ITie purpose of this paper is to examine the usefulness of factor analysis in developing and evaluating personality scales that measure hmited domain constructs The approach advocated follows from several assumptions that a smgle scale ought to measure a smgle construct, that &ctor analysis ought to be applied routinely to new personality scales, and that the factors of a scale are important if it can be demonstrated that they are diiFerenbally related to other measures A detailed study of the Self-Monitonng Scale illustrates how &ctor analysis can help us to understand what a scale measures A second example uses the selfesteem hterature to illustrate how factor analysis can clarify the proliferation of scales within a single content domain Both examples show how fJEictor analysis can he used to identify important conceptual distinctions Confirmatory techniques are also introduced as a means for testing specific hypotheses It is concluded that &ctor analysis can make an important contnbuhon to programmatic research m personality psychology One of the major research traditions m personality psychology has been the measurement of individual differences Rather than attempting to create vanance between groups through the use of expenmental manipulation and control, personality researchers typically focus instead on measunng existing vanation across individuals Consequently, statistical analyses more often involve measures of relationship (e g , correlatWHi coefficients) than tests of mean differences (e g , analysis of vanance) In the terminology of Cronbach's (1957) classic article, personality is part of the "correlational" (as opposed to the expenmental) discipline of psychdogy Consistent with this emphasis on a correlational or differential research strategy, personahty psychologists have gravitated toward statistical techniques that allow them to examine xaore thoroughly and moare ngorously the relaticniships m multivanate data sets Factor analytic proHie authors gratefully acknowledge the helpful comments of

A factor analysis of the Self-Monitoring Scale yielded three replicated factors: Acting, Extraversion, and Other-Directedness. Acting includes being good at and liking to speak and entertain. Other-Directedness is a willingness to change one's behavior to suit other people, and Extraversion is self-explanatory. Other-Directedness correlates positively with Shyness and Neuroticism and negatively with Self-Esteem. Extraversion correlates negatively with Shyness and positively with Self-Esteem and Sociability. Two of the scale's three factors, therefore, have opposite patterns of correlations with other personality dimensions. The three distinct factors help to explain certain discrepancies found in previous research with the Self-Monitoring Scale. For future research, we suggest that scores for each of the factors are more appropriate than full scale scores. We conclude that there may be a gap between the construct of Self-Monitoring and how it is operationalized in the scale.Stage actors make gestures, speak dialogue, and respond to cues-all in the service of portraying a particular character or role written by the author. Several decades ago, Erving Goffman (1956) used the stage metaphor to characterize the way each of us acts in social contexts. He suggested that we behave the way others expect us to, that we are alert to subtle cues in our social environment, and that in general we engage in self-presentation.A sociologist, Goffman ignored individual differences, but surely people vary in the extent to which they stage-manage their social behavior. To assess such individual differences, Mark Snyder developed the Self-Monitoring Scale (1974). A description of the high selfmonitor is also a definition of the concept:The prototypic high selt-monitoring individual is one who, out of concern for the situational and interpersonal appropriateness of his or her social behavior, is particularly sensitive to the expression and self-presentation of relevant others in social situations and uses these cues as guidelines for selfmonitoring (that is, regulating and controlling) hisWe thank Robert Helmreich and John Loehlin for their generous assistance.Requests for reprints should be sent to Arnold

We investigated the evidence for a general factor of self-monitoring in six samples (N = 3,615). The results indicate the presence of two, relatively orthogonal, major factors (General Factors A and B). General Factor A, which presumably accounts for most of the empirical validity of the scale, correlated strongly with measures of the traditional personality dimension of social surgency such as exhibitionism, social potency, and extraversion. General Factor B is a weakened version of an earlier identified Other-Directedness factor, and it correlated positively with shyness and negatively with self-esteem. Additional analyses examined the 18-item revision of the Self-Monitoring Scale in terms of the extent to which it strengthens General Factor A, its relationship to the original version, its factor structure, and correlations with other personality measures. We discuss three flaws in the construct of self-monitoring: its assumption of the bipolarity of social and personal orientations, its assumption of uniformity among those who score high on the Self-Monitoring Scale, and its lack of clarity concerning the role of intentionality in self-presentational processes. We suggest that it is time to move beyond the construct presented by Snyder (1987) toward a more comprehensive investigation of self-presentation and social behavior.

No abstract

Several approaches to assessing the dimensions of the five-factor model are reviewed and evaluated. The items in the assessment instrument may be adjectives or phrases, and the instrument itself may have been developed specifically to measure the five factors or may have been reinterpreted in terms of the five-factor model. Data are presented comparing an adjective-based measure of the model (Goldberg, 1990) with two phrase-based measures (Costa & McCrae, 1985; Hogan, 1986), and recommendations are made for the choice of an instrument in different research contexts. Allport's (1937) distinctions between the structure of the trait lexicon and the structure of personality in individuals are reiterated.

The concept of shyness and its measurement were investigated in a series of studies (total N = 1,687). Data collection and analysis proceeded in three phases: the revision and continued development of a measure of shyness, the Social Reticence Scale; a psychometric comparison among five measures of shyness; and an examination of the factor structure underlying the construct of shyness. Phase 1 assessed the reliability and validity of the Social Reticence Scale, including ratings of videotaped monologues and ratings by significant others. Phase 2 compared the five shyness measures with one another on indices of internal consistency and with other relevant measures of emotionality, personality, relationships, and behavior. Items from the five shyness measures were combined in a factor analysis in Phase 3, and the resulting factors were correlated with the self-report and rating data obtained in Phase 2. Overall, the results from these studies confirmed that the shyness measures were valid, reliable, and empirically distinct from measures of related constructs. Behavioral validity was observed for several of the shyness scales. Additional analyses suggested that three interpretable factors underlie responses to the shyness scales but provided little support for drawing conceptual distinctions among types of shyness. Discussion focuses on the implications of these data for the measurement and conceptualization of shyness.

Convergent and discriminant validity were examined for 8 widely used preschooler, toddler, and infant temperament questionnaires in a cross-sectional design that included both mothers and day-care teachers as respondents. The analyses focused on activity level, negative emotionality, and approach–sociability. Surprisingly strong evidence was uncovered for convergence among scales intended to measure similar concepts, with most convergent validity coefficients in the .50s, .60s, and .70s. Most of the scales also showed reasonable discriminant properties; however, some scales intended to measure distinct concepts also intercorrelated highly enough (in the .40s and .50s) to raise concerns. These cross-scale correlations could often be understood by reference to the actual item content and the contexts of day care versus home. The main analyses were supplemented with a structural modeling approach.

No abstract