Sie sind auf Seite 1von 9

Journal of Applied Psyeholoffy 1986, Vol. 7 I, No.

3, 402--410

Copyright 1986 by the American Psychological Association, Inc. 0021-9010/86/$00.75

A Meta-Analysis of the Relation Between Personality Traits and Leadership Perceptions: An Application of Validity Generalization Procedures
Robert G. Lord, Christy L. De Vader, and George M. AUiger
University of Akron This article reexamines the relation between personality traits and leadership perceptions or extent of leader emergence. We maintain that prior research on trait theories and leadership has been misinterpreted as applying to a leader's effect on performance, when it actually pertains to the relation of leadership traits to leadership emergence. Further, based on current theories of social perceptions, several traits were expected to be strongly related to leadership perceptions. Using the rectaanalytic technique of validity generalization, results supported this expectation with intelligence, masculinity-femininity,and dominance being significantly related to leadership perceptions. Also, findings showed that variabilityacross studies in the relation of these traits to leadership perceptions could be explained largely by methodological factors, indicating that contingency theories of leader. ship perceptions may not be needed. Both of these results contrast with the conclusions of earlier nonquantitative literature reviews on traits and leadership perceptions and with conventional thinking in the leadership area.

Trait theories have not been seriously considered by leadership researchers since Mann (1959) and Stogdill (1948) reported that no traits consistently differentiated leaders from nonleaders across a variety of situations. The thesis of this article is that, first, these reviews have often been misinterpreted, and second, there are both theoretical and methodological reasons for reconsidering the relations between the traits of potential leaders and their tendency to be perceived as leaders by others. The findings of the Mann and Stogdill reviews have been misinterpreted in three respects. First, though both reviews dealt with only leadership emergence or the perception of leadership in groups with no formal leader, many current theorists (Landy, 1985; Muchinsky, 1983) report that their conclusions pertain to the topic of leadership effectiveness. This confusion probably stems from the title of Mann's (1959) review, "A Review of the Relationships Between Personality and Performance in Small G r o u p s " Though Mann mentions performance, the relations he investigated were between personality and attained leadership status as indexed by peer ratings, observer ratings, or by being formally nominated as a leader by group members. None of the leadership studies he reviewed used group task performance or ratings of group effectiveness by observers as dependent variables. Second, there were many consistently significant

The authors would like to acknowledge the helpful comments of Ralph Alexander on an earlier version of this article. Christy De Vader is now at the Department of Psychology, University of Wisconsin, Oshkosh. George Alliger is now at the Department of Psychology, SUNY-Albany. Correspondence concerning this article should be sent to Robert G. Lord, Department of Psychology, University of Akron, Akron, Ohio 44325. 402

relationships between personality and leadership emergence in both reviews. For example, Mann (1959, p. 247) reported that 88% of the 196 relations between intelligence and leadership were positive, 92 of these 196 relations were significant, and 99% of the significant relations were in the positive direction. Nevertheless, Mann chose to emphasize the low median correlations between leadership and traits such as intelligence rather than the consistency in the trends he uncovered. Others have primarily restated his conclusions rather than critically reevaluating his empirical evidence. Third, the number of studies upon which Mann based his conclusions is far fewer than most people realize. Mann's unit of analysis was the relation between every measure of leadership and every measure of personality, not the aggregate findings of a particular study. Thus, whereas he investigated 196 relations between intelligence and leadership, these numerous relations came from only 28 independent studies, For other traits, such as masculinity-femininity, only 9 independent studies produced the 70 relations he investigated. The net impact of Stogdill's and Mann's reviews was substantial. Current experts in the leadership area (Mitchell, 1982, p. 370; Yukl, 1981, pp. 69-70) have noted the substantial impact of Stogdill's review on subsequent trait research. This negative perspective on trait theory is also reflected in current texts on applied psychology. For example, Muchinsky (1983, p. 403) notes that there is "little or no connection between personality traits and leader effectiveness," and Landy (1985, p. 428) cites the Stogdill and Mann reviews as having demonstrated "no relationship between personality factors and leadership effectiveness,' In short, what has occurred in the scientific literature is an overgeneralization of findings on personality and leadership perceptions to the issue of how personality relates to leader effectiveness. Moreover, the actual empirical results seem to have been interpreted too pessimistically by Mann, and even StogdiU

TRAITS AND LEADERSHIP PERCEPTIONS (1974, p. 72) has noted that his 1948 review was interpreted too negatively. But if one focuses on the question of how personality relates to leadership perceptions or emergence, as we do here, there are sound theoretical and methodological reasons for a careful reevaluation. Theoretical work in the social-cognitive area suggests that traits should affect social perceptions. For example, Mischel (1973), in a cogent criticism of trait theory, suggests that traits are important constructs for perceivers, helping them to organize perceptions of others. More recent work (Winter & Uleman, 1984) indicates that people unintentionally make trait inferences when encoding information into memory. Further, work applying categorization theory to the leadership domain (Lord, Foti, & De Vader, 1984) found several traits that were thought to characterize leaders in many situations. Thus, from the perceiver's perspective, the assessment of the traits of others should affect perceptions of other's leadership qualities. Hence, a cognitive perspective would argue that the traits of potential leaders should affect the extent to which they are perceived as leaders by others. Methodologically, there have been a number of important developments in statistically aggregating results across studies since the Mann and Stogdill reviews were published. Techniques such as meta-analysis give better estimates of population parameters than the median correlation used by Mann and by Stogdill, and they can also indicate the proportion of variability in results across studies explainable by methodological artifacts (Hunter, Schmidt, & Jackson, 1982; Pearlman, Schmidt, Hunter, 1980; Schmidt & Hunter, 1977). The purpose of the present work is to apply such techniques to the literature investigated by Mann in his 1959 review and to subsequent studies relevant to this area. In the following sections, we take a more detailed look at the social-cognitive perspective regarding leadership perceptions, provide more information on Mann's and StogdiU's reviews, and explain the advantages of validity generalization as a form of recta-analysis. We then use the validity generalization approach to reexamine the Mann data base and discuss the implications of the resulting findings.l Social-Cognitive Explanation o f Leadership Perceptions The social-cognitive perspective explains social perceptions in terms of perceiver information processing, paying particular attention to selective attention, encoding, and retrieval of information. Following this perspective, Cantor and Mischel (1979) argue that perceivers develop systems of cognitive categories in which people can be grouped. Categories are defined in reference to prototypes, which are abstract collections of the attributes most commonly shared by category members. Because classifying others into categories involves matching stimulus characteristics to appropriate perceiver prototypes, prototypes should be key constructs for understanding person perception. If prototypes are widely shared in our culture and if they include many trait terms, traits should be important perceptual constructs, and our perceptions of others should be based on their match with the traits in our prototypes. This general theory of person perception is consistent with work in the leadership area. Hollander and Julian (1969) argued

403

in an important theoretical paper that leaders emerged in group situations by fitting the shared conceptions of followers, emphasizing the role of perceiver constructs in leadership processes. Though not stated in cognitive terms, Hollander and Julian's view is quite compatible with the prototype matching view of leadership emergence explained above. Stated somewhat differently, Hollander and Julian's work implies that followers would tend to allow others to lead when those others matched followers' ideas of what good leaders should be. This perspective on leadership emphasizes lay people's conceptualizations of leadership---that is, their implicit leadership theories (ILTs). Extensive work on ILTs has demonstrated widely shared beliefs about leader behaviors and traits (Eden & Leviatan, 1975; Lord et al., 1984; Rush, Thomas, & Lord, 1977; Weiss & Adler, 1981), that guide perceiver's encoding of relevant information (Phillips, 1984; Phillips & Lord, 1981, 1982), their formation of leadership perceptions (Fraser & Lord, 1983; Lord et al., 1984), and their reconstructive recall of leadership information (Lord, 1985). Attempting to develop a cognitive definition of ILTs, Lord, Foti, and Phillips (1982) argued that ILTs were simply a type of category system. Their theory emphasized the role of leadership categories and prototypes, showing that the prototypicality of behavior or trait information explained both leadership perceptions and distortions in memory about leaders. More recent work along these lines extensively investigated the specific attributes associated with leadership in different situations. Lord et al. (1984) argued that leadership was a superordinate category and that leadership, in conjunction with various contexts (business, military, education), defined basic level categories. They found that several traits characterized leaders in many of these contexts. Although scientific and lay definitions of traits differ, many of these traits can be associated with the trait constructs investigated by Mann (1959). Based on the correspondence of traits presented in Table 1, intelligence should predict leadership emergence or perception in almost all situations; extroversion-introversion and masculinity-femininity should predict leadership perceptions in many situations. Interpersonal sensitivity, dominance, conservatism, and adjustment probably would not predict leadership. In short, consistent with the social-cognitive perspective, research on ILTs shows that cognitive schema composed primarily of traits are important perceptual constructs that should predict leadership perceptions or leadership emergence. This implication contrasts with more traditional thinking in the leadership area based on common interpretations of the Stogdill and Mann reviews. Reviews by Stogdill and by M a n n The prevalent belief that traits do not predict leadership perceptions can be traced directly to two widely cited reviews by Stogdill (1948) and by Mann (1959). Stogdill reviewed studies The Mann data was chosen in preference to the data reviewed by Stogdill because it (a) was more recent, (b) included studies that tended to provide more of the information necessary for a meta-analysissuch as N size, and (c) did not include studies in which children were subjects, as did Stogdill.

404 Table 1

R. LORD, C. DE VADER, AND G. ALLIGER Several findings from Mann's (1959) review are noteworthy. First, trends were strongly supportive of relationships between personality variables and leadership perceptions. For example, the percentage of studies indicating positive relationships with leadership was 88% for intelligence and was greater than 70% for all other personality variables except conservatism. Results were even more dramatic when the percentage of significant resuits (which probably had larger N sizes) in the positive direction was examined. For intelligence, 99% of the significant resuits were in the positive direction, and this was 96% for adjustment, 94% for sensitivity, 92% for masculinity, 85% for extroversion, 71% for dominance, and 15% for conservatism. Thus, there were significant and consistent trends in the relation of personality to leadership emergence. However, in spite of these impressive trends, Mann emphasized the variability in resuits and the low median correlations he found. For example, he concluded his review by reporting that "in no case is the median correlation between an aspect of personality covered here and performance higher than .25, and most of the median correlations are closer to. 15" (p. 266). The thesis of the present article is that Stogdill's and Mann's findings have been misinterpreted. The variability in results they report may be due as much to methodological artifacts as to the variable relationship between personality and leadership perceptions. Further, the low relationships often found may refleet unreliability of measures or range restriction in the sampies investigated. Finally, median correlations probably do not provide very good estimates of population parameters, and thus, conclusions based on median correlations may be quite misleading. Current methodology provides more precise ways to aggregate results over studies and to estimate population parameters. These approaches are described in the following section. Validity Generalization Validity generalization, one type of meta-analysis, was developed within the context of industrial/organizational psychology (Schmidt & Hunter, 1977; Schmidt, Gust-Rosenberg, & Hunter, 1980). (See Appendix for studies included in the Validity Generalization Analysis.) Validity generalization has several advantages over other meta-analytic techniques. First, it corrects for some sources of artifactual variance across studies. Second, it provides an estimate of the population effect size, whereas many meta-analyses focus only on cumulating significance indices. Third, it provides a test of homogeneity of variance, which gauges whether the studies' results came from a single population or whether situational variables may moderate the effect size distribution (Hunter et al., 1982). Validity generalization postulates seven sources of artifactual variance that may act to reduce or attenuate individual study results: sampling error, differences across studies in criterion and predictor reliability, differences across studies in range restriction, clerical errors, criterion contamination, and mistakes in analysis. The three sources of error that are practical to estimate when cumulating studies are error due to sampling, predictor unreliability, and range restriction. It is interesting to note that sampling error has accounted for the majority of explained variance in validity generalization

Correspondence Between Traits in Implicit Leadership Theories and Predictors of Leadership Perceptions From Mann (1959)
Family resemblance values .91 .55 .45 .36 .27 .18 .00 .27 .45 .09 .09 .36 .09 .09 .09 .09 .09 .18 .18 .09

Mann's traits Intelligence Extroversion-introversion Masculinity-femininity Interpersonal sensitivity

ILT traits Intelligence Outgoing Verbal skills Aggressive Decisive Unemotional Sensitive Caring Understanding Interested Concerned Determined Directive Tough Cooperative Flexible Conservative Strict Authoritarian Disciplined

Dominance

Conservatism

Adjustment

Note. ILT = implicit leadership theory. Numbers are Family Resemblance Values from Lord et al. (1984), which indicate the proportion of I 1 different situations in which subjects' combined implicit leadership theories indicated that a trait characterized leaders.

from 1904 to 1947 that used physical characteristics or personality traits to differentiate leaders from nonleaders. For most predictor variables, he found quite variable results across studies, although it should be noted that the measures of leadership, predictor variables, and population investigated also varied widely across these studies. His most consistent findings pertained to the relationship between intelligence and leadership. He cited 23 studies that indicated that the average child or student leader surpassed the average member of his group in intelligence, 5 studies that found no difference in intelligence, and 5 studies that indicated that too large a difference was detrimental to leadership. Of the 17 studies reporting correlations, the highest correlation was .90, and the average correlation was .28. Clearly, there was a significant trend indicating that leadership and intelligence were associated. Mann (1959) updated Stogdill's review, focusing on the relationship between personality variables and leadership perceptions in small groups. His review covered over 1,400 relations (p. 245) between personality and leadership, popularity, or task activity. In the reviewed studies, personality was defined in numerous ways, but Mann was able to organize over 350 different personality measures into groupings involving seven major dimensions: intelligence, adjustment, extroversion-introversion, dominance, masculinity-femininity, conservatism, and interpersonal sensitivity. He then catalogued the trends in results between measures of these personality dimensions and leadership.

TRAITS AND LEADERSHIP PERCEPTIONS studies. Since sampling error is simply a function of the N-size of the different studies, it is always possible to precisely estimate for sampling error, since all studies report N-size. Reliability and range restriction in predictors, however, are often left unreported, so that assumed distributions must be used to estimate these additional sources of artifactual variance (Alexander, Carson, Alliger, & Cronshaw, 1984; Pearlman et al., 1980; Schmidt & Hunte~ 1977). Applying the Schmidt-Hunter validity generalization technique requires the calculation of an N-size weighted mean r as the best estimate o f the population correlation. Next, the N-size weighted total variance of the sample correlations around this mean is calculated. This is usually called the observed variance. It is this observed variance that is inflated to a certain extent by sampling error and other artifacts. Then the variance due to these artifacts is estimated (see Hunter et al., 1982, for this and other formulas), and this estimate is subtracted from the observed variance. The issue then becomes whether the remaining, residual, variance is small enough to warrant the conclusion that all the observed correlations came from a single population. If 75% or more of the variance c a n be explained by methodological artifacts, then the assumption of homogeneity is warranted (Pearlman et al., 1980; Schmidt & Hunter, 1981).2 In addition to these corrections to the observed variance, the frequency-weighted mean r is corrected for attenuation caused by criterion unreliability and range restriction. This corrected mean r is a final estimate of the unattenuated population r. Finally, the standard deviation of this corrected mean r is estimated, allowing confidence intervals to be calculated. It is then possible to see whether these intervals include zero. Thus, the confidence intervals represent a test of significance for the final corrected mean r. Method The articles identified by Mann (1959) as investigating the relationship between personalitytraits and leadership were used as the basis for initial meta-analyses. These studies were then pooled with subsequent studies in an additional set of meta-analyses. A list of the studies providing the minimal amount of information necessary for a meta-analysis is included in Table 2. Studies that only provided a verbal summary of results and those that reported information in such a way that it was not possible to transform the results into a correlation coefficient were excluded. Some studies reported data on more than one sample, resulting in 19 independent samples altogether based on the studies reviewed by Mann. Studies subsequent to Mann's review provide additional independent samples. For each sample only one correlation coefficient was analyzed. In the instances in which several coefficients were reported, such as when several measures of leadership perceptions were correlated with the same personality trait, we used only one data point, the average correlation (see note to Table 2). Frequently, several of the correlations that were averaged were based on different sample sizes even though many were the same subjects. In these situations as suggested by Hunter et al., (1982) we used the smallest sample size. Results Table 3 shows how our data base was derived from Mann's (1959) review.3 The first row of Table 3 pertains to the number o f results as given in Mann's original analysis. What is immedi-

405

ately striking is the large number of results that Mann reports for each of the six traits. They range from a high of 196 for intelligence to a low of 39 for dominance. However, Mann apparently defined "results" as all possible independent/dependent variable correlations, permitting the inclusion of multiple results from the same studies. In fact, in some instances as many as 53 results seem to have come from a single study. This procedure violates the meta-analytic requirement of sample independence. The bottom row of Table 3 shows the number of independent correlations used for each of the six validity generalization analyses we performed. Table 4 presents the results of the validity generalization program which was applied once to each of the six traits. The table displays for each trait the total number of subjects, the number of results (independent samples), followed by the observed variance that exists in the sample correlations, the predicted or estimated variance due to methodological artifacts and the difference between the observed and predicted variances or the residual variance. This is followed by the total percentage of explained variance; the frequency weighted mean r; and the true r (which is corrected for range restriction and unreliability); and, finally, for comparison purposes, the median rs reported by Mann. These results present a somewhat different picture of the strength of the relationship between personality and leadership perceptions than the median values reported by Mann. For example, two of our frequency weighted rs were substantially higher than the median rs reported by Mann, being .38 for intelligence and .24 for masculinity-femininity. Corrections for attenuation raised values to .52 for intelligence, .34 for masculinity-femininity, and .21 for adjustment. Though these differences in correlations are not great, they are important, especially because values should be squared to be properly interpreted in terms of effect size. For example, our corrected correlations show that intelligence explains four times as much variance in leadership perceptions as do Mann's findings. Part of these differences undoubtedly reflects the use of different methodologies, however, we also found other differences where identical analyses were used. For example, for three traits, the highest r that we found was substantially greater than the maxim u m correlations noted by Mann, being .78 rather than .50 for intelligence, .60 rather than .53 for adjustment, and .68 rather than .42 for extroversion. Conclusions concerning the strength, significance, and variability of the relations between personality and leadership perceptions based on including the data from studies subsequent to those reviewed by Mann are presented in Table 5. This should be viewed as the most accurate estimate of trends based on the currently available literature. Results show that intelli2 Hunter, Schmidt, and Jackson (1982, p. 45-47) acknowledge the existence ofa chi-square test for homogeneity of correlations that could, in theory, provide additional information on whether a distribution of correlations arise from a single population. However, they recommend against its use because it yields results too conservative to be useful. 3 Although Mann included a seventh personality trait sensitivity, it was excluded from our analysis because the construct was not as clearly defined as the other traits, and there was little consistency in how sensitivity was measured.

406
Table 2

R . L O R D , C. D E V A D E R , A N D

G. ALLIGER

Study Description and Actual Correlations Used in the Validity Generalization


Sample Study N Type I M A Trait D E C

Mann's Review
1. A r b o u s & M a r e e , 1951 219 Administrative trainees f o r a large i n d u s t r i a l corporation College women Supervisors at a large oil r e f i n e r y College women High school men College w o m e n College women High schoolwomen College w o m e n Graduates ofa military school Air ROTC instructor trainees College men College men College s t u d e n t s ( 1 , 3 7 4 ) collegc s t u d e n t s College men Students Students Students .47 . . . .

2. Bass, M c G e h e e , H a w k i n s , Young, & G e b e l , 1953 3. B a s s & Wurster, 1953 4. Bass, Wurster, Doll, & Clair, 1953 5. C a r t e r & N i x o n , 1949 6. D e x t e r & Stein, 1955 1 2 7. H e m m i n g , 1935 8. G o r d o n , 1952 9. G o w a n , 1955 10. G r e e n , 1 9 5 0 11. H o l t z m a n , 1952 1 2 12. Howell, 1 9 4 2 13. M c C u e n , 1 9 2 9 14. ~ r s t e r & Bass, 1953 15. Zeleny, 1939 I 2 3

76 87 49 100 80 64 71 168 485 54

-.50 b .27 e .01 b --.44 .21 .45 .45

--- . 12 b .23 b .47 b .3 lb ---. .

.20~ ~ .1 lb ~ -.06 .05 .21 .22 .ll . .

.22 b .35 d __ .06 ~ --.06 --

.17 t~ -.07 b --.3 l --. 12 .40 b .08 .ll

.20

.09

24 22 34 58 65 21 35 12

--.39 .40 f .25 .78 s


.44 h

--. . . ---

. . .

.60 b .51 b . . . -~

~ _ . . . ~ m

-_

.58 .17 .68

Subsequent studies
16. A s h o u r & E n g l a n d , 1972 19. G r e e n , 1948 1 2 20. G r e e n w o o d & M c N a m a r a , 1969 2 I. Heisler & G e m m i l l , 1977 l 2 22. L a m o n t & L u n d s t r o m , 1977 24. M i t c h e l , 1975 25. Page, 1935 26. R o w l a n d & Scott, 1968 27. T u r n e r & M a r t i n e z , 1977 1 2 101 23 23 100 Management students Male graduate students F e m a l e college s t u d e n t s Premanagerial personnel Managers Managers Industrial salesmen Managers West P o i n t c a d e t s Civil Service e m p l o y e e s Both samples are a random sample of questionnaire data on employed men and women ---.06 ---. -.72 b .68 b . .13 m -. -__ -. -__ --

34 52 71 454 115 53 500 418

. -.12 .48 b .23 .22 ---

. --------

. ----~ ---

.20 -.16 .17 -~ .00 .08 .10

--.03 .44 .06 ----

---------

Note. I = intelligence; M = m a s c u l i n i t y - f e m i n i n i t y ; A = a d j u s t m e n t ; D = d o m i n a n c e ; E = e x t r o v e r s i o n - i n t r o v e r s i o n ; C = c o n s e r v a t i s m . C o r r e l a t i o n r e p o r t e d w a s c o r r e c t e d f o r r a n g e r e s t r i c t i o n . b T h e c o r r e l a t i o n r e p o r t e d in t h e t a b l e is a n a v e r a g e o f t w o o r m o r e c o r r e l a t i o n s , c n = 49. d n = 76. n = 136. f 1,374 s u b j e c t s w e r e u s e d i n t h e study, b u t t h e c o r r e l a t i o n is b a s e d o n t h e n u m b e r o f g r o u p s (n = 58). s n = 19. h n = 34.

pence,

dominance,

and

masculinity-femininity

were

signifi-

In addition to strength and significance, the variability of the relations between correlations personality and leadership indicate perceptions is o f interest. High variability ters. It would would that mean corrected parame-

cantly associated with leadership perceptions. very close to being significant for adjustment. is based on the lower boundary being greater than zero.

Also, results were This conclusion interval

of a 95% confidence

were still not good estimates of population also suggest that contingency

theories of leader-

TRAITS AND LEADERSHIP PERCEPTIONS Table


3

407

BriefNumerical Description of the Data Base


Trait Descriptor Mann's number of results Averagenumber of results per study Number of studies in Mann's analysis Number of studies located for current analysis Number of studies that did not contain useable effect sizes or did not quantify the relationship between personality and leadership Number of studies in the present analyses Number of samples from which data was obtained for the recta-analysis I 196 7.00 28 22 M 70 7.77 9 6 A 164 7.45 22 16 D 39 3.25 12 10 E 119 5.60 21 16 C 62 3.65 17 11

10 12

3 3

9 7

6 4

9 7

9 2

13

10

Note. I = intelligence; M = masculinity-femininity; A = adjustment;


D = dominance; E = extroversion-introversion; C = conservatism.

ship perceptions might be warranted. According to Pearlman et al. (1980), if 75% or more of the variance is explained by artifacts, then homogeneity can be accepted and corrected, frequency-weighted means can be used as estimates of population parameters. As shown in Table 5, this criterion is clearly met for dominance, where 80% of the variance is explained. A sizeable amount of variance is also explained for intelligence and masculinity-femininity. Hence, contrary to popular thinking, for these three traits, there are fairly consistent and significant relations to leadership perceptions. Discussion The results of our validity generalization analyses produce a very different picture of the relationship between personality traits and leadership perceptions than that provided by Mann's review. To summarize, first, we found that the number of independent samples was generally far fewer than the number of relations reported by Mann. Second, we found several instances where the highest correlation was substantially greater than that reported by Mann. We can find no obvious explanation for this discrepancy, as we based our analyses on the same sources reported by Mann. Third, several frequency-weighted mean correlations (whether corrected or uncorrected) were higher than the median correlations reported by Mann. Fourth, using all quantitative data currently available, the "true" correlations between leadership perceptions and intelligence, and masculinity-femininity, and dominance were significant. Finally, for in-

telligence, masculinity-femininity, and dominance, most of the variance in results from study to study could be attributed to artifactual sources. In short, personality traits are associated with leadership perceptions to a higher degree and more consistently than the popular literature indicates. The issue of the amount of variability around population values is also of substantial interest, since it bears directly on the need for contingency theories of leadership perceptions. If this variability is substantially greater than one would expect based on artifactual sources, contingency theories to explain leadership perceptions would be warranted. However, if most of the variability around this parameter can be explained in terms of artifactual sources, contingency theories are not needed. On the basis of our results, we would accept the assumption of homogeneity of variance for dominance as a predictor, and we would be very close to accepting it for intelligence and masculinityfemininity as well. On the other hand, for adjustment, introversion-extroversion, and conservatism the mean correlations are not good estimates of population values. For these three traits the search for moderators may be warranted. It should be noted that the above conclusions concerning contingency theories pertain to leadership perceptions, not to leadership effectiveness or to group performance. As noted earlier, many scientists have overgeneralized results from leadership perceptions to the topic of leadership effectiveness. Heslin (1964) published a follow-up to Mann's (1959) review that did focus on group performance as the dependent variable. He found intelligence to be strongly and consistently related to group performance (rs generally in the .60s), and adjustment was also consistently related to group performance. Thus, personality may consistently predict performance as well as leadership perceptions. Interestingly, our validity generalization results are fairly consistent with conclusions based on the social-cognitive literature. Both would suggest that intelligence is a key characteristic in predicting leadership perceptions. Also, both suggest that other

