the inverse and compound functors).
Of course, this matching implies two-way inferences from A to B and also retroactively from B to A. Either A or B has like ability to estimate or predict the other--hence in part their naming as the "like-ability" model.
Fuller specifying for using this powerful "matched transacts" technic can be found in the author's two volumes systematically devoting seventeen hundred pages to such dimensional analysis (Refs. 1, 2). A danger in the matching technic lies in superficial and overly mechanical matching without deeper understanding of the provisos that the matching be "relevant" and "appropriate" and done by the full rules specified in those two volumes (and in a paper (Ref. 5) discussing some misapplications)
D. The Testing
Towards testing the predictivity of the likabilities model an empirical survey and a logical argument-will be reviewed here.
This poll compared the correlations from the nice questions in the likability model (in less developed form as shown in Exhibit B with correlations from 162 rivals (Ref. 21). The Transaction B to be predicted was the respondent's degree of support or opposition to a national voluntary membership organization dedicated to one side of a highly controversial issue in the U.S. Jewish community. The 4-point index of this supportive behavior, B, was the organization's records for the 1364 respondents, classifying them into "Paid-up members," "Lapsed members”, “non-members” possibly "Susceptible" for recruiting, and "Resigned ex-members."
The mailing went to a random sample of a universe defined by full lists of persons supplied by the organization as in the four categories above. These were taken to represent the four-point range from effectively supportive to non-supportive behavior.
The questionnaire measuring the likability questions was substantially the nine "mode-and-tense" questions shown here as Exhibit B. The prediction from these nine questions--each a 7-point Rating Scale--was to be compared with the predictive correlations from 162 other questions, oath a 4-point scale. These questions had been formulated from extensive interviews by a clinical, psychologist to tap a comprehensive range of psychological mechanisms of identification, protest, etc., which were hypothesized to be correlates, if not partially causes, of the differing degrees of supportive behavior.
The comparison of the nine likes ratings with the other 162 "Comparison" ratings was stacked against the former to make the test more convincing if the likes questions showed superior predictivity as the likes hypotheses expected. For the nine likes questions were single questions (see Exhibit a) while each of the Comparison sets were 9 scales composed of around ten questions each. These Comparison scales with their greater length and range should be significantly more reliable by the Spearman-Brown formula and more valid than single questions.
The findings are shown in Table 5:
Table 5. Criterial Correlations From Modal Vs. Non-Modal Ratings (Ref 21)
| Predictor Sets of Variables |Multiple R |
| 9 modes questions |0.78 |
| 9 randomly-selected non-modal scales (of 10 questions each) | 0.75 |
| 9 most predictive non-modal scales (of 10 questions each) |0.75 |
| 9 least predictive non-modal scales (of 10 questions each) |0.47 |
|17 Thurstone And Likert non-modal scales (of 10 questions each) |0.75 |
| | |
|Predictor Variables When Equalized for “Length” |Multiple R |
| 9 nodes questions | 0.79 |
|17 non-modes scales | 0.72 |
| | |
The nine single questions from the likes model out-correlated each of four sets of nine scales each comprising some ten questions. The nine likes questions showed better correlation against heavy odds whether compared with a randomly representative set of nine scales or a best correlating set of nine scales, or finally the whole set of 17 Thurston or Likert types of scales each comprising some ten questions. When equalized for length the superior predictivity of the 9 modes questions is even clearer, yielding a multiple correlation of .79 compared with .72 for the 17 non-modes, psychoanalytic types of scales.
Clearly and against odds the nine likes questions out-correlated a large field of 162 Comparison questions even then the later were optimally organize in scales with optimal regression weightings.
This poll was not strictly a testing of our likability model, which not been formulated at that time. But it does imply such testing in the following respects:
1. It involved phrasing largely in likes terms as called for by the likes model;
2. It involved rough statements of amounts of things-liked with give/get tensions roughly probed as called for by the likables model;
3. It involved 162 context variables in a questionnaire which was mailed back from the respondents' customary home situations in full "natural contexts";
4. It involved a high degree of matching to the criterion behavior (of paying dues to support the organization) as called for by the likability model.
Logically, we argue that the more alike transacts A and B are in each of n respects, the more their common elements must be in each of those n distributions, and hence the greater their appropriately built-up net correlation will tend to be in general. Thus, in trying to predict the transaction called "supporting organization X" we postulate that predictive correlations will rise according as:
1. The predictive acts match the acts in the outcome (e.g., the polled respondent's intention to pay his dues, Question i in Exhibit E., closely matched the independently observed records of the organization as to his being a paid up member;
2. The date of the predictor matching (i.e., being near the date of the outcome; here the organization's records preceded the poll by several months.)
3. The predictor population observed matches (i.e., overlaps with common membership) the predictand population;
4. The same four subsets of the population are observed in the poll as in the criterion;
5. Mean behaviors are observed in both predictor and predictand;
6. etc. (Ref's 7, 8, 14, 16, 20)
The principle that matching all features of predictor and the predictand improves their correlation is necessarily true at the limit where they become one and the same. At the limit the correlation is perfect by definition and trivial for practical predicting (though if the two observations are independent, their correlation usefully tests their reliability, of course). But the principle is basic to all social predicting which seeks a practical balance between:
a. higher prediction from predictors that are near in time and similar to the outcome versus
b. lower prediction from predictors usefully earlier in time and more available or more independently observable or causally related but often not as similar to the outcomes in (a).
In short, the prediction-by-matching hypothesis seems useful but always in need of better specifying to avoid superficial matching and include more comprehensive matching of all features and deeper functional and even cause-effect matching.
In fact we suspect, without rigorous experimental proof as yet, (Ref. 18, 22, 23), that the close matching of Transacts A and B here are mostly responsible for the higher correlation of the likes questions.
E. The Applying
The likabilities model is in such constant daily use everywhere in informal pieces that a formal list would far exceed the bounds of an article. We merely note that any aptitude tests for a job, or college admission, or marital prospects, etc., seek to match and measure whatever is most like the predicted behavior. Inquiries as to an applicant's experience and skills and interests try to match what he has been likely to do and has liked with what is most like his future role. Unconsciously do not most people when forecasting their own or other persons' probable acts in the future try in a common sense way to base the forecast on as closely matched past behavior in as closely matched contexts as they can?
F. The Systematizing
The likability dimensional formula can now be written as a special case of our full transact formula, (Ref. 13, 14)
The 8-factor transact formula is:
B A=B AFKD V-1; P; T; L; M; W; C] A=B Eq. 9
The transactive likability model where BA=B stands for the matched predictor and predicted transacts or behaviors-in-context (which is the chief feature of the likability model);
Stands for the four facets which specify by corner-scripts on each factor, its cases, degrees, relations and systeming in the situation;
AFKD stands for the 3 likes acts of Liking, Likening and Likelifying;
V-1 stands for the valued-objects or things-liked in the denominator of a give/get ratio (which is the chief feature of the Likables model)
; stands for any functor relating variables such as +, -, /, =, etc.
P, T, L, stand for categories of standardizing factors respectively called.
M, W, C, People, Timings, Spacings, Materials (especially Money!), Words, and residual Circumstances.
This means that the likability models can be exactly viewed, in both their symbolic and their referent behavioral features, as a subsystem within the larger system of human behavior. This subsystem of the acts of men is offered as a useful analyzing and resynthesizing technic for both the behavioral scientist interested in rigorizing laws and the humane psychologist interested in people and what they value most, such as catalogued in Exhibit for example.
Exhibits
Exhibit A: The Likes-Ratings Scales
[pic]
Exhibit B: Ratings of Organizations
[pic]
Exhibit Ca: Baseline Experiment
[pic]
Exhibit Cb: Simulated Experiment
[pic]
Exhibit D: Project Worth, Test 1 1963
I. What is liked Most:
Our Human Wants are here listed to help each person choose the best in life
0. Our Personal Present
We all, the people of our Earth,
want now a life of greater worth
for each in every place and time,
We seek ten ends as means to climb:
1. Hygenic
We want TO LIVE in health—
We want to live both well and long;
We want to grow both safe and strong;
We want less sick of any kind;
We want more whole in body and mind.
2. Economic
We want TO GET more wealth—
thru work and trade we freely choose,
ourselves to feed, clothe and amuse,
to fill all needs from high to low
as we from child to adult grow.
3. Political
We want TO RULE ourselves—
thru rulers chosen by us all
who let no rights nor freedoms fall;
with justice and security
in local or world community.
4. Recreational
We want TO ENJOY life—
each day, in work, or play, or rest
on peak or plain, with fullness and zest,
with social gain at cost that’s due,
with joy for us and others too.
5. Scientific
We want TO LEARN the truth—
to learn how things—as a scientist tries—
do move, or breathe, or symbolize;
to learn to make men free from war
fearless, joyful, and aspiring more.
6. Domestic
We want TO LOVE and be loved—
as wives or husbands, parents, children,
as kin or friends or fellowmen,
as living, dead, or yet to be,
each in due ways and due degree.
7. Philanthropic
We want TO GIVE wisely—
to help men raise themselves: the healthless,
the loveless, lawless, joyless, wealthless,
and those who lack the uttermost
of what their own groups value most.
8. Religious
We want TO WORSHIP sincerely—
our God, our good, or goals in life.
We want to worship, with no strife,
what as most holy each conceives,
and freely tell what each believes.
9. Artistic
We want TO BEAUTIFY—
ourselves, our homes, our native strands—
thru music, pictures, garden lands,
thru what we touch or taste or smell,
so lovelier feelings may upswell.
10. Educational
We want TO TEACH wisdom—
to children, here and yet unborn,
and all who want from us to learn,
the better ways to live and grow
that we in humbleness now know.
11. Our Social Future
If we, the people of our Earth
would now no longer live in dearth,
then each a greater part must play
to help all win a better way.
II. Specifications of this "Likables Scaling”
This statement hypothesizes 200 human values, or items of "things-wanted," for the reader to rate. This value-system is intended to be highly:
1. Comprehensive--sampling somewhat typically all ten institutions of any culture--each stanza sampling one stratum or all human culture;
2. Universal --satisfying men in most cultures, religions, ideologies, statuses, and periods;
3. Exact -- specifying, in just 400 words, exactly 200 measurable items of preferring behavior, distributed with just 4 bits of decision or 16 (=24) va1ued items (between apostrophes) per stanza which deals with 1 of the ten major social institutions;
4. Understandable--writing simply with words-per-syllable ratio = 400/440 =.90 and so likely to be grasped by over 90% of English-speaking persons;.
