Sunteți pe pagina 1din 11

Journal of Personality and Social Psychology 1990, Vol. 59, No.

5, 960-970

Copyright 1990 by the American Psychological Association, Inc. O022-3514/90/$00.75

Gender Differences in the Accuracy of Self-Evaluations of Performance


Sylvia Beyer
University of Oregon
Research has shown that gender differences in self-perceptions exist. However, interpretational ambiguities make it impossible to determine whether these gender differences are due to the operation of biases. The present research investigated whether gender differences in biased selfperceptions exist by assessing the accuracy of posttask self-evaluations of performance. In accordance with self-consistency theory, it was hypothesized that Ss' expectancies affect their posttask self-evaluations. For example, men who generally have high expectancies on masculine-gendertyped tasks were hypothesized to evidence overly positive self-evaluations. Women, who generally hold low expectancies on masculine tasks, were hypothesized to hold overly negative self-evaluations. The results confirmed that self-consistency tendencies can partially explain self-perception biases. The implications of these findings for women's achievement behavior and self-confidence are discussed.

It is a well-established finding that women have lower expectancies of success than men in many areas of achievement (Crandall, 1969; Erkut, 1983; Gitelson, Petersen, & Tobin-Richards, 1982; Mura, 1987; Sleeper & Nigro, 1987). These low expectancies may be indicative of women's tendency to underestimate their abilities (Carr, Thomas, & Mednick, 1985; Crandall, 1969; Lenney, 1977). Research on causal attributions of performance also produced evidence for women's putative underestimation of abilities. Women tend to attribute success more externally (Feather, 1969; Meehan & Overton, 1986; Pasquella, Mednick, & Murray, 1981; Simon & Feather, 1973; Sohn, 1982; Viaene, 1979; Zuckerman, 1979) or more to effort than do men (LaNoue & Curtis,1985; Parsons, Meece, Adler, & Kaczala, 1982; Wiegers & Frieze, 1977). The implication is that by making external attributions for success, women do not take credit for their performancethereby denigrating their abilitieswhereas men stress the importance of their ability in achieving success. This has been interpreted as evidence for the operation of a self-enhancing bias in men and a self-derogatory bias in women (Bar-Tal & Frieze, 1977; Berg, Stephan, & Dodson, 1981; Erkut, 1983; Heilman & Kram, 1978; Levine, Gill-

Part of this research was presented at the 1989 joint meeting of the Western Psychological Association and the Rocky Mountain Psychological Association. This research was conducted in partial fulfillment of the requirements for the doctoral degree at the Department of Psychology, University of Oregon. This research was supported in part by a grant from the Center for the Study of Women in Society at the University of Oregon. I would like to thank Beverly Fagot, Shinobu Kitayama, and the anonymous reviewers for their thoughtful comments on drafts of this article and Robert Mauro and Sriram Natarajan for statistical advice. I would like to give special thanks to Edward Bowden for his valuable suggestions during all stages of this research. Correspondence concerning this article should be addressed to Sylvia Beyer, who is now at the Department of Psychology, University of Michigan, Flint, Michigan 48502. 960

man, & Reis, 1982; Levine, Reis, Sue, & Turner, 1976; Simon & Feather, 1973; Zuckerman, 1979). These findings suffer from an important interpretational problem. Terms such as under- and overestimation and bias imply that women and men hold inaccurate expectancies and causal attributions. To establish the existence of biases, it is therefore necessary to demonstrate that expectancies and attributions are in fact inaccurate. However, the accuracy of expectancies is difficult to assess because expectancies can affect performance (Battle, 1965; Feather, 1966; Meehan & Overton, 1986; Vollmer, 1986). For example, low expectancies can cause people to decrease effort so that they can legitimately attribute failure to lack of effort rather than to lack of ability. This selfhandicapping strategy results in poor performance (Baumgardner& Levy, 1988; Frankel&Snyder, 1978; A. Miller, 1985). Expectancies can thereby become self-fulfilling prophecies. In one sense, people using self-handicapping strategies may be considered accurate predictors of performance. However, by exerting little effort, their ability remains untested and may be greatly underestimated by low expectancies. In this sense, self handicappers are inaccurate. Because true ability cannot be determined, measuring the accuracy of expectancies is difficult. The accuracy of attributions is also difficult to establish. The true cause of a performance is usually indeterminable and may not even be the same for different people. It is not known how important ability, effort, task difficulty, luck, and sundry other factors are in determining performance on, say, an anagrams task. It is conceivable that women correctly identify the causes of performance as external. Alternatively, the true cause of performance may be different for men and women, so that, for example, women might try harder than men and therefore accurately rate effort as an important determining factor, whereas men may consider effort relatively less important. In both cases, it would be erroneous to consider women biased toward self-derogation. Thus, the demonstration that gender differences in causal attributions exist does not prove the existence of gender differences in biases.

ACCURACY OF SELF-EVALUATIONS

961

Recent research on depression and self-esteem has recognized the importance of assessing the accuracy of self-perceptions. Some theories of depression (Beck, 1976) and self-esteem (Fitch, 1970) presumed that the self-perceptions of depressives and people low in self-esteem were negatively distorted. However, when the accuracy of self-perceptions was assessed, depressives were found to be more accurate than nondepressives in their social self-evaluations (Lewinsohn, Mischel, Chaplin, & Barton, 1980), estimates of future success and failure (Alloy & Ahrens, 1987), and in assessments of the degree of control over external stimuli (Abramson & Alloy, 1981; Alloy & Abramson, 1979; Alloy, Abramson, & Viscusi, 1981). Similarly, people low in self-esteem had more accurate self-evaluations of performance than did people high in self-esteem (Shrauger & Kelly, 1988; Shrauger & Terbovic, 1976). Theories of depression, self-esteem, and women's self-perceptions would have to be revised if it could be empirically established that depressives, people low in self-esteem, and women are in fact accurate rather than inaccurate self-evaluators. An intriguing practical implication regarding therapeutic interventions with people from these groups would be that they should be taught to be inaccuratein essence, to overestimate themselves! Research has indeed shown that self-enhancing biases (positive illusions) rather than accurate self-perceptions are conducive to psychological health (Janoff-Bulman, 1989; Snyder, 1989; Taylor & Brown, 1988; Taylor, Collins, Skokan, & Aspinwall, 1989). Self-enhancing biases may engender hope for the future and thereby also result in improved motivation and task persistence (Abramson & Alloy, 1981; Greenwald, 1980). But before such recommendations can be made, it is imperative to empirically address the issue of gender differences in selfperception biases. The present experiments were designed to address this question by measuring the accuracy of posttask self-evaluations of performance. Self-evaluations often have to be arrived at in the absence of immediate, unambiguous feedback regarding the quality of performance and thus are vulnerable to biased interpretations of performance. People have to decide for themselves, for example, whether the decision to recommend psychological treatment for a client was correct, whether their latest essay really deserved the low grade assigned by the instructor, or whether a new manuscript is good enough to warrant submitting it for publication.l In the latter case, if one believes that the manuscript does not meet the journal's standards, he or she may decide not to submit it. Thereby, one precludes the possibility not only for receiving feedback but also for having the manuscript published. Thus, low self-evaluations can have undesirable consequences. There are several reasons to expect that women and men fall prey to different self-perception biases. First, women are socialