SOME BASIC CONCEPTS IN PSYCHOLOGICAL RESEARCH

[Pages:24]Gavin-Ch-03.qxd 11/21/2007 2:31 PM Page 39

CHAPTER 3

SOME BASIC CONCEPTS IN PSYCHOLOGICAL RESEARCH

Contents

Who takes part in research? Populations Samples Variables Independent Fixed Random Dependent Confounding Level of measurement Quantitative research design Temporal Precedence Covariation Alternative Explanation Internal validity Replicability and reliability External validiy

Gavin-Ch-03.qxd 11/21/2007 2:31 PM Page 40

40 Understanding Research Methods and Statistics in Psychology

Hypotheses Handling statistical data Qualitative and quantitative data Why use qualitative methods? Reflexivity Validity in qualitative research Types of qualitative research Ethnography Phenomenology Hermeneutics Grounded theory Qualitative data collection In-Depth Interviews Direct observation Participant observation Case studies Diaries Handling qualitative data Summary

Gavin-Ch-03.qxd 11/21/2007 2:31 PM Page 41

Some Basic Concepts in Psychological Research 41

Some Basic Concepts in Psychological Research

Like other occult techniques of divination, the statistical method has a private jargon deliberately contrived to obscure its methods from non-practitioners. (G.O. Ashley)

Learning Objectives

? To examine the basic vocabulary of research. ? To examine some of the range of quantitative and qualitative methods. ? To understand the way that terms are used in research. ? To examine the debate over the nature of quantitative and qualitative research and the ten-

sion between them. ? To understand the various forms of variables. ? To introduce the concept of measurement and type of measurement. ? To introduce the concept of design and types of design. ? To introduce the concept of hypothesis generation and testing. ? To examine some of the ways that psychological data can be transformed into meaningful

summaries. ? To discuss what each type of summary can be used to describe.

KEY TERMS

? Case studies ? Central tendency ? mean, median and mode ? Design ? temporal precedence, covariation alternative explanations ? Dispersion ? maxima, minima, range, variance, standard deviation ? Display ? graphs, tables ? Distribution ? normal, skew, kurtosis ? Ethnography ? Grounded theory ? Hermeneutics ? Hypotheses ? alternative, null ? Interviews

Gavin-Ch-03.qxd 11/21/2007 2:31 PM Page 42

42 Understanding Research Methods and Statistics in Psychology

? Levels of measurement ? nominal, ordinal interval/ratio ? Observation ? Phenomenology ? Populations and samples ? Variables ? independent, fixed, random, dependent, confounding

Learning about research methods is like learning a new language, and learning a new language cannot start without understanding some basic rules and equipping ourselves with some basic vocabulary. This chapter will outline some of that vocabulary. Firstly, we will discover the language of quantitative research and examine some fundamental concepts that comprise the vocabulary on which we can build fluency in any area of statistical data analysis. We will then look at the ways in which qualitative methods allow us to approach research questions differently, not simply as a non-numerical alternative to quantitative methods, but as valuable ways of addressing the investigation of topics.

WHO TAKES PART IN RESEARCH?

Populations

KT

In psychology, quantitative research is almost exclusively carried out on samples

drawn from populations. Here the term population has a slightly different mean-

KT

ing from the one we use in everyday speech. It need not refer only to people or

creatures, for example the population of London, or the population of hedgehogs

in Huddersfield. In research, we can also refer to a population of objects, events,

procedures or observations. A population is thus an aggregate of things.

We must always clearly define the population we are interested in, but we may

not be able to describe and enumerate it exactly. For example, we might want to

know the average IQ of psychology students, but who are these people? At any

one time, the population of psychology students may contain people of different

sexes, ages, socioeconomic and ethnic background, etc. Also, at one time, every

psychology lecturer has been a psychology student. The researcher needs to pro-

vide a precise definition of a population and the constraints on that definition

(such as time and location) in order to draw valid inferences from the sample that

was studied, to the population being considered. Statistics that we will consider

when taken from populations are referred to as population parameters. They are

often denoted by Greek letters: the population mean is denoted by ? (mu) and the

standard deviation denoted by (lower case sigma).