5. Important--organizing 200 highly preferred items into a hierarchic system; '
6. Standardizing -- so polls can measure each person’s deviant value-system in terms of bits, H, of decision or choice:
n
H = -Σ pi lg PL
1
This statement for humanist value systems seems usable for:
7. Personal Choices--Sept. 1: Rating each item; and step 2: Revising any item (while keeping wordage constant) can convert this statement into the readers Own Value-system.
8. Group Consensus – reformulating values convergingly by discussion and joint choice in controlled experiments can promote agreement on a Common Value-System.
9. Human Unity—re-systematizing value-system gradually by extended experiments can help develop a World Value-system.
III. Trial Scorings of Likables
To help compare averages from whole value systems (and. Changes in them); different persons or groups who have differing personal histories and cultures, the following scoring schemes may be tested, improved, and used.
a. An “Unlike-Iess” or "Percent Individuality Score" might be the number of items marked off apostrophes which a valuer changes in meaning (each item counting .5%) or changes in phrasing (each such synonym change = .25%) while keeping total wordage constant at 400 words.
b. "Like-Items” or "Percent Commonality Score" is the complement, i.e., 100 Unlike-Items score. This measures people's agreement on items when each is liked more than the alternatives considered for it, thus specifying a common inventory of values or preferred items.
c. A "Like-Ratings" score is the correlation, such as Robinson agreement index, between one's ratings of the 200 items and group's ratings tikes as norms or standards.
d. A "Fervent-hating" can be decided on by the Valuer distributing 1,000 tokens on a "Token Abacus" among the 200 items, rating how relatively important or preferred, within any specified subset of items, he sees each item to be, on a scale of 1 to 10,
e. A "4-Bits-Rating" or ranking is decided on by the Valuer splitting the 16 items of each stanza in turn into a "More important half" and a "Less important half," then re-splitting each half into quarters, then into eighths, and finally into sixteenths which are individual items.
f. Other scorings should be devised, tested, and standardized.
IV. Hypothesis on Likables
If this 400-word statement of 2C10 human valuer or "things-liked most is compared with any like list of just 200 valued items expressed in 400 words by such appropriately constructed and applied indices as their:
a. relative reliability coefficients when re-observed in people's behavior, including speech behavior
b. relative validity coefficients against criterion variables of what people most choose and pursue in life
c. relative satisfaction coefficients
d. relative use of time pursuing each valued item;
Then this "likables-list" will tend to show superior indices of comparison by experiment. The claims in these hypotheses should be tested, not by argument, but by experiment.
Products Predict Interaction Where Sums Do Not
by
Stuart C. Dodd
and
Stefan C. Christopher
Institute for Sociological Research
University of Washington
Seattle
Sums or Products: Plurals or Groups?
The whole idea here put forth is that multiplicative thinking is a far more useful tool in the analysis of group behavior than is additive thinking; on the other hand, additive thinking is a more useful tool in looking at plurals than is multiplicative thinking. By plural is meant any plural number of persons or category of people to which we respond as a whole, in contradistinction to the single units of which the whole is composed.1 By group is meant any plural within which observable interaction of stimulus and response among the members is present.1 Any namable set of people is a plural; it becomes a group to the extent that its members interact or behave in ways that influence each other. Groups have dynamic relations, among the members and social relations are a form of product, not a sum, as will be shown.
Quantitatively inclined sociologists will not find much that is surprising or instructive in this mathematically trivial thesis. But many sociologists studying groups are not yet fully aware of the appropriate use of products instead of the corresponding sums in a set of indices or variables. Much of what this paper says is well known to statisticians, but it is presented here in everyday speech to make the ideas known to a wider audience.
Just what is a product? Think first of a product of two all-or-none kinds of things which logicians call a "logical product" or "intersect" of sets in the new mathematics. We symbolize a set by the zero exponent2, X0, meaning any set of elements (i.e., class with members, property with instances of it, or kind of symbolized things). In Figure 1A let two circles point to two sets (Which we may call X0 and Y0) and let dots mean the elements or members or instances of each set. Then the overlap of the two circles as in Figure 1A, shows what has been variously called their common membership, or joint result, or logical product, or Boolean intersect, or qualitative product (=X0Y0). If X0 in Figure 1A is the set of married folk and Y0 is the set of women, then X0Y0, their product (or class in common, or intersect) is the set called "wives." If X0 is the set of marriage relations and Y0 is the set of sonship relations of a man to a mother or father, then their product or joint happening is the set of the son-in-law relations of a man to his wife's mother or father. If X0 denotes the act of a man's giving his seed and Y0 denotes a woman's reception of it in an ovum, then the product in time, X0Y0 (along with some further generative factors) denotes a baby. If X0 means an act of telling a message and if Y0 means the react of hearing it, then X0Y0, their interact or social product, stands for a simple form of the group behavior called "communicating," namely, a-telling-and-a-hearing a social behavior. If X0 means broadly the raw materials factor and if Y0 means all the handling of them as a second factor, then X0Y0 means their economic product.
Figure 1
[pic]
[pic]
Note also how every adjective-and-noun, or adverb-and-verb or modifier-and-modified make verbal or logical products. (Many further factors are also needed to make some of these products, of course, such as nourishing the unborn child in the making of a baby.)
Next think of how these all-or-none joint happenings of two or more kinds of things became measurable products whenever the factors became amounts of some kind of thing (as by the ordering or counting of the things in a set)(8:12). If X0 stands for the cardinal class of twos (shown in the second part of Figure 1B as dots-in-pairs) and if Y0 stands for the cardinal class of threes (shown as dots-in-threes), then their intersect or logical product and also their arithmetic product is th cardinal class of sixes (shown as the rectangle of dots-in-sixes). This intersect is often called simply "two times three make six." Geometrically, Figure 1B still, if X1 is two meters and Y1 is three meters, then their product or cross-classifying in space is six square meters., This diagram 'suggests how the notion of an intersect, or common elements in two sets, extends the familiar notion of a quantitative product, of 2 quantities or variables to the qualitative product of 2 classes (or 2 properties, or 2 attributes, or 2 concepts, or 2 systems, or any other pair of symbols).
Since the thesis argued here is not only that products tell about groups better than sums, but also that sums tell about plurals better than products, it will be well for us to pause a while to make sure the reader knows what is here meant by the various kinds of sums. .A sum in arithmetic means a count or number of units. In set theory the sum of two sets, X0 + Y0 (called a union"), denotes two alternatives as in saying "People are either men (X0) or women (Y0). This statement in folk speech can be restated as: A set of people (=Z0) is the sum, or "union," of the set of men plus the set of women, i.e., Z0 = X0 + Y0. Some sociologists still talk primitively about social "patterns" and "organizations" as "Wholes that are more than the sum of their parts" without realizing that mathematicians have devised far better tools than sums for dealing with combinations of social things (which may form intricate systems). Logical and mathematical products (and other functions) such as dimensional systems(8:2,3,7) can show just wherein the two-way interworking of parts makes the whole more than their mere sum. In short, for interactive phenomena, such as the mutually influencing behavior in human groups and organizations, products are more appropriate than sums(8:8,9,10,11).
As partial proof, let us tell of three tests and outline a mathematical proof for one of them. Then let us go on to see the conditions under which one should add or should multiply in dealing with ambiguous sociological phenomena.
II. Some Experimental Evidence
A. Interaction of People thru Space - the case of intercity telephoning
Among Cavanaugh's twenty-seven tests(8:1) finding much support for the thesis here proposed, one set of data on telephoning between towns was chosen to make an experimental test to decide between the relative effectiveness of using a sum vs. using a product for prediction. If products do make better models of group behavior than do sums, then one should find much greater fit between expected and observed indices of human interaction whenever the model is a product rather than a sum of its variables. In the phoning between towns test, Cavanaugh studied the populations of a set of cities in all possible pairs, PA,PB, and the distances, LAB, between the members of each pair. In this clear-cut case of interaction the observed number of intercity phone calls was expected (by the suns vs. products hypothesis) to fit better the products model PA • PB • 1/L, than the sums-model, PA + PB - L, for exactly the same data.
To test this, the interaction matrix or sociomatrix (8:4, 5, 6) was computed. This cross-classified the twenty-seven chief U.S. cities (for which complete data were available) against each other. Three matrices (designated X1, X2, Y) were compared. Matrix Y recorded the (n2 -n)/2 observed numbers of intercity phone calls as supplied by the AT&T. Matrix X1 was the "product model" or hypothesis, which computed each of the 351 different intercity number of expected calls by the unweighted interactance formula, X1 = PA • Pt • 1/L. The third matrix, X2, was the "sums model" computing the same variables as in X1 but putting them, together additively in each cell as X2 = PA + Pt - L. The Pearsonian Correlation of the sums model with the observed data was .58 (=) correlation of the product model with just the data jumped to .93 (=). A gain of thirty-five points of correlation – and that gain near the ceiling of correlation – was observed merely by appropriately treating the same data of a group's behavior as products rather than as sums.
It is here forecast that this will be found to be so most of the time. Whenever sociologists put together interactional indices of stimulation and response between set of persons the right way by multiplication instead of by addition (as has been customary in our usual linear statistics), they may hope for markedly better foretelling of social or group behavior. The correlation of the prediction by the model with the observed outcomes or later facts are likely to rise from middle ranges around .5 to upper ranges nearer to .9.
A sums model (X2) yielding correlations of .58 accounts for only about 34 percent of the variance of the data (since r2 + k2 = 1 Where •14 is the coefficient of alienation so that .34 + .66 = 1). But a products model (X1) yielding correlations around .93 (as here) accounts for 86 percent of the variance of the data (i.e., r2 + k2 = .932 + .372 = .86 + .14). This dramatic decrease of the errors of estimate in the regression equation from 66 percent of error to 14 per cent of error simply by operating more appropriately upon interactional indices is the kind of progress towards an exact and predictive science which scientists seek. It represents the increased predictivity that was foreshadowed to result from our dimensional analyses of social behavior(8:2, 3, 10)
B. Interaction of People thru Time - the case of logistic message diffusion.
A second set of data supporting this "sums vs. products" hypothesis (for putting together the indices from plurals vs. groups so as to predict human group behavior better) is the logistic model. This can be stated both, as a mathematical theorem and as a controlled experiment that can be replicated and checked by any teacher in a classroom hour(8:9,14-18) Again one can forecast that dimensional analyses will yield many more such well supported hypotheses building a growing body of sociological laws. For this logistic model is only a special case of the general transact for social action which, in turn, is a special case of the completely inclusive and generalized dimensional formulas(8:2, 3) for any symbolizable knowledge in the field of the social sciences.
Since this logistic model has been much tested and published in several sources(8:8' 9, 14,18), only its decisive bearing on the sums vs. products subhypothesis need be pointed out here. A class of sixty students each started a message and traded all the messages which each had heard with one partner chosen at will in two-minutes time-spans following upon each other. Within eight time-spans all sixty messages were heard and recorded by each of the sixty students, showing full diffusion of every message. The number of item-communications in the group grew from 60 to 1,770 in the predicted S-shaped logistic growth curve. This S-shaped curve can be plotted on the blackboard beforehand so as to picture the expected growth of the knowing according to the hypothesis of the logistic growth curve. The increments in observed diffusion in an early testing correlated with the logistic model's expected increments at an average correlation coefficient of .90 in five separate replications of this experiment. The discrepancies were not statistically significant by the chi square test at the one percent level.
In this experiment there were three ill-controlled variables which made the actual conditions fall short of the "equal opportunities" for everyone to interact which was called for by the model. The confounded variables were
1. sex differences,
2. friendship, and
3. fluctuations from the small sampling of only six starters.
A later test controlled these three variables by replicating the experiment ten times among a set of strangers of one sex who chose partners at will in each time-span with spatial proximity well randomized(8:9,14) (78 freshman girls on the first day of classes at the University of Washington). Taking the mean of the ten replications gave more stable increments in the observed S-shaped growth curve and reduced sampling error. As predicted, this high degree of control of the three conditions called for by the model (i.e., pairing off with equal speed and equal chance for everyone) yielded an even higher fit of the model to the data. The exacting intraclass correlation between the expected and the observed increments in knowers of the message rose from .90 to .996+.
This seems a high point in experimental sociology. An hypothesis, written in an equation foretelling period by period a growth curve of one-way action between people under conditions approximating random multiplex), steady, pairing, was well confirmed in a new setting with a new population and a new message never before so tested. It was a controlled experiment of a fully rationalized stochastic model, not just another observing of an empirical model.
This excellent fit of the S-shaped curve to interaction data, when observed under conditions of steady pairing with equal opportunity for all to spread the item, will happen again (within sampling limits) whenever sociologists look for it again. For it is a mathematical necessity. The S-shaped curve is defined as the cumulation of most probable increments (= pq) under the three specified conditions in the diffusion or total, proportion (= p) of persons knowing a message at time t.
Thus:
pt+1 = pt + ptqt the logistic model
The increment (pq) in any time-span is by the law of joint probability the product (pq) of the part of persons knowing (= p) and the other part not knowing (r= q) the item told and heard. The S-shaped curve is then a product of two indices (p and q) by which the diffusion behavior is to be foretold. So all data gathered under the conditions of random,3 steady, pairing will always have to correlate with the logistic model as close to unity (r = 1.0) as sampling error will allow, (Lower correlations, of course, may be foretold insofar as the three conditions are only approximated as when people interact at will, as in the experiment above, instead of choosing partners by full "multiplex" causation.)
Now if one puts together the same predicter variables (p, q) by addition, he will get the sum, p + q = 1, which is always 1, since q is the complement of p. Since a constant has no correlation with a variable, this summative model will always correlate not at all with the data.
Thus either the worst possible prediction or the best possible prediction – a correlation of zero or of unity – will always tend to result (to the extent that the three logistic preconditions exist fully and solely) depending only on whether the selfsame data are combined by adding or by multiplying – as a sum (p + q) or as a product (pq) - as separate acts, by a plural or joint interacts in pair-groups.
This sort of mistake in assuming p + q for the model is unlikely to be made consciously, but often occurs concealed in computing a linear (i.e., additive) sort of statistic such as in combining variables by a multiple regression equation. So to test for such a concealed mistake, we computed the telephone data in a multiple correlation of the three predicter variables (the two populations and their distance apart) against the observed outcome. Even with these optimal weights this rose only to .65.
Likewise, all models which build on the mathematical law of joint probability with its assumption of equal opportunity for all the elements will tend to hold in any social setting only insofar as their preconditions operate fully and solely.
In testing the fit of any model to data one is really testing how far its conditions are there, or have been set up, fully and solely, in the setting tested. Among these conditions a weighty one is whether the behavior at issue is either the individualistic behavior of a plural or the interactive behavior of a set of pair-groups and therefore whether their indices should be put together either as a sum or as a product. (Cf. Chap. V-VII in 8:2 and Chap. 4-8 in 8:3)
C. interaction in a Group vs. Action in a Plural
A third bit of published proof(8:8) from Project Revere underpins the "sums vs. products" hypothesis from tests of message diffusion in open communities. This a posteriori test called for four separate correlations of roughly given size to be observed between model and data. This experimental setting was the dropping of leaflets by planes on people who reacted either individually in a plural by picking up leaflets, or socially in pair-groups by telling-and-hearing the leaflet message. An exponential curve was expected to fit "physical diffusion" data in a plural with a correlation as close to unity as experimental errors would allow – a correlation of .87 was observed. That exponential curve was expected to fit partially interactive or group data, less well with a correlation "below .5" and the actual result was r = .33. An S-shaped logistic curve was expected to fit those "social diffusion" data from a set of pair-groups with a correlation "above .5" – r = .78 was observed. That logistic model was expected not to fit the data in the plural as shown by a correlation "near zero" – r = .01 was observed. Although not highly reliable due to few cases, this testing shows that crucial tests which can be done over and checked, can be made of sociological hypotheses in open communities. This testing took in two double hypotheses – that a summative model for a plural should fit data from a plural and not fit data from a group; and that a multiplicative model for a group should, fit data from a group and not fit data from a plural. Such rigor of doubled testing, and high confirmation found in this case, again evidences to us that better modeling (for which our dimensional analysis offers one route(8:2,3)), and better experimentation on it, can more and more build up sociology as a hard science.
D. Other Testings
Further evidence is growing. Thus almost any testing of a dimensional formula or its subcase of a transact model – of which the testings above are examples – has to do with a test of a product as the most predictive way of putting the variables of social behavior together. This is so because a (two-level)4 dimensional formula is taken to be by the authors for sociology, and by Bridgman5 for physics, as "a product of powers" of the categories of variables, called "dimensions", which are basic categories to a field of science. Such a product singles out a (two-level) dimensional formula both in sociology and in physics and points out a unitary setting whether societal or physical. Similarly throughout the formulas of chemistry a product of atomic symbols point to a chemical union, a molecule; whereas a sum of atomic symbols means a physical mixture with no new chemical properties (which only came with a product). Similarly throughout speech a logical product of words makes up almost every sentence and communicates some single pattern of meaning which is the joint meaning of all the words of the sentence. A logical sum of those words, on the other hand, would be a mere list, a whole with no new pattern of meaning as in saying almost any sentence with "or" (which means addition in folk speech) put in between every word. In daily speech we talk of producer good as "products," not "sums"; and of their economic causes as "factors in production," not as "addends in summation." In fact, throughout the semiotic sciences and so throughout human thinking in symbols, products are found so vastly more important than sums (which merely flag the plural number) that for convenience we usually omit any multiplication sign and assume the product relation by simple juxtaposition between words in sentences, between quantities in mathematics between the symbols for chemical elements between physical quantities in their dimensional formulas, between statistical variables, and between sociological variables.
III. Remedial Criteria
If one grants that appropriate use of sums vs. products in handling sociological data can greatly improve scientific predicting, the next question is what does "appropriate" mean? Just when should one add and when should he multiply the variables? Just what are the social conditions calling for computing sums or products? What operational rules will tell what the word "appropriate" should mean and so tell the social researcher beforehand whether to add, to multiply, or otherwise to put together the symbols of his variables? What behavioral relations among men and things do the adding or multiplying of the symbols stand for?
As answers to these questions, a priori rules are needed – not a posteriori rules. These rules should be fitted beforehand to new settings – not tests that fit only after gathering and treating the data. Thus to say that rectilinearly related variables call for addition while curvilinearly related variables may call for multiplication needs first plotting the variables or running a test for curvilinearity. So also to say, that a product formula fitted similar data in the past better thin a summative formula is to utter a purely actuarial rule. Such actuarial rules cannot guide research as well as rational rules should in new settings which may seem to be unlike old ones. A rational and a priori rule should state the empirically observable conditions under which either a sum or a product in new settings will give the most predictive formula.
A. The "Cross-Classifying" Test of Interaction
Two a priori and rational rules for knowing when to add or when to multiply (in relating sociological variables to each other in a theory) are the "cross-classifying of people test" or "the population matrix test" and the "vanishing result" test.
The "cross-classifying test" of stimulus and response interaction of people is carried out by asking: Do these two variables interact as shown by cross-classifying them? Do they change each other or do they act alone with no effect on each other? Does A and B simply change their quantity (as in changing a singular into a plural number of same noun) or does it change the quality, the nature of either factor, somehow? Does their combination have new properties or only a list of properties of each alone? Thus, if one adds the plurals of a ;million whites and a million Negroes, a mixed population results, half of whom are white and half of whom Negroes. But let the whites and Negroes interact (whether in conversation or in work or in marriage) and a partially new result emerges which we call "a group," a social product, a result of interaction.
This interaction or group behavior is clearly shown in an interaction matrix or sociomatrix which cross-classifies actors and reactors with the acts which interrelate them as the cell entries. If this cross-classifying is needed to show pair relation between people within the data in hand, then those data stand for a group; they show interaction; they must be treated by multiplying in some form (such as adding their logarithms appropriately) to yield a product. But if cross-classifying is not needful or if subclassifying the two variables separately in two lone arrays is enough, then the data stand for a plural; they show two pluralic actions; they should be treated by addition in some form to yield a sum.
