Quantitative Methods in Personality Research - New Mexico State University

Quantitative Methods in Personality Research

R. CHRIS FRALEY AND MICHAEL J. MARKS Volume 3, pp. 1637?1641 in

Encyclopedia of Statistics in Behavioral Science ISBN-13: 978-0-470-86080-9 ISBN-10: 0-470-86080-4 Editors

Brian S. Everitt & David C. Howell

John Wiley & Sons, Ltd, Chichester, 2005

Quantitative Methods in Personality Research

The aims of personality psychology are to identify the ways in which people differ from one another and to elucidate the psychological mechanisms that generate and sustain those differences. The study of personality is truly multidisciplinary in that it draws upon data and insights from disciplines as diverse as sociology, social psychology, cognitive science, development, physiology, genetics, clinical psychology, and evolutionary biology. As with other subdisciplines within psychology, quantitative methods play a crucial role in personality theory and research. In this entry, we identify quantitative methods that have facilitated discovery, assessment, and model testing in personality science.

Quantitative Methods as a Tool for Discovery

There are hundreds, if not thousands, of ways that people differ from one another in their thoughts, motives, behaviors, and emotional experiences. Some people are highly creative and prolific, such as Robert Frost, who penned 11 books of poetic works and J. S. Bach who produced over 1000 masterpieces during his lifetime. Other people have never produced a great piece of musical or literary art and seem incapable of understanding the difference between a sonnet and sonata. One of the fundamental goals of personality psychology is to map the diverse ways in which people differ from one another. A guiding theme in this taxonomic work is that the vast number of ways in which people differ can be understood using a smaller number of organizing variables. This theme is rooted in the empirical observation that people who tend to be happier in their personal relationships, for example, also take more risks and tend to experience higher levels of positive affect. The fact that such characteristics ? qualities that differ in their surface features ? tend to covary positively in empirical samples suggests that there is a common factor or trait underlying their covariation.

Quantitative methods such as factor analysis have been used to decompose the covariation among

behavioral tendencies or person descriptors. Cattell [3], Tupes and Christal [14], and other investigators demonstrated that the covariation (see Covariance) among a variety of behavioral tendencies can be understood as deriving from a smaller number of latent factors (see Latent Variable). In contemporary personality research, most investigators focus on five factors derived from factor analytic considerations (extraversion, agreeableness, conscientiousness, neuroticism, openness to experience). These `Big Five' are considered by many researchers to represent the fundamental trait dimensions of personality [8, 16].

Although factor analysis has primarily been used to decompose the structure of variable ? variable correlation matrices (i.e., matrices of correlations between variables, computed for N people), factor analytic methods have also been used to understand the structure of person ? person correlation matrices (i.e., matrices of correlations between people, computed for k variables). In Q-factor analysis, a person ? person correlation matrix is decomposed so that the factors represent latent profiles or `prototypes,' and the loadings represent the influence of the latent prototypes on each person's profile of responses across variables [17]. Q-factor analysis is used to develop taxonomies of people, as opposed to taxonomies of variables [11]. Two individuals are similar to one another if they share a similar profile of trait scores or, more importantly, if they have high loadings on the same latent prototype. Although regular factor analysis and Q-factor analysis are often viewed as providing different representations of personality structure [11], debates about the extent to which these different approaches yield divergent and convergent sources of information on personality structure have never been fully resolved [2, 13].

Quantitative Methods in Personality Assessment and Measurement

One of the most important uses of mathematics and statistics in personality research is for measurement. Personality researchers rely primarily upon classical test theory (CTT) (see Classical Test Models) as a general psychometric framework for psychological measurement. Although CTT has served the field well, researchers are slowly moving toward modern test theory approaches such as item response theory (IRT). IRT is a model-based approach that relates

2 Quantitative Methods in Personality Research

variation in a latent trait to the probability of an item or behavioral response. IRT holds promise for personality research because it assesses measurement precision differentially along different levels of a trait continuum, separates item characteristics from person characteristics, and assesses the extent that a person's item response pattern deviates from that assumed by the measurement model [5]. This latter feature of IRT is important for debates regarding the problem of `traitedness' ? the degree to which a given trait domain is (or is not) relevant for characterizing a person's behavior. By adopting IRT procedures, it is possible to scale individuals along a latent trait continuum while simultaneously assessing the extent to which the assumed measurement model is appropriate for the individual in question [10].