Table 4

Results of the Validity Generalization Analysisfor Each of the Six Personality Traits
Trait Measure I M A D E C

N 1533 293 1039 420 1034 125 Number of results i3 4 9 4 10 2 Observed variance .019 .030 .014 .045 .032 .003 Predicted variance .018 .020 .009 .009 .009 .017 Residual variance .001 .009 .005 .035 .023 -.014 Percentage of explained variance 94.328 69.327 69.502 20.809 27.601 100.00 Mean r .381 .241 .145 .10 .101 .157 True r .518 .338 .210 .168 .145 .223 Median r (reported by Mann) .25 .15i .15 .20 .15 .15"

Note. I

Median not reported, but Mann (1959,p. 266) states that most medians were close to. 15.

= intelligence; M = masculinity-femininity; A = adjustment; D = d o m i n a n c e ; E = extroversion-introversion; C = conservatism.

408 Table 5

R. LORD, C. DE VADER, AND G. ALLIGER Our conclusion that some traits may indeed be important predictors of leadership perceptions agrees with conclusions based on other recent work in the leadership area (Kenny & Zaccaro, 1983) and in the general personality domain (Mischel, 1984). This work suggests that traits may be important organizational constructs for perceivers. However, as Mischel (1984) has noted, this does not imply that traits are arbitrary constructs of perceivers. Rather, as Mervis and Rosch (198 l) noted, perceivers based categories on structure existing in the environment, but their perceptual systems accentuate this structure. One should be careful in generalizing our findings on the relation of traits to leadership perceptions to other areas of leadership. They do not directly imply that there are also traits that would generally predict the performance of a leader's work group or organization, nor do they imply that there are certain types of leadership behaviors that will generally produce superior performance. We would expect that part of the variability of extant results predicting leadership performance could be explained in terms of methodological artifacts, hence validity generalization analyses may yield more accurate conclusions. However, our study was grounded in a strong theoretical and empirical rationale arguing for some consistency in explaining leadership perceptions. We know of no theory of leadership performance that is analogous to the social-cognitive perspective that guided this investigation. Despite this caution to generalizing conclusions, we believe the findings of the present study are practically important. Leadership perceptions are important in their own right, being a major component of the social fabric of many organizations. Being perceived as a leader allows one to exert greater influence in business or government, and leadership perceptions are particularly important in the political area (Foti, Fraser, & Lord, 1982). Leadership perceptions may also have important symbolic value (Pfeffer, 198 l) that fosters commitment to an organization on the part of followers. Our findings are also important from a methodological perspective. There is a close parallel between trait approaches to leadership perceptions and work on the general topic of using traits to predict behavior. For example, the extensive criticism of trait approaches to leadership noted earlier redirected the efforts of researchers toward contingency theories and the search for situational moderators. Similarly, work on personality traits has examined individual differences as moderators (Bem& Allen, 1974) and argued that person-situation interactions (Endler & Magnusson, 1976) are required to explain the variability in empirical findings. Extending this parallel would suggest that some of the variability in the relations between personality traits and behaviors may be due to the same type of methodological artifacts that this study has investigated in the leadership area. Correcting for sample size variations, for differences in range restriction, or for differences in reliability of predictors may be important in producing better estimates of population values and in estimating the amount of nonartifactual variance requiring substantive explanations. The validity generalization approach, along with other conceptual (Mischel, 1984) and methodological improvements (Rushton, Brainerd, & Pressley, 1983) may well yield a more optimistic conclusion concerning the ability of personality traits to predict behaviors.

Results of Validity Generalization Analyses Adding Studies After 1959 to Mann's Data Base
Trait Measure N Number of results Percentage of explained variance Meant True r I 2239 18 62 .368 .503 .095 .317 A 1085 11 40 .168 .238 .123 -.005 D 1660 11 80 .093 .133 .042 .051 E 1701 13 21 .185 .261 .184 -.100

SD

Lower bound

sion-introversion. Findings for masculinity-femininity and conservatism would not change from Table 3 results because no new studies using these variables were found.

Note. I = intelligence; A = adjustment; D = dominance; E = extrover-

personality traits such as masculinity-femininity should be fairly important in forming leadership perceptions. The major difference between predictions from Table 1 and our validity generalization findings deserves comment. On the basis of work on implicit leadership theories, one would expect extroversion to be predictive on leadership perceptions. Yet quantitative findings from our review indicated nonsignificant and very inconsistent findings for this trait. This discrepancy could be explained by differences in the way scientists and lay people define the construct. Work on ILTs indicates that verbal skills and outgoing personalities are generally characteristic of leaders, and other empirical findings indicate that verbal activity is very strongly predictive of leadership perceptions (Stein & Heller, 1979). However, Mann used a very broad definition of this construct, which pulls together the dimensions of sociability, surgency, and cyclothymia versus schizothymia (Mann, 1959, p. 243). It is unlikely that lay people (or current researchers) would define extroversion-introversion this way. The fact that estimated population correlations were still much less than 1.0 probably reflects three factors. First, correlations would be attenuated, since the perceived traits of leaders and the measured traits of potential leaders would not agree perfectly. We have argued that traits as perceived by others would affect leadership emergence, but the empirical results we examined used standardized tests of potential leaders' traits. Second, it has recently been argued that leadership and other person categories are probably "fuzzy" (Cantor & Mischel, 1979; Lord et al., 1984), being based on the principle of family resemblance rather than being defined on critical features required by all category members. Hence, a particular prototypical trait such as intelligence may not be present, but a person could still be perceived as a leader if he or she possessed other traits that were prototypical of leaders. Third, leadership, like other person categories, is probably defined in terms of a prototype involving several traits. Hence, bivariate relationships should not be expected to yield correlations close to one. Multivariate prediction of leadership perceptions might be much better. This possibility should be investigated by future researchers.