This device of cross-classifying not only helps to apply the interacting test but it also enlarges the concept of products from the well-known one of arithmetic products of numbers to take in logical products of classes (or "intersects" as they are called in the theory of sets) and on to take in highly complex products of systems of diverse sorts. It else helps to unify the concept of "products" as used in daily speech (as when we say that "culture is all the products of interhuman learning") with the concept of product in various logical and mathematical forms. Thus by cross-classifying, as made clear below, one can see some of the unity that is called "a product" among such seemingly different kinds of products as the following thirteen:
| |Examples of Products: |Its Name: |Corresponding Sums: |
|1 |2 X 3 = 6 |(an arithmetic product of two factors) |2 + 3 = 5 |
|2 |XY = k, the simple hyperbolic model |(an algebraic product) |X + Y = k1 = a line |
|3 |Σ z1z2/N = r12 = a correlation |(a statistical product-moment also a |Σ (z1 + z2)/N = a mean |
| | |scalar product of two vectors) | |
|4 |(p + q)N = the binomial (and normal curve when n|(an algebraic self-product or nth power) |(p + q) |
| |is large) | | |
|5 |(pq)t the logistic model |(a stochastic or chain product of t |p1qt |
| | |factors in statistics) | |
|6 |"woman" ∩ "married" = "wife" |(a logical product of two classes) |Women or married ( a nonsense phrase) |
|7 |The marriage tie and the sonship tie = the |(a logical product of two relations) |The marriage or sonship ties |
| |son-in-law tie | | |
|8 |A set of actors cross-classified with a set of |(a set of pair-products, a Cartesian |Actors or reactors |
| |reactors = a sociomatrix |product of two sets) | |
|9 |An act of telling an item and a react of hearing|(a social behavior product) |A telling or a hearing |
| |it make together the interact of communication | | |
|10 |an automobile is the |(an economic product) |an automobile as a mere sum of its parts |
| |joint result of all the factors of production | | |
|11 |a joint act of a man and a woman makes a baby |(a biological behavioral product) |two alternative auto-erotic acts do not |
| | | |produce |
| | | |a new organism |
|12 |the executive departments of a corporation or. |(an organizational product) |either Dept. A or Dept. B |
| |government | | |
|13 |the component hypotheses in an integrated theory|(a systemed product) |a listing of the hypotheses as alternates|
| |(8:19) | |– which they are not! |
The more compounded products in language, in logic, and in mathematics can be looked at as outgrowths of the above technique diagrammed in Figure 1 of cross-classifying joint happening. As the kind and number of factors and sequence of operatings on them change, one gets the products called products of sets, of vectors, of matrices, of lattices, and many others. Thus our "Likability Models"(8:19) provide a full predictive theory of human behavior that is
1. testable by polling; and
2. built as a complex product or joint occurrence of what its ten factor hypotheses specify.
One of the most important products for sociologists seems likely to be the qualitative-quantitative product. This means the clear statement of the joint occurrence of kinds of things with amounts, of each in any patterning. With the help of the zero exponent(8:2, 3, 20) our interdisciplinary and standardizing "corner-script" notation, any quantity (=X1) can be "qualified" by multiplying it by any quality(=X0). Thus the product, X0X1, means both a quantified quality and a qualified quantity. Thus any model can include and express clearly in its formula the conditions under which it holds by just writing in the qualifying factors, X , X , etc. , which
mean those conditions.6
This zero exponent notation(8:20, 2, 3) when coupled with the other three corner scripts(8:2,3) in "S-notation" can become, it is claimed a powerful tool for sociologists and scientists generally. With it they can classify and specify more precisely and fully than has been possible hitherto the combined qualitative and quantitative aspects of human behavior. The charge that quantifying leaves out of account the humanly important qualitative aspects need no longer be true. Any quality the human mind can know or feel, any value man can perceive or conceive or imagine, any aspiration or ideal he can speak of or otherwise symbolize can be recorded as an X0 with suitable scripts. Then it can be treated by harmonized logical and algebraic rules to aid man powerfully in adapting scientifically to all that is qualifiable and quantifiable and relatable in the universe.
The zero exponent denoting a set offers flexible and powerful symbolizing for extensional thinking that has so revolutionized semantics, the sciences, the "new math," and all computers with their dependence on itemizing knowledge into card holes punched or not punched or into magnetized points on tape.
In short, we hereby go on record as predicting that: As sociologists use the zero exponent notation appropriately they gain in ability to, describe and predict exactly the qualitative aspects of as much as use of the zero digit in the past millennium has advance our ability to deal exactly with all quantitative aspects of human knowledge.
With the above notation in mind, let us consider another type of product which can give a sharper tool for sociological thought than has been available in the past. This is the self-product or dimensional power of any symbol. The power is shown by the exponent which tells how many times its base (representing a category of variables such as Acts, People, Time, Space, etc.) has been taken as a factor or cross-classified against itself.
A major form of this cross-classifying can be a "III-matrix." This Ill matrix has three axes making a solid much as the geometric cross-classifying of three feet of length against two feet of breadth against four feet of height yield a spatial product, or "three dimensional volume" of twenty-four cubic feet.
In sociology a use of the III-matrix would be as a "role-matrix" for arraying all the basic data from which any and all summarizing indices about human organizations could be computed (8:7) This role matrix is a sociomatrix showing one kind of interaction between actors (represented in the rows) and reactors (represented in the columns) for every role-actor or any specialized interactor (represented in the sagittal arrays). If there are PI actors, PII, reactors, and PIII role-actors, then this III-matrix would have PIII ordinary socio-matrices (i.e., II-matrices). Its triple numerical product P • P • P would be the total number of cells. More compounded products can then describe compounded relations among these data. Each cell would show the amount of some act in triple relation to an actor, a reactor, and a role-actor. The cells record indices of acts in the context of the actors, reactors and role-actors that are named in the array headings.
Thus every act and its consequences in any human organization could all be systematically arrayed in such matrices using as many axes as needed(e.g. 8:13). All organized human behavior consists of specialized and correlated interests among people which maybe called "role-acts." These include all behaving in some amount of conformity to whatever one sees to be the behavior expected of him by the fellows of his in-groups. Thus, all cultural behavior can in principle be systematically arrayed in n-matrices of sufficient size and number with whatever interlocking and hierarchical relations that may be needed. Then indices about the various aspects of any organization will be sections or summaries from some larger organization matrix. Thus in the hygienic sub-cultures of rural Syria(8:21) our n frequency distributions of the-villagers' acts relevant to their health and mortality rates specified an n-matrix of n axes. Then specified selections and computations from these matrix cells defined "patterns of behavior" and measured the "net social forces" modernizing and reorganizing those total communities.
Such organization n-matrices are in essence highly compounded cross-classifications, or products, of n sets. Such n-matrices can spell out the internal structuring of any complex factors in still more complex products.7
B. The "Vanishing Result" rest
The second criterion for products vs. sums is the "vanishing result" test. This says simply: "If the result or combination of variable (i.e., becomes null or zero) whenever one of the variables vanishes, then the combination is a product, and not a sum. For throughout mathematics and logic any product, XY, vanishes if either X or Y vanishes. But any sum, X + Y, does not vanish if only one of its addends vanishes. Therefore, to apply this test one thinks: If any one of these variables becomes non-existent, or absolutely absent in this situation, will the result cease to exist? If either the act of telling a message or the react of hearing it, is non-existent, then there is no interest of communication – so this communication is a product (i.e., a tell-and-hear act), not a sum: (i.e., a tell-or-hear). Again, for example, one may ask himself: If you or I do not die this year, will the death rate of our city become zero? The answer being "no" indicates that deaths should be added up to get the plural’s death rate, not multiplied together.
Again, for a most general example, our transact model for any behavior-in-context is formulated as a product of six basic dimensional factors – namely the acts of actors, for wants or things-valued in time and space, and under residual circumstances. For every transact vanishes whenever there is absolutely no act nor actor, no time nor space for it to happen in, no values (i.e., desiderata or wants), and no other circumstances whatever before, during, or after the act.
Any apparent omission of basic factors from sub-models of the general transact model simply means that these factors are implicitly present as a parameter or constant for that situation which does not affect the fluctuation of the variables under observation (just as the coefficient, 1, is customarily omitted from algebraic formulas).
IV: Summary
The product (which is much more than the sum) of what has been said in this paper can be restated more briefly, as follows:
Sociologists, studying groups and organizations as the heart of sociology, need sharper tools to analyze, synthesize and predict what men do and will do under specified conditions. One new and sharp tool, is the most effective way of putting together the words, numbers and other symbols which stand for the human behavior that is studied in sociology.
The lone doings of men in a plural (as in demographic data) should have their symbols added to make a sum; the joint or mutually influencing acts of men in a group should have their symbols multiplied together to make products. These products may be of whatever complexity is needed to improve the scientists' descriptions and explanations, such as to improve man's predicting and controlling of his group and organized behaviors. Several controlled experiments showed dramatic improvements in the predictions from models using these tools.
Two rules which make either the sums or else the product way of combining concepts, variables, and indices the more predictive, can be called the "cross-classifying test" and the "vanishing result test." These state: If whenever one variable vanishes their result vanishes, then those variables form a product.
If cross-classifying is necessary to show interactions within pairs of people, then multiply to get the indices of interaction; but if cross-classifying as in a matrix is unnecessary to specify every person's behavior, then add the indices of pluralic action.
Both criteria show that the acts of plurals should be treated as suns, while the interacts of people in groups should be treated as products, and the role-acts in organizations should be treated as dimensional powers, i.e., as products of multiple factors of one kind.
Footnotes
1. Adapted from Lundberg, G.A., Foundations Macmillan, 1939, pp. 340-1.
2. This notational usage conveniently unifies the handling of qualitative and quantitative products by laymen. It also helps to unify the rules for operating in Boolean and ordinary algebras for mathematicians. Thus, with the zero exponent notation, products in both algebras obey the rule to add exponents, etc.
3. Note that randomness in human affairs maybe well approached in large communities by a population that is either homogeneous with respect to the diffusing at issue or highly heterogeneous. For randomness and normal distribution will result from causation by many small independent influences – a condition of "multiplex influences" that a large city may approximate well.
This suggests that laboratory conditions in this logistic controlled experiment do simulate real community conditions. The more numerous and diverse the influences playing in each person to react to a given item message and the more multiplied in many people, the more a large community will tend to approximate the condition of randomness or equal opportunity to tell and hear that item.
4. The two levels are (1) products of (2) powers of basic variables. These are a part of our larger four-level dimensional system of sets of sums of products of power,(8:2, 3, 10)
5. Bridgman, P.W., Dimensional Analysis, Yale University Press, 1931, pp. 16, 22 ff.
6. Thus we may specify the linear recursive logistic model in complete form as (pt+1 = ptqt) X0 Y0 Z0 by letting X0, Y0, Z0 denote respectively the three qualifying preconditions or "random, steady, pairing." This equation then states the "logistic law" which will always be true or hold regardless of whether the things whose parts are named "p" and "q" are men, mice, or molecules. The logistic law is a mathematical principle that is surer, in a sense, than the law of gravity in that it does not depend on empirical events but only on syntactic ties among symbols. It must always be observed wherever its conditions are observed. It is as sure as 2 + 2 = 4, which will hold as long as the definitions of "2," "4,,"+," and "=" are not changed. It is thus a purely formal or syntactic law that goes beyond the things which may embody the symbols in any setting looked at. Sociologists can thus formally get a law-abiding science. The practical problem is to find models whose conditions are embodied in human behavior.
7. The core dimensional "III matrix" for any human group or organization behavior that is studied can be built up as follows for scientific purposes of improving prediction and control:
Let P denote a person, one element in a set of people, a cell of a population matrix, a " matrix "
Let PI denote a plural, a list or set of people, one array or axis of a matrix, a "I-matrix
Let PII denote a group, a set of interactors, a socio-matrix, a "II-matrix"
Let PIII denote an organization, a set of role-actors an organization matrix or "III-matrix"
Let XX denote the set of all other dimensions in any transaction or situation studied.
8. Cited bibliography of author and students.
8:1 Cavanaugh, Joseph A., "Formation, Analysis, and Testing of the interactance Hypotheses," American Sociological Review, Vol. XV, o. 6, December, 1950.
8:2 Dodd, S.C., Dimension of Society, Macmillan, 1943, 944 pp.
8:3 Dodd, S.C., Systematic Social Science, American University Social Science Series, No. 6, 1947, 985 pp., University Store, Seattle, Agents).
8:4 Dodd, S.C., "The Interrelation Matrix" Sociometry, Vol. III, 1940.
8:5 Dodd, S.C., "Analyses of the Interrelation Matrix by Its Surface and Structure," Sociometry, Vol. III, No. 2, 1940.
8:6 Dodd, S.C., "The Interactance Hypothesis: A gravity model fitting physical masses and human groups," American Sociological, Vol. XV, No. 2, April, 1950.
8:7 Dodd, S.C., "Sociomatrices and Levels of Interaction – For with plurals, groups, and organizations," Sociometry, XIV, Nos. 2 and 3, May-August, 1951.
8:8 Dodd, S.C., "Can the Social Scientist Serve Two Masters – An answer through experimental sociology," Proceedings of the Pacific Sociological Society, Research Studies of the State College of Washington, Vol. XXI, No. 3, September, 1953.
8:9 Dodd, S.C., "Diffusion Is Predictable," American Sociological Review, August, 1955.
8:10 Dodd, S.C., "The Transact Model – A predictive and testable theory of social action," Sociometry, December, 1955.
8:11. Dodd, S.C., "The Counteractance Model, American Journal of Social Psychology, Vol. LXIII, No. 3, November, 1957.
8:12. Dodd, S.C., "The Reiteration Rule – A cyclic system for syntax, neurograms, and all laws," Synthese, Vol. XI, No. 1, March, 1959.
8:13 Dodd, S.C., "How Random Interacting Organizes a Population – Exploring a simple chance model to relate diffusion theory to information theory," Synthese, Vol. XII, No. 1, March, 1960.
8:14 Dodd, S.C., "The Logistic Law of Interaction When People Pair Off 'At Will:" with Garabedian, P. G., Journal of Social Psychology, 1961, No. 53.
8:15 Dodd, S.C., "The Logistic Law. in Communication," with McCurtain, Marilyn, The Symposia Studies No. 8 of the National Institute of Social and Behavioral Science. Series Research in Social Psychology, Washington, D.C., Sept.., 1961, pp. 1-9.
8:16 Dodd, S.C., "Clique Size As a Factor in Message Diffusion," with Garabedian, Peter G., Sociological Inquiry, Vol. XXXII, No. 1, Winter 1962, pp. 71-81.
8:17 Dodd, S.C., "How Momental Laws Can be Developed in Sociology – By Deducting testable and predictive 'Actance' models from transacts," Synthese, Vol. XIV, No. 4, Dec., 1962.
8:18 Dodd, S.C., "Logistic Diffusion of Information Through Randomly Overlapped Cliques," with McCurtain, Marilyn, Operations Research Quarterly, Vol. 16, No. 1, 1965.
8:19 Dodd, S.C., "Likability Models for Predicting Probable Acts of Men – A theory of values," Systematics, Spring, 1965.
8:20 Dodd, S.C., "Something Out of Nothing," Main Currents Modern Though, Vol. 8, No. 4, Feb., 1952.
8:21 Dodd, S.C., Syria, American University of Beirut, 1934, pp. 336.
Appendix
Among the further criteria for deciding how best to combine the symbols for sociological variables, four practical and largely a posteriori techniques can be noted here. Statistical textbooks and the author's Dimensions of Society (8:2) give a fuller systematic treatment of these questions.
1. Note the overall test of science as to good methodology: Those methods or behaviors of the researcher are most scientific which yield the highest predictivity indices, i.e., which help most to predict recurrence of the behavior, B, at issue under recurrent antecedent conditions, A.
Scientific methods should help produce the "If A, then B" sorts of statements that constitute the core of any science. Will adding or multiplying one's data appropriately improve their predictivity more?
2. A second suggestion is to apply appropriately the usual additive and linear techniques of statistics (including correlation, factor analysis, etc.) to the ha of the variables. This in effect treats those variables as combining multiplicatively. This "log variables" technique seems to us highly promising for sociologists to explore in group and organizational studies whenever the data seem excellent enough to justify such refinements of analysis.
3. A third note is to interpret any correlation in multiple languages (such as the 24 versions of r in Chap. VI in 8:2). A squared correlation when interpreted as a scalar product of two vectors
(rxy = σxσy Cos Q)
or as a percentage of common elements
(rxy = nxy /√nx • ny
measures the percentage degree to which the two variables are related as a blend of a product and a sum. As r2 rises towards unity, the common elements or product (measured by r2) rises towards becoming the whole relation; while as r2 falls towards zero, the non-common elements or disjoint sum (measured by k2 = 1 - r2) rises towards becoming the entire relation between x and y. Thus, correlation indices measure a posteriori the mixture of additive and multiplicative combining of underlying sets of the relevant elements.
4. A fourth suggestive note is to distinguish when cross-classifying between interrelation matrices (specifying human groups and organizations) versus 'correlation matrices or "scattergrams" (specifying cross-product moments of any two variables). Interrelation matrices cross-classify people against people PI x PI = 101 with behavioral and other, indices (I) in the cells specifying the interrelations within that population. Correlation matrices in social data cross-classify any index, ix, against any other index with frequencies of people in the cells specifying the correlation within that 2-variable situation. The dimensional formulas clearly distinguish these as
|Interrelations = |[pic] |
| | |
|Correlations = | |
Then the four levels of these behavioral interrelations or interactions of people or populational products, can help systematize sociology neatly into situations involving no more than:
|A Person |[P] |a matrix cell |
|A Plural |[PI ] |a 1-array matrix |
|A Group |[PII ] |a 2-array "sociomatrix" |
|An Organization |[PIII ] |a 3-array "role matrix" |
Of course, these four "pure" levels will yield a host of mixtures (or sums) and blends (or products) with all the variants sociologists observe.
The Transact Model: A Predictable and Testable Theory of Social Action, Interaction and Role-Action
Sociometry, Vol. 18, No. 4, Sociometry and the Science of Man (Nov., 1955), pp. 432-447
[Editor's Note: There was no Table 1 included with the published article. Equation 2 would have been in Table 1.]
Preview of the Model-An Operational Definition
This paper outlines our general transact model in six dimensions as an operationally defined and testable theory of social action. It is intended to go beyond the current description of social behavior by developing the predicting of it.1 "Predicting" here broadly includes:
1. forecasting in time,
2. estimating from samples to parent populations, and
3. inferring from independent to dependent variables.
This means, in general, inducing and deducing from a known part to an unknown whole along as many dimensions as possible. Our theory specifies empirical and rational operations, as spelled out in the five sections below, to improve the predicting of social action.
This model for any transact (which is our name for an act-in-context when observed in six dimensions) will be developed below, under the five steps in scientific modeling, these specifying testable propositions about:
1. its variables-seeking the most general set of categories for predicting repetitive human behavior;
2. its pre-conditions-seeking the most general hypotheses of relationship among those variables;
3. its formulas-deducing solely from those variables and pre-conditions the model proper for predicting any social action taken as criterion;
4. its experiments-seeking techniques to isolate the model and the criterion with all other variables controlled;
5. its fits-seeking statistical indices measuring the predictivity of the model under controlled conditions.
As a foretaste of this development, our transact model can be simply previewed as a typical operational definition in the following sentence: Any social behavior is best predicted by an earlier transact in six dimensions which means telling, in reproducible terms, six kinds of facts, each in three tenses, namely: Who did, does, and will do what, when and where, why and how, about that criterion behavior.2
II. Categories for Social Action-the Variables
A. Six "Factors"
In building comprehensive theories the problem of what set of fundamental categories to adopt has concerned the systematizing philosophers and scientists alike. Aristotle had ten categories. Kant had four (quality, quantity, relation, and modality) which subdivided into twelve. Physicists can convert all their formulas into five dimensions (space, time, mass, temperature, and electric charge). Sociologists have proposed simple sets such as LePlay's "folk, place, work," or Eubank's "societary composition, causation, change, and culture." Von Wiese, Sorokin, Parsons, have elaborated larger and less definitely bounded sets of categories. Authors of textbooks in chapter and section headings use sets of organizing concepts that specify subfields such as: "demography," "ecology," "interaction," "institutions," "rural communities," "change," "social problems," "disorganization," "culture," "collective behavior," etc. Seldom are these classifications analyzed by the canons of classification so as to be a minimum non-overlapping set whose subclasses and combinations can subsume all other concepts in sociology. Longer lists of concepts have been offered by Eubank and by a Committee of the American Sociological Society seeking to standardize the first course in sociology (Ref. 3).
In Dimensions of Society (Refs. 4, 5) four categories, called "sectors," were taken as people (P) in time (T) and space (L, a length), with indicators (I) of all residual characteristics of the people or of their environment. These four top level classes of variables, especially the indicators, were then sub-classified as finely or to as many levels as needed. Two sub- indicators of desiring (D) and of desiderata (i.e., objects of value, V) were also distinguished and treated at length because of their importance in the social sciences. The sub-classes and combinations of these six categories (T, I, L, P, D, V) were shown in descriptive formulas to redefine some five hundred of the most-used concepts in sociology. Later in Revere Studies on Interaction (Ref. 17) the "desiring" category was broadened to "behavior" or acts (A), since all behavior can be viewed as indicating some degree of desire (whether conscious or not) for whatever that behavior satisfies. This set of six categories we call "factors" of a transaction both to avoid confusion with the four "sectors" and to denote that they form a logical product rather than a mere sum or set of categories. Among these six categories we have come to put the "acts" in the central position since we study social action and take the other five as context of the act. A transaction or set of transacts here is an action-in-contest, a behavior observed as part of a whole situation. Our meaning of Dewey's term "transaction" is more exactly defined by specifying the six standard factors of the context. These six necessarily present factors specify what is taken as the "whole situation" of which any act at issue is a part. This means that the content of our transact theory, the variables studied, are the probable behavior and the values of people in a context of time, space, and other circumstances. This may be paraphrased in many ways in order to transcend any one phrasing. Thus we also say that a transaction datum is an explicit record of an action analyzable into a set of all-or-none acts (including the degrees of an action) and distributed by actors, by their wants or values, by time, by places, and by any other circumstances. The logical product (i.e., joint occurrence) of these six factors of actions, people, time, space, values, and all the residual circumstances ( = C) is our definition of a transaction.3 It is an action taken jointly with its context of actors and of temporal, spatial, motivational, and all other observed circumstances. In interrogative phrasing a transact answers the questions: Who? does what? when? and where? why? and how? The "unit transact" is any all-or-none act of one actor, at one time and place, for one value or want, and at one point of each other relevant circumstance. The unit-transact can be enlarged along any of its six dimensions. Thus a "larger" transaction might be "50 men working 100 days mending a mile of road for X dollars of total wages during the winter in wartime." A still larger and less definite transaction might be: "All the activities of the inhabitants of town X during one year under whatever motivating and other circumstances existed there." A precise and predictable transaction might be our experiment on logistic message diffusion. Here the transact specified as "telling a news item as people meet in pairs at a steady rate and with equal opportunity" fully predicts a logistic S-shaped growth curve of news diffusion (as expanded below). A transaction is thus some behavior so comprehensively observed as to include all aspects of its attendant situation insofar as these aspects help to predict the later criterion behavior. The testable proposition summarizing this substantive part of our transact model might be stated as: Any behavior of a large population will be better predicted in general from these six categories of variables than from single ones or subsets of these six. Each category is expected in most situations to improve the predicting in some way or in some degree.
B. Four "Facets" of Each Factor
Each factor has four "facets"' or aspects. These are:
1. its power-level specified by an algebraic exponent (such as the powers of a population, namely: P0 denoting a person, P1 a plural of actors, P2 a group of pair interactors, P3 an organization of role-actors (Ref. 7));
2. its classes or many varieties (such as men, women, children, in a population; feeling, knowing, doing, modes of speech behavior; positive, neutral, and negative values; etc.);
3. its degrees or amounts (whether in all-or-none, ordinal, or cardinal terms);
4. its cases or elemental points (such as a person, an act, an instance of an average amount of a desideratum, a date, a spot, a score, etc.).
A fuller notion of these 24 facets of the six factors may be had from studying Table 1 which lists some three sample alternatives for each facet of each factor of any transact. For further detail of how these four facets were derived, how they are specified by scripts at the four corners of a base letter in our interdisciplinary and standardizing "S-notation," and how they work to give the six factors enormous flexibility and coverage of social situations, the reader is referred to the two systematic volumes, Dimensions of Society and Systematic Social Science. Table 1 suggests the vast number of situations that can be described or symbolized by the orderly combinations of just six factors and their four facets each with three or four of their many possible subdivisions. Actually this number of combinations of items in Table I works out to millions of millions. This would seem more than are needed to symbolize the social situations of society for some time to come. A summarizing and testable proposition for this facets part of the transact model might be: Earlier transacts will predict later transacts, or criterion behaviors, in general, best when their facets have been adequately specified and appropriately used.
C. Translation of Categories between Systems
In choosing basic categories to systematize a science one seeks as few or simple a set as will describe the whole field. For this, one seeks categories which can subsume or translate most other concepts from those in daily use up to those of rival systems of the upper range.4
Just how can categories be translated? The answer is: By finding their meaning(s) in common (which are their logical product or "intersect" as Boolean algebra calls the members-in-common of two classes). Let an example from the field of communication theory illustrate. In communications a standard frame of reference is to ask: Who says what to whom, how, and with what effect? Let this sentence be rearranged as: Who with whom (actor and reactor) says what with what effect (acts and reacts) how (residual circumstances, mostly means). This indicates that the words in this statement coincide in meaning with our factors of people acting in circumstances when people are cross-classified as actors and reactors (developing their second "power-level"-see Table 1, Row 1, Col. 1, line 3, "a group," p2).
More sociological translations5 of categories may be found in the author's two volumes (Refs. 4, 5). These translate five hundred sociological concepts into transact terms in some five hundred definitional equations. These descriptive formulas only lack the specifying of pre-conditions to become calculative and predictive formulas. Such specifying of pre-conditions is the next step in modeling and is developed in the next section.
III. Pre-Conditions Connecting the Factors-The Hypotheses
To convert descriptive and classificative formulas into calculative and predictive formulas called models, requires specifying the pre-conditions or hypothesized relations which connect the variables to each other. (Out of all the interconnecting conditions that may exist those that are known and specified in the model we call the "pre-conditions.")
A. Probability Pre-conditions
For this requirement in modeling we propose using probability conditions as a general base line. Wherever more special conditions connecting the factors are not known, transact models will use probabilistic pre-conditions as a base line. Then all the many overlaying variables which produce systematic or non-random deviations from that base line can be better isolated, measured, and progressively built into fuller models.
Let the simplest possible transact illustrate a bit of what a base line of probabilistic conditions mean. Consider any human behavior observed as an all-or-none act, A ( = 1, 0), among P people at one time, place, motivation, and residual set of circumstances. Suppose the act is "hearing a news item" which is universal to all cultures and easily measured by counting the knowers of the item. Taking the act as all-or-none means that we assume or create a dichotomous distribution as the relation of the acts to the people. From this all-or-none condition one can deduce seven "laws" of elementary probability by computing the statistical moments or average powers of the index of the act.6 These probability principles are known as "simple probability," "complementary probability," "alternative probability," "nul probability," "joint probability," "difference probability," and "the inverse of the joint probability." Which of these seven probability principles holds depends not at all on the acts of these people but solely on the researcher's response to them. If he observes and counts the knowers, he thereby computes the simple probability, p, of being a knower; if he observes the non- knowers, he then computes the complementary probability, q. If he observes "either," he computes the alternative probability, p + q (= 1); if he observes "neither," he gets the nul probability, 0; if he observes "both," (as when knowers and non-knowers meet during a period of time in conversational pairs and some pairs contain both) he gets a joint probability, pq. Thus the researcher's conditions, or choice of what to observe, determines what probability formula describes and will "predict" the result of his observing. The transact model specified as "the statistical moment of any order of an all-or-none act" will measure the outcome of his observations in general and will measure the outcome more particularly depending on whether he observes "one,' "the other," "either," "neither,' "both,' "one but not the other," or "all not both." Next let this partial transact involving only the dimensions of acts and people be enlarged to include the time factor. Let it illustrate how to predict the growth of an act in a population during a period under conditions of elementary probability. The relation of the acts to time is here assumed to be a steady one for the simplest case at first. (This means a rectangular distribution of the acts over the successive unit periods.) From this condition of a steady speed of acting, the formulas, mathematically deduced (as by integrating the probabilities over time), enable one to predict that the growth of the act or diffusing of news items will tend to follow:
1. a normal probability distribution of news items known after t days (if a random half of the people hear one new item each day); or
2. an exponential growth curve of knowers (if one and the same item is heard with equal opportunity for all each day); or
3. a logistic S-shaped growth curve of knowers (if one item spreads with equal opportunity from person to person as through conversational pairs).
Which of these three growth curves or mathematical "laws" is followed depends essentially upon whether the acts are observed as alternative probabilities among persons-predicting a normal probability curve; or as a simple probability in a plural-predicting an exponential curve; or as a joint probability in a group (of dyads)-predicting a logistic curve.
All three growth curves for diffusing here depend alike on the two conditions of random and steady distribution of the acts in the population and period. If the steadiness condition changes, an altered model results; if the randomness condition changes, the model must change to match.
But just what is meant by the word "randomness?" The layman thinks of a random distribution as one given "by chance," "by lot," "by pitching pennies." Philosophers have long searched for the bases of "probability"- is it an empirically observable frequency or an intuitive degree of belief? Statisticians think of probability as a relative frequency of one kind of event, the proportion of times it occurred in some past series of possible occurrences. The mathematical statistician refines this as the limiting proportion in infinitely many samples. Building on the frequency concept of probability, sampling theory, actuarial practice and a host of highly successful and predictive applications of probability theory have grown up in the physical, biological, and social sciences.7
Our transact theory proposes to weave probability into the warp and woof of social theory and not use it merely as an afterthought when testing the significance level of some finding. We use a probability, which is but the mean all-or-none behavior in a large statistical population, as used throughout the natural sciences. We apply probability principles to the mass prediction of social action because from past experience probabilities mean by definition what is most likely to happen in the future under like conditions. We expect in life situations that elementary probabilities will be greatly compounded and structured, that they will be often unstable and hard to unravel. But we further expect as a major working hypothesis in trying to systematize sociology that by operating with probabilities in modeling we can get on with building an experimentally based and predictive science of interhuman behavior.
Note the three tenses of the operational definition of the word "random." As the past tense of any instance of it, randomness may have been generated as a controlled artifact by drawing lots or by a table of random numbers. Alternatively, it can arise naturally from many small uncorrelated influences. In any situation in proportion as we control, or strip off the variation of, the few, large correlated and visible influences, we get nearer to a homogeneous or purely random "base-line" situation. Thus consider our logistic interaction experiment in a classroom described below which assumes just three conditions, namely:
1. random
2. steady
3. pairing.
Here the experimenter tries to hold constant:
1. the kind and amount of the acts (held constant by doing nothing but telling and hearing news items throughout the experiment),
2. the set of actors (held constant as the same students),
3. the time, and
4. the space (held constant in one class period and room),
5. the motivation (held constant at "sufficient willingness to interact as instructed"), and
6. all the visible residual influences on interacting such as
a. sex differences (held constant by using one sex only),
b. friendships (held constant by using all strangers),
c. subtly sensed congeniality perceived from dress, expressions, or conduct (held constant perhaps by darkening the room).
Insofar as the experimenter controls these factors (which are experimentally known to correlate with the interaction), then whatever further residual influences determine who meets whom when choosing partners "at will" may be expected to be highly random. They seem due to the residual many, small, diverse, and as yet undetectable influences in one's nervous system whose aggregate effect in a set of persons we call chance.
As the operational definition of randomness in the present tense, there are all the statistician's parametric and non-parametric tests for identifying degrees of probability in diverse statistical structuring. These sampling indices, together with appropriate descriptive indices of correlation and the like, can tell us the probable contribution of each factor (at its given facets within the data in hand) in predicting the criterion behavior. These indices can further tell us the share of the residual unknown or random elements challenging to further research.
The operational definition of randomness in its future tense as its uses or adaptations is only beginning to be realized by sociologists. For randomness can mean social equality of opportunity in specified respects. When the poller selects a randomized sample, he has given every person in that universe a mathematically equal opportunity in respect to being drawn. Randomness is thus close to the heart of a democracy insofar as that word of many meanings means equality of opportunity in specified respects for its citizens in some of its political, or economic, or other institutions. Randomness can thus be viewed as the absence of biases, causes, restraints, or anything making for inequality or visible "lumps" of causation in specified respects in the population at issue. Thus if the p knowers of a news item interact randomly with the q non-knowers, the law of joint probability expects that their product, pq, will represent the proportion of knower-non-knower meetings where diffusing of news can go on. So the expected increment in a unit period is pq under the condition of randomness or equal opportunity to interact. This may be paraphrased in respect to the interacting at issue as "a homogeneous population" or "an extremely and atomistically heterogeneous population" or "thorough mixing" or "meetings with equal opportunity" or "democratic equality" or "most probable interacting" or "random pairing," or "chance meetings" or a "situation without biases, restraints, or any structuring," or an "equalitarian condition." "Randomness" can be viewed as an operationally specifiable word for that part of the ideal of democracy, socialism, communism, Christianity, Islam, justice, fraternity, or any concept or creed insofar as it stands for equality of people before the law or "in the eyes of God," or in other specified respects.
A summarizing and testable proposition of a substantive sort for this probabilistic pre-condition part of the transact model might be: Insofar as the measurable correlates of the criterion are held constant, the residual variation of the criterion will tend to be predicted best by the laws of chance.
B. The Product Pre-Condition
In addition to probabilistic pre-conditions, a second essential precondition in our transact model of social action is that the basic factors are combined by multiplication, not addition. A transact is a product of factors, not a sum of addends. The acts, actors, values, time, space, and residual circumstances are factors combining in some form of logical or mathematical product. Depending on whether they are observed qualitatively or quantitatively, their product may be a logical product asserting joint existence (whereas a logical sum asserts alternative existence) or an algebraic product including product-moments or correlations, set products, scalar products of vectors, matrix products, etc. The justification for assuming a transact to be a product and not a sum of its six basic categories is the "vanishing result" test. A product vanishes if any factor vanishes (i.e., becomes zero), whereas a sum does not. Every transact must vanish if there is no act, or no actor, no finite time or place, no causation, and no material or symbolic context of any sort. Each category may have many alternative forms which are logical sums, but since each category must exist in some form in any transact (defined as an act-in-context), the category must be a factor in a product. This definition of a transact as a product results in transacts being highly unitary situations with interworking parts that modify each other and often create new emergent wholes. Mere sums or aggregates of the same parts do not interwork and so form a looser unity like a pile of stones. Transacts (products) correspond somewhat to molecules in chemistry since molecules are always formularized as products of atoms (whereas chemists write a sum to denote a mere physical mixture, not a chemical union). Furthermore, combining the basic variables in transacts in the appropriate form of product can be shown experimentally to improve predictive correlations with the criterion behavior dramatically.8 This expectation is a testable hypothesis summarizing the "product pre-condition" of the transact model.
IV. Formulas Deduced from Categories and Pre-Conditions
The third step in our modeling deduces a calculative and predictive formula by taking the general descriptive formula and qualifying or restricting it to the case of the specific pre-conditions that are assumed in any model at issue. The most general descriptive formula in our system9 says that any recorded social situation whatever is symbolizable by some set of variables, their four facets (specified in the corner scripts) and any operators. Since this formula, by definition, can describe any recorded set of data whatever, it does not predict any specific situation until some pre-conditions are specified and these limit Eq. 4 to more specific values for the range of its symbols. Thus suppose the simple pre-conditions sketched above for studying an all-or-none act are chosen. Let us choose to study, as an example, the probable spreading of an all-or-none act such as a news item diffusing from person to person in a specified population and period. Let us choose
1. pairing off
2. randomly and
3. steadily as three base line pre-conditions connecting these three factors of an act, actors, and time.
Then the formula as derived from these pre-conditions will be the simplest logistic curve (Refs. 7, 10, 12, 15). This illustrative formula cumulates the most probable increments of new hearers, pq, in successive unit periods. It is one example among many of a formula predicting the course of some social behavior (here the conversational spreading of gossip, rumor, or news) under specified pre-conditions (here of random, steady, pairing). Note that these pre-conditions are irrespective of personalities, the situation, or the culture. They are universally statable pre-conditions, applicable to men, mice, or molecules alike. The model can apply in any science. Note further that "pre-conditions" in this paper do not mean further circumstancing variables (the C in Eq. 1 and 2). "Pre-conditions" mean relations such as forms of distribution between variables. Thus "pairing" means a dyadic relation of population elements; while "random" and "steady" denote rectangular distributions of the act in the population or in time.
Before discussing the experimental testing of this illustrative logistic formula, let us note the vast range of possibilities opened up to the experimental social scientist by converting descriptive formulas such as in our Dimensions of Society into predictive models. Conceivably any repetitive social phenomena can be modeled, given sufficient ingenuity (whether or not we can control things well enough to test them). And all science is limited to the repetitive, to classes of phenomena, leaving to history and literature the study of the unique. Already our moment-powers model has been generalized to a "powers" model which begins to specify the pre-conditions under which human behavior may be expected to follow (or can be predicted by) linear, logarithmic, hyperbolic, parabolic, and other common curves, as well as the normal, exponential and logistic curves sketched above for diffusing an act among persons, plurals, or groups.10
V. Experiments Testing the Fit of the Formula-The Tests
To test probabilistic models by controlled experiments on people one devises situations which hold every known factor, except the factor-at-issue, constant at some one amount. This tries to achieve the goal of leaving only the one independent factor-at-issue to vary along with the residual many, small, uncorrelated and currently undetectable influences operating. Then these become increasingly predictable by pure probability considerations - and testable similarly by probability theory.
Thus in studying social diffusion to get at some laws or predictive generalizations of culture growth, the Washington Public Opinion Laboratory has experimented extensively on the simple logistic model of diffusing behavior (Refs. 8, 11, 13). Experiments isolated and either indicated or measured the effects on diffusing (defined by the changing percent of message knowers) of all six of the factors and their four facets."11 In short, we observed or in some cases believed from experience that diffusion could be altered by any of these twenty four influences, some of them subdivisible much further. We further have not found and cannot conceive of any influences on diffusion which are not subsumable within these twenty-four factors and facets (as a result of defining the residual category, C). This means that all our six basic factors of a transact and each factor with its four facets, and nothing else, were potential influencers of the diffusion to be predicted. All twenty-four "factor-facets" must be controlled or held constant by our model. This is the sort of complexity which has often been urged as preventing sociology from becoming an experimental or exact science. Yet every one of these twenty-four factors and facets was perfectly controlled in a simple classroom experiment repeatable by any instructor at any time. This published experiment (Refs. 13, 15, 17) on logistic diffusion of an act achieved demonstrable 100 percent control of all these twenty-four influences so that the specified pre-conditions of random steady pairing existed fully and solely. By getting twenty students to pair off by lot in each of ten one-minute periods and tell their partners all the news items (such as each other's birth dates) which each had heard during these pair-meetings, the diffusion (measured by the percent of knowers) grew, minute by minute, in an S-shaped logistic curve-just as predicted beforehand on the blackboard. The correlation of increments in the observed fact with the expected increments by logistic theory was r .99.12 The small discrepancies from the model were random ones as indicated by a non-significant chi square test at the 5 percent level. Here, we submit, in this checkable illustrative social experiment is a promise that transact models can help to build an experimentally based, increasingly exact, and predictive science of sociology.
References
1. Ashby, Ross, Design for a Brain, John Wiley & Sons, Inc., 1952. 260 pp.
2. Bridgman, Percy W., Dimensional Analysis, Yale University Press, 1931.
3. Committee on the Introductory Course, "Report to the American Sociological Society," Journal of Educational Sociology, Vol. VII, No. 1, Sept. 1933.
4. Dodd, Stuart C., Dimensions of Society, Macmillan, 1942.
5. Dodd, Stuart C., Systematic Social Science, American University of Beirut, Social Science Series, #6, 1947 (University Bookstore, Seattle, Agents)
6. "Tilp -- A Ten-Letter Alphabet of Meanings," General Semantics Bulletin, Spring-Summer, 1951, No. 6 & 7,
7. Dodd, Stuart C., "Sociomatrices and Levels of Interaction - for dealing with plurals, groups, and organizations," Sociometry, Vol. XIV, No. 2 & 3, May-Aug. 1951.
8. Dodd, Stuart C., On All-or-None Elements and Mathematical Models for Sociologists," American Sociological Review, Vol. 17, No, 2, April 1952.
9. Dodd, Stuart C., Something out of Nothing," Main Currents in Modern Thought, Vol. VIII, No. 4, February 1952.
10. Dodd, Stuart C. "Model English," General Semantics Bulletin, No. 10, Summer 1952
11. Dodd, Stuart C., "Can the Social Scientist Serve Two Masters - an answer through experimental sociology," Proceedings of the Pacific Sociological Society, Research Studies of the State College of Washington, Vol. XXI, No. 3, Sept. 1953.
12. Dodd, Stuart C., "Model English - an example of a national language regularized for machine translation," M.I.T. Transactions on Mechanical Translation, Technology Press, 1955. –
13. Dodd, Stuart C., “Diffusion is Predictable,” American Sociological Review, 1955
14. Dodd, Stuart C., "The 'Steps-and-Parts' Model for Polling," UNESCO publication in1956.
15. Dodd, Stuart C., "Formulas for Spreading Opinions."
16. Dodd, Stuart C., A Dimensional System of Diffusion Models."
17. Dodd, Stuart C., Revere Studies on Interaction (volume ready for press)
Notes
1. This predictive transact model using six dimensions is a specific outgrowth from our earlier purely methodological "S-theory" based on four dimensions as presented in Dimensions of Society (Ref. 4) which was developed with more substantive content but still as a largely descriptive and not a predictive system in Systematic Social Science (Ref. 5).
2. The amount of detail and technical language in an operational definition thus depends in part on the hearer for whom the definition is pitched. This brief presentation in this paper is pitched from pretests on six graduate students or journal editors to be reproducible by sociologists who have some statistical and semantic training.
3. The dimensional formula for a complete "transactance" (which is an index of the behavior-in-context (Be) as expected or predicted by the six factors) is thus the product:
[Be Aa Pp Tt Ll Vv Cc] The Transactance Equation 1
The difference between a dimensional formula and a statistical formula should be noted – Equation 1 in square brackets is the former; Equation 2 in Table 1 is the latter. Dimensional formulas in three levels are "sums of products of powers of factors basic to a field of science." This is the physicist's definition (Ref. 2) and ours for sociology (Ref. 4, 5, 17). (Though we also use "4-level formulas" specified as "sets of sums of products of powers of factors.") A dimensional formula specifies the factors and exponents only, omitting the three other facets of classes, degrees, and cases which a statistical formula may also specify. Dimensional formulas thus show the shapes of curves when graphed, omitting details of size of units and of intercepts from the origin point which our pre-scripts specify for statistical formulas. Dimensional formulas are useful for classifying families of curves; statistical formulas are more useful for computing individual members of such families. A still more complete and useful set of eight categories is formed by taking all material things (things having mass) and symbols (words, numbers, records) out of the residual class, C, and using them as compound categories coordinate with the other six. We call this set of eight categories "type-parts" in order to avoid collision with the subset of six factors and further subset of four sectors (Ref. 14).
4. For systematic terminology we use:
an "act" for any behavior observed as all-or-none
an "action" for any compound of acts, including a behavior in degrees
an "actor" for any human doer
an "acting" for any on-going act (present active participle)
an "activity" for any set of actions and actings
an "actance" for any product of the factors when taken as a predictor of a criterion behavior
a "unit-act" for an instance of any act of one person at one time and place and at one degree of each motivation and residual circumstance
a "react" for any response of a reactor to an act
a "contact" for a one-way act by A and react by B
an "interact" for any product (i.e., joint occurrence) of an act and a react and its reverse; a two-way contact as when A acts, B reacts by an act to which A then reacts
a "role-act" for any act of a specialist or role-actor
a "transact" for an act-in-context of its five further factors, i.e., any behavior when explicitly distributed in the six dimensions of any "total situation"
a "partid transact" for any transact with some of its dimensions controlled, i.e., observed at just one point
a "counter-act" for any conflicting act cancelling a given act
a combination of a prefix with a suffix, e.g., "transactance" for a predicter-act-in-context; "interactance" for a predicter index of interacts, etc.
5. In a companion paper, Professor A. A. Lundberg marshals convergences in the categories used in the systems of Parsons, Bales, and Dodd.
6. The general formula for a discrete statistical moment,
Am/P Equation 3
about zero computed for any exponent or power of the attribute, A, gives p, the proportion or simple probability of knowers in that population, i.e., Σ Am/P = p if A = 1,0 for any m. Similarly every moment, about zero, of not-A gives q, the proportion of non-knowers of the news item. Then the central moments (about the mean, p) yield in simplest form (Ref. 16):
Zeroth moment = = p + q, an alternative probability, asserting "A or not-A"
First moment = = 0, a nul probability, asserting "neither A nor not-A"
Second moment = = pq, a joint probability, asserting "both A and not-A"
Third moment = = q – p, a difference probability, asserting "not-A without A"
Fourth moment = = 1/pq, an inverse of a joint probability, asserting "all not both A and not-A"
7. Note especially in mathematical-physiological-psychology, Ross Ashby's Design for a Brain (Ref. 1), which expects an adaptive system to arise best from
a. many
b. part and
c. step functions
d. randomly combined.
We see our transact theory based on random combinings of elements among our factors as spelling out a possible base of the adaptive system we call "human society."
8. Thus, for example, the logistic predicts growth of diffusion in each unit-period by using the product, pq, not the sum, p + q, because interacting and not alternative acting is going on. The product will always correlate (aside from sampling error) at unity (r = 1.0) with the data of diffusing under pre-conditions of random, steady, pair- ing; while the sum will always correlate at zero (r = 0) in identical data. For the sum (p + q = 1) is a constant and will always correlate at 0 with any variable. Experimentally and rationally a model adding the variables is here worthless for prediction, while a model multiplying the identical variables predicts perfectly. Merely treating identical data here as either a sum or a product results in the worst possible or best possible predictivity. Such excellent predictivity in social science requires models for formulating it rigorously.
9. S general descripted formula Equation 4
where S means a name for any recorded data or set of symbols on the right-hand side
X means any variables
means the four corner scripts specifying the four facets of each factor
; means any logical or mathematical operator such as + - X etc.
10. This "powers model" described elsewhere (Refs. 13, 15, 16, 17) is:
Be Equation 7
which operationally defines (for calculation and prediction) the normal, exponential and logistic curves according as a and t denote one or more acts and as m is 0, 1, or 2.
11. These twenty-four factor-and-facets may be spelled out in the definitional Eq. 6 as follows:
|1 |the power-level of observing the act |= Aa |= |= the four facets of | |
| | | | |the acts factor |= Eq. 6 |
| | | | | |the factors and |
| | | | | |facets of transact, |
| | | | | |an expanded example |
| | | | | |of Eqs. 1, 2, 4 |
|2 |the number of different acts |= Aa | | | |
|3 |the degrees of each act |= aA | | | |
|4 |the boundaries of each act |= aA | | | |
|5 |the powers of the people (as persons P0, plurals P1, groups P2) |= Pp |= |= the four facets of | |
| | | | |the population factor | |
|6 |the different kinds of plurals |= Pp | | | |
|7 |the number of people |= pP | | | |
|8 |the presence of deviant persons |= pP | | | |
|9 |the power-level of the value, i.e., satisfactory frequency, mean, and |= Vv |= |= the four facets of | |
| |dispersion | | |the values factor | |
|10 |the kinds of values, i.e., motives to interact here |= Vv | | | |
|11 |the degree of each value |= vV | | | |
|12 |the threshold and limits of tor the value |= vV | | | |
|13 |power-levels of the time factor, i.e., instants T0, durations T+1, |= Tt |= |= the four facets of | |
| |speeds T-1, and celerations T-2 of acting | | |the time factor | |
|14 |the patterning of the periods |= Tt | | | |
|15 |the units of time |= tT | | | |
|16 |the initial and terminal instants |= tT | | | |
|17 |the powers of space (involving locations, distances, or areas) |= Ll |= |= the four facets of | |
| | | | |the space factor | |
|18 |the kinds of space, i.e., its homogeneity or absence of barriers |= Ll | | | |
|19 |the amounts of distance between actors |= lL | | | |
|20 |the initial location of actors |= lL | | | |
|21 |the power-levels of specified circumstances (i.e., qualitatively, |= Cc |= |= the four facets of | |
| |quantitatively, or correlationally observed) | | |the acts factor | |
|22 |classes of circumstances (e.g., sex, friendships, willingness to |= Cc | | | |
| |interact | | | | |
|23 |degrees of each circumstance |= cC | | | |
|24 |limits, etc. of each circumstance |= cC | | | |
12. The equation for the simple logistic in recursive form is:
Pt + 1 = Pt + ptqt Equation 5
where p = the knower proportion or mean of all-or-none act of "hearing the news item,"
q = the non-knowers, and
t = each time period in turn.
[pic]
-----------------------
•
sired.
(inorganic)
vii
t
................
................
In order to avoid copyright disputes, this page is only a partial summary.
To fulfill the demand for quickly locating and searching documents.
It is intelligent file search solution for home and business.
Related searches
- dodd 8 hour training
- dodd eight hour training session
- dodd 8 hour free training
- john stuart mill perfect science
- from john stuart mill s viewpoint
- dodd my learning portal
- my dodd log in
- dodd portal billing
- dodd training online
- dodd training ohio
- brian stuart houston long island
- helping people succeed stuart fl