Another development in quantitative methodology that is poised to influence personality research is the development of taxometric procedures ? methods used to determine whether a latent variable is categorical or continuous. Academic personality researchers tend to conceptualize variables as continua, whereas clinical psychologists tend to treat personality variables as categories. Taxometric procedures developed by Meehl and his colleagues are designed to test whether measurements behave in a categorical or continuous fashion [15]. These procedures have been applied to the study of a variety of clinical syndromes such as depression and to nonclinical variables such as sexual orientation [7].

Quantitative Methods in Testing Alternative Models of Personality Processes

Given that most personality variables cannot be experimentally manipulated, many personality researchers adopt quasi-experimental and longitudinal approaches (see Longitudinal Data Analysis) to study personality processes. The most common way of modeling such data is to use multiple linear regression. Multiple regression is a statistical procedure for modeling the influence of two or more (possibly correlated) variables on an outcome. It is widely used in personality research because of its flexibility (e.g., its ability to handle both categorical and continuous predictor variables and its ability to model multiplicative terms).

One of the most important developments in the use of regression for studying personality processes was

Baron and Kenny's (1986) conceptualization of moderation and mediation [1]. A variable moderates the relationship between two other variables when it statistically interacts with one of them to influence the other. For example, personality researchers discovered that the influence of adverse life events on depressive symptoms is moderated by cognitive vulnerability such that people who tend to make negative attributions about their experiences are more likely to develop depressive symptoms following a negative life event (e.g., failing an exam) than people who do not make such attributions [6]. Hypotheses about moderation are tested by evaluating the interaction term in a multiple regression analysis. A variable mediates the association between two other variables when it provides a causal pathway through which the impact of one variable is transmitted to another. Mediational processes are tested by examining whether or not the estimated effect of one variable on another is diminished when the conjectured mediator is included in the regression equation. For example, Sandstorm and Cramer [12] demonstrated that the moderate association between social status (e.g., the extent one is preferred by one's peers) and the use of psychological defense mechanisms after an interpersonal rejection is substantially reduced when changes in stress are statistically controlled. This suggests that social status has its effect on psychological defenses via the amount of stress that a rejected person experiences. In sum, the use of simple regression techniques to examine moderation and mediation enables researchers to test alternative models of personality processes.

During the past 20 years, an increasing number of personality psychologists began using structural equation modeling (SEM) to formalize and test causal models of personality processes. SEM has been useful in personality research for at least two reasons. First, the process of developing a quantitative model of psychological processes requires researchers to state their assumptions clearly. Moreover, once those assumptions are formalized, it is possible to derive quantitative predications that can be empirically tested. Second, SEM provides researchers with an improved, if imperfect, way to separate the measurement model (i.e., the hypothesis about how latent variables are manifested via behavior or selfreport) from the causal processes of interest (i.e., the causal influences among the latent variables).

Quantitative Methods in Personality Research 3

One of the most widely used applications of SEM in personality is in behavior genetic research with samples of twins (see Twin Designs). Structural equations specify the causal relationships among genetic sources of variation, phenotypic variation, and both shared and nonshared nongenetic sources of variation. By specifying models and estimating parameters with behavioral genetic data, researchers made progress in testing alternative models of the causes of individual differences [9]. Structural equations are also used in longitudinal research (see Longitudinal Data Analysis) to model and test alternative hypotheses about the way that personality variables influence specific outcomes (e.g., job satisfaction) over time [4].

Hierarchical linear modeling (HLM) (see Hierarchical Models) is used to model personality data that can be analyzed across multiple levels (e.g., within persons or within groups of persons). For example, in `diary' research, researchers may assess peoples' moods multiple times over several weeks. Data gathered in this fashion are hierarchical because the daily observations are nested within individuals. As such, it is possible to study the factors that influence variation in mood within a person, as well as the factors that influence mood between people. In HLM, the withinperson parameters and the between-person parameters are estimated simultaneously, thereby providing an efficient way to model complex psychological processes (see Linear Multilevel Models).