TRAITS AND LEADERSHIP PERCEPTIONS

409

References
Alexander, R. A., Carson, K. E, Alliger, G. M., & Cronshaw, S. (1984). Empirical distributions of range restricted SDX. Paper submitted for publication. Bern, D. J., & Allen, A. (1974). On predicting some of the people some of the time: The search for eross-situational consistencies in behavior. PsychologicalReview, 81, 506-520. Cantor, N., & Mischel, W. (1979). Prototypes in person percoption. In L. Berkowitz (Ed.), Advances in experimental social psychology (Vol. 12, pp. 3-52). New York: Academic Press. Eden, D., & Leviatan, U. (1975). Implicit leadership theory as a determinant of the factor structure underlyingsupervisory behavior scales. Journal of Applied Psychology,, 60, 736-741. Endler, N. S., & Magnusson, D. (1976). Toward an interactional psychology of personality. PsychologicalBulletin, 83, 956-974. Foil, R. J., Fraser, S. L., & Lord, R. G. (1982). Effects of leadership labels and prototypes on perceptions of political leaders. Journal of Applied Psychology,, 67, 326-333. Fraser, S. L., & Lord, R. G. (1983). The effect of stimulus prototypicality on leadershipperceptions and behavioral ratings. Manuscript submitted for publication. Heslin, R. (1964). Predicting group task effectiveness from member characteristics. PsychologicalBulletin, 62, 248-256. Hollander, E. P., & Julian, J. W. (1969). Contemporary trends in the analysis of leadership processes. PsychologicalBulletin, 71, 387-397. Hunter, J. E., Sclimidt, F. L., & Jackson, G. B. (1982). Meta-analysis: Cumulating researchfindings across studies. Beverly Hills: Sage. Kenny, D. A., & Zaccaro, S. J. (1983). An estimate of variance due to traits in leadership. Journal ofApplied Psychology,, 68, 678-685. Landy, F. J. (1985). Psychology of work behavior. Homewood, IL: Dorsey Press. Lord, R. G. (1985). Social information processing and behavioral measurement: Application to leadership measurement. In B. M. Staw & L. L. Cummings (Eds.), Research in Organizational Behavior (Vol. 7, pp. 87-128), Greenwich, C~. JAI Press. Lord, R. G., Foil, R. J., & De Vader, C. L. (1984). A test of leadership categorization theory: Internal structure, information processin~ and leadership perceptions. Organizational Behavior and Human Performance, 34, 343-378. Lord, R. G., Foil, R. J., & Phillips, J. S. (1982). A theory of leadership categorization, In J. G. Hunt, U. Sekaran, & C. Schriesheim (Eds.), Leadership: Beyond establishment views (pp. 104-121). Carbondale: Southern Illinois University Press. Mann, R. D. (1959). A review of the relationships between personality and performance in small groups. Psychological Bulletin, 56, 241270. Mischel, W. ( 1973). Toward a cognitive social learning reconceptualization of personality. PsychologicalReview, 80, 252-283. Miscbel, W. (1984). Convergences and challenges in the search for consistency, American Psychologist, 39, 351-364. Mitchell, T. R. (1982). People in organizations: An introduction to organizational behavior. New York: McGraw-Hill. Mervis, C. B., & Rosch, E. ( 1981). Categorization of natural objects. In

M. R. Rosenzweig & L. W. Porter (Eds.). AnnualReview of Psychology (pp. 89-115). Palo Alto, CA: Annual Reviews. Muchinsky, E M. (1983). Psychology applied to work. Homewood, IL: Dorsey Press. Pearlman, K., Schmidt, E L., & Hunter, J. E. (1980). Validity generalization results for tests used to predict job proficiency and training success in clerical occupations. Journal of Applied Psychology, 65, 373--406. Pfeffer, J. (1981). Management as symbolic action: The creation and maintenance of organizational paradigms. In L. L. Cummings & B. M. Staw (Eds.), Research in organizational behavior (Vol. 3, pp. 1-52). Greenwich, CT: JAI Press. Phillips, J. S. (1984). The accuracy of leadership ratings: A cognitive categorization perspective. Organizational Behavior and Human Performance, 33, 125-138. Phillips, J. S., & Lord, R. G. (1981). Causal attributions and perceptions of leadership. Organizational Behavior and Human Performance,28, 143-163. Phillips, J. S., & Lord, R. G. (1982). Schematic information processing and perceptions of leadership in problem-solving groups. Journal of Applied Psychology, 67, 486-492. Rush, M. C., Thomas, J. C., & Lord, R. G. (1977). Implicit leadership theory: A potential threat to the internal validity of leader behavior questionnaires. Organizational Behavior and Human Performance, 20, 93-110. Rushton, J. P., Brainerd, C. J., & Pressley, M. (1983). Behavioral development and construct validity: The principle of aggregation, Psychological Bulletin, 94, 18-38. Schmidt, E L., Gust-Rosenber~ I., & Hunte~ J. E. (1980). Validity generalization: Results for computer programmers. Journal of Applied Psychology,, 65, 643-661. Schmidt, E L., & Hunter, J. E (1977). Development of a general soluilon to the problem of validity generalization. Journal ofApplied Psychology, 62, 529-540. Schmidt, E L., & Hunter, J. E (1981). Employment testing. American Psychologist, 36, 1128-1137. Stein, R. T., & Heller, T. (1979). An empirical analysis of the correlations between leadership status and participation rates in the literature. Journal of Personality and Social Psychology,, 37, 1993-2002. Stngdill, R. M. (1948). Personal factors associated with leadership; a survey of the literature. Journal of Psychology, 25, 35-71. Stngdill, R. M. (1974). Handbook of leadership. New York: The Free Press. Weiss, H. M., & Adler, S. (1981 ). Cognitive complexity and the structure of implicit leadership theories. Journal ofApplied Psychology,, 66, 6978. Weiss, H. M., & Adler, S. (1984). Personality and organizational behavior. In B. M. Staw & L. L. Cummings (Eds.), Research in Organizational Behavior (Vol. 6, pp. 1-50). Greenwich, CT: JAI Press. Winter, L., & Ulernan, J. S. (1984). When are social judgments made? Evidence for the spontaneousness of trait inferences. Journal of Personality and Social Psychology,, 47, 237-252. Yukl, G. A. ( 1981). Leadership in organizations. Englewood Cliffs, NJ: Prentice-Hall.

(Appendix follows on next page)

410

R. LORD, C. DE VADER, AND G. ALLIGER

Appendix
References for Studies I n c l u d e d i n the Validity G e n e r a l i z a t i o n Analyses Arbous, A. G., & Maree, J. (1951). Contribution of two group discussion techniques to a validated test battery. Occupational Psychology, 25, 73-89. Ashour, A. S., & England, G. (1972). Subordinate's assigned level of discretion as a function of leader's personality and situational variables. Journal of Applied Psycholog)z, 56, 120-123. Bass, B. M., McGehee, C. R., Hawkins, W. C., Young, P. C., & Gebel, A. S. (1953). Personality variables related to leader-lessgroup discussion behavior.Journal of Abnormal Social Psychology,, 48, 120-128. Bass, B, M., & Wurster, C. R. (1953). Effects of company rank on LGD performance of oil refinery supervisors. Journal of Applied Psycholog3z,37, 100-104. Bass, B. M., Wurster,C. R., Doll, P. A., &Clair, D. J. (1953). Situational and personality factors in leadership among sorority women. Psychological Monographs, 67( 16, Whole No. 366). Carter; L. E, & Nixon, M. (1949). Ability, perceptual, personality, and interest factors associated with different criteria of leadership. Journal of Psychology, 27, 377-388. Dexter, E. S., & Stein, B. (1955). The measurement of leadership in white and Negro women students. Journal of Abnormal Social Psycholo~,, 51, 219-221. Flemming, E. G. (1935). A factor analysis of the personality of high school leaders. Journal of Applied Psycholog~, 19, 596-605. Gordon, L. V. (1952). Personal factors in leadership. Journal of Social Psychology,, 36, 245-248. Gowan, J. C. (1955). Relationship between leadership and personality factors. Journal of Education Research. Green, G. H. (1948). Insight and group adjustment. Journal of Abnorreal Social Psychology, 43, 49-61. Green, N. G. (1950). Verbalintelligence and effectivenessof participation in group discussion.Journal of Educational Psychology,, 41, 440445. Greenwood, G. M., & McNamara, W. J. (1969). Leadership styles of structure and consideration and managerial effectiveness.Personnel Psychology, 22, 141-152. Heisler, W. J., & Gemmill, G. R. (I 977). Maehiavellianism,job satisfaztion, job strain, and upward mobility: Some cross-o~anizational evidence. Psychological Reports, 41, 592-594. Holtzruan, W. H. (1952). Adjustment and leadership: A study of the Rorschach test. Journal of Social Psychology, 36, 179-189. Howell, C. E (1942). Measurement of leadership. Sociomet~ 5, 163168. Lamont, L. M., & Lundstrom, W. J. (1977). Identifying successful industrial salesmen by personalityand personal characteristics.Journal of Marketing Research, 14, 51%-529. MeCuen, T L. (1929). Leadership and intelligence.Education, 50, 8995. Mitchel, J. O. (1975). Assessment center validity: A longitudinal study.

Journal of Applied Psychology, 60, 573- 579.


Page, D. P. (1935). Measurement and prediction of leadership. The

American Journal of Sociology, 41, 31-43.


Rowland, K. M., & Scott, W. E., Jr. (1968). Psychologicalattributes of effectiveleadershipin a formal organization. PersonnelPsychological, 21, 365-377. Turner, C. E, & Martinez, D. C. (1977). Socioeconomic achievement and the machiavellianpersonality.Sociomet~ 40, 325-36. Wurster, C. R., & Bass, B. M. (1953). Situational tests: IV. Validity of leaderless group discussion among strangers. Educational and Psychological Measurement, 13, 122-132. Zeleny, L. D. (1939). Characteristics of group leaders. Sociology and SociaI Research, 24, 140-149. Received May 6, 1985 Revision received September 12, 1985

Das könnte Ihnen auch gefallen