Gavin-Ch-03.qxd 11/21/2007 2:31 PM Page 43

Some Basic Concepts in Psychological Research 43

Samples

Even if a population can be defined, it will usually contain too many individuals to study, so research investigation is commonly confined to one or more samples drawn from it. A good sample will contain the information that the population does, so there must be an effective relation between the sample and the population. One way of providing this is to ensure that everyone in the population has a known chance of being included in the sample, and also it seems reasonable to make these chances equal. We also want to be certain that the inclusion of one population member does not affect the chance of others being included. So the choice is made by some element of chance, such as spinning a coin, or in large populations and samples by use of tables of random numbers. These are widely published alongside other tables used in statistical analysis.

VARIABLES

KT

Variables are things that we can measure, control or manipulate in research

because they can have more than one value. There are different types of variables,

and we can consider numerical variables such as IQ, where the values would be

the score measured, or non-numerical variables such as sex (values are male or

female). Types include independent, fixed, random, dependent and confounding.

Independent

KT

Independent variables are those that could have an effect on other variables. For

example, it is possible that, as people get older, their short-term memory becomes

less effective, and we could test that by comparing the performance on a memory

test of several people who are deemed to be young adults and older adults. The

variable that may be having an effect here is the age of the people being tested.

Fixed

A fixed variable is one where we have specific set values for the independent variable included in the study. For example, in our aging and memory study, although age changes every year, at the moment in which the participants take part they each have a particular measured age. Therefore, we can fix our variable `age' in terms of specific groups, such as 18?25 and 65?75.

Gavin-Ch-03.qxd 11/21/2007 2:31 PM Page 44

44 Understanding Research Methods and Statistics in Psychology

Random

In the study on aging and memory, we could fix our values at the age ranges in the fixed variable example. This might be representative of the aging process, but there are many more values available with the variable `age'. With a variable that has many values, we would not necessarily wish to use every value, and could randomly select values from it. With age, this might not be appropriate as we wish to see the effect that aging has; a random selection of age groups does not necessarily give us the structure we need for our independent variable and we would prefer to use a fixed set for our variable.

Dependent

KT

A dependent variable is one that might be affected by the variations in the inde-

pendent variable. In our memory study, the independent variable, age, might have

an effect on short-term memory; our dependent variable is the scores achieved on

a memory test. So a dependent variable is one whose values may depend on the

different values in the independent variable.

Confounding

In some cases, there are variables that can affect the outcome, but which are not

strictly part of the study. For example, in our memory study it might be thought

that the effect of aging could be altered by certain types of drugs, so we might

want to exclude people who are on certain types of medication. Or it may be that

even though there is an established sex difference in memory, this difference

varies with age (see Bleecker, Bolls-Wilson, Agreus and Meyers, 1988), so we

might want to ensure that we have equal numbers of men and women in our sam-

ples of each age group to ensure we can compare them to see if this is the case. In

KT

this way, we control for confounding variables.

LEVEL OF MEASUREMENT

Variables differ in the way they can be measured, firstly in the amount of error that is inherent in the measurement (we will examine this later with respect to specific types of measurement) and secondly the amount of information that can be provided by a variable. This last is referred to as the type of measurement scale.

Gavin-Ch-03.qxd 11/21/2007 2:31 PM Page 45

Some Basic Concepts in Psychological Research 45

Variables are classified as nominal, ordinal, interval or ratio, and this distinction is

KT

referred to as the levels of measurement. Each variable that we might investigate

in psychology has a set of characteristics that indicate their nature, as a separate

characteristic, irrespective of what we might do with them.