The Future of Quantitative Methods in Personality Research

As with many areas of behavioral research, the statistical methods used by personality researchers tend to lag behind the quantitative state of the art. To demonstrate this point, we constructed a snapshot of the quantitative methods in contemporary personality research by reviewing 259 articles from the 2000 to 2002 issues of the Journal of Personality and the Personality Processes and Individual Differences section of the Journal of Personality and Social Psychology. Table 1 identifies the frequency of statistical methods used. As can be seen, although newer and potentially valuable methods such as SEM, HLM, and IRT are used in personality research, they are greatly overshadowed by towers of the quantitative past such as ANOVA. It is our hope that future researchers will

Table 1 Quantitative methods used in contemporary personality research

Technique

Frequency

Correlation (zero-order)

175

ANOVA

90

t Test

87

Multiple regression

72

Factor analysis

47

Structural equation

41

modeling/path analysis

Mediation/moderation

20

Chi-square

19

ANCOVA

16

Hierarchical linear modeling and

14

related techniques

MANOVA

13

Profile similarity and Q-sorts

9

Growth curve analysis

3

Multidimensional scaling

3

Item response theory

3

Taxometrics

2

Note: Frequencies refer to the number of articles that used a specific quantitative method. Some of the 259 articles used more than one method.

explore the benefits of newer quantitative methods for understanding the nature of personality.

References

[1] Baron, R.M. & Kenny, D.A. (1986). The moderatormediator variable distinction in social psychological research: conceptual, strategic and statistical considerations, Journal of Personality and Social Psychology 51, 1173 ? 1182.

[2] Burt, C. (1937). Methods of factor analysis with and without successive approximation, British Journal of Educational Psychology 7, 172?195.

[3] Cattell, R.B. (1945). The description of personality: principles and findings in a factor analysis, American Journal of Psychology 58, 69?90.

[4] Fraley, R.C. (2002). Attachment stability from infancy to adulthood: meta-analysis and dynamic modeling of developmental mechanisms, Personality and Social Psychology Review 6, 123?151.

[5] Fraley, R.C., Waller, N.G. & Brennan, K.G. (2000). An item response theory analysis of self-report measures of adult attachment, Journal of Personality and Social Psychology 78, 350?365.

[6] Hankin, B.L. & Abramson, L.Y. (2001). Development of gender differences in depression: an elaborated cognitive vulnerability-transactional stress theory, Psychological Bulletin 127, 773?796.

4 Quantitative Methods in Personality Research

[7] Haslam, N. & Kim, H.C. (2002). Categories and continua: a review of taxometric research, Genetic, Social, and General Psychology Monographs 128, 271?320.

[8] John, O.P. & Srivastava, S. (1999). The big five trait taxonomy: history, measurement, and theoretical perspectives, in Handbook of Personality: Theory and Research, 2nd Edition, L.A. Pervin & O.P. John, eds, Guilford Press, New York, pp. 102?138.

[9] Loehlin, J.C., Neiderhiser, J.M. & Reiss, D. (2003). The behavior genetics of personality and the NEAD study, Journal of Research in Personality 37, 373?387.

[10] Reise, S.P. & Waller, N.G. (1993). Traitedness and the assessment of response pattern scalability, Journal of Personality and Social Psychology 65, 143?151.

[11] Robins, R.W., John, O.P. & Caspi, A. (1998). The typological approach to studying personality, in Methods and Models for Studying the Individual, R.B. Cairns, L. Bergman & J. Kagan, eds, Sage Publications, Thousand Oaks, pp. 135?160.

[12] Sandstrom, M.J. & Cramer, P. (2003). Girls' use of defense mechanisms following peer rejection, Journal of Personality 71, 605?627.

[13] Stephenson, W. (1952). Some observations on Q technique, Psychological Bulletin 49, 483?498.

[14] Tupes, E.C. & Christal, R.C. (1961). Recurrent personality factors based on trait ratings, Technical Report No. ASD-TR-61-97, U.S. Air Force, Lackland Air Force Base.

[15] Waller, N.G. & Meehl, P.E. (1998). Multivariate Taxometric Procedures: Distinguishing types from Continua, Sage Publications, Newbury Park.

[16] Wiggins, J.S., ed. (1996). The Five-Factor Model of Personality: Theoretical Perspectives. Guilford Press, New York.

[17] York, K.L. & John, O.P. (1992). The four faces of eve: a typological analysis of women's personality at midlife, Journal of Personality and Social Psychology 63, 494?508.

R. CHRIS FRALEY AND MICHAEL J. MARKS

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

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

Google Online Preview   Download