Nominal variables allow for only categorisation into named sets, and all we know

is that individual items belong to some distinctively different categories, but there is

no quantifying or ranking of items. So we can know whether individuals are male or

female, but there is no indication that being in one category is better than the other

(irrespective of your personal feelings!). We can know whether one category has

KT

more in it than another, so we can know the frequency of each category.

Ordinal variables are those in which we place items in rank order in terms of

which has less and which has more of the quality represented by the variable, but

not how much more. A typical example of an ordinal variable is socioeconomic

status. We know that middle class is higher than working class but we cannot say

that it is, for example, 30% higher.

Interval variables not only give rank order but also quantify and compare

the sizes of differences (or interval) between them. Temperature is an interval scale.

A temperature of 50 degrees is higher than a temperature of 40 degrees, and the

increase from 40 to 50 degrees is half as much as an increase from 20 to 30 degrees.

In addition to all the properties of interval variables, ratio variables have an iden-

tifiable absolute zero point, thus they allow for statements such as 100 kg is two times

more than 50 kg. Typical examples of ratio scales are measures of time or space. For

example, as the Kelvin temperature scale is a ratio scale, not only can we say that a

temperature of 200 degrees is higher than one of 100 degrees, but we can correctly

state that it is twice as high, though this does not apply to the Fahrenheit scale. The

zero point must be meaningful, and 0 degrees Fahrenheit is arbitrary. Most statistical

data analysis procedures we use in psychology do not differentiate between the inter-

val and ratio properties of the measurement scales and the distinction is unimportant.

One way to remember the distinction between different levels of measurement

is to compile them all into one example. The one I liked to use when learning the

differences is the idea of a foot race. There is a certain number of people who will

take part, some professionals and some amateur, the numbers of whom are mea-

sured on a nominal scale, the number of different types of people, or the frequency

with which each type is counted is the nominal category of measurement. When

the race is ended, all the people will have finished in a particular order, first, sec-

ond, third, etc.; these are ordinal numbers and in the ordinal category of measure-

ment. Each person will have finished in a certain time and we can place them in

Gavin-Ch-03.qxd 11/21/2007 2:31 PM Page 46

46 Understanding Research Methods and Statistics in Psychology

order due to this, but also we can say that one person's time is less than another's so they have intervals between them that can be measured in seconds, or even minutes for a long race.

In psychology, although there may be more scope for flexibility than other sciences, we still need a systematic approach to research. If we are to make statements about human behaviour, we must do so in the light of good methods or we are simply speculating.

QUANTITATIVE RESEARCH DESIGN

Most empirical quantitative research belongs clearly to one of two general cate-

gories. In correlational research we do not (or at least try not to) influence any vari-

KT

ables, but only measure them and look for relations (correlations) between some set

of variables, such as weight and cholesterol level. If we see a relationship, we might

conclude that being overweight causes high cholesterol levels, but it could be just as

valid to say that high cholesterol means it is difficult to lose weight or prevent it

going up. So correlation research does not seek to establish causal relationships

between variables, just the strength and direction of the relationship.

In experimental research, we manipulate some variables and then measure the

effects of this manipulation on other variables; for example, a researcher might have

participants deliberately increase their weight and then record cholesterol level (not

the most ethical of studies!). Only experimental data can conclusively demonstrate

causal relations between variables. For example, if we found that whenever we

change variable A then variable B changes, then we could conclude that A influences

B. Data from correlational research can only be interpreted in terms based on some

theories that we have; correlational data cannot conclusively prove causality.

In order to claim causality we need three elements:

? Temporal precedence ? Covariation of cause and effect ? No alternative explanation

Temporal Precedence

In order to be able to say that one thing caused another, the first thing has to happen before the second. This might not be quite as easy as you think and it could be a classic case of chicken and egg. For example, our study of weight and blood must establish which state happened first, the obesity or the high cholesterol.

 ................
................

In order to avoid copyright disputes, this page is only a partial summary.

Google Online Preview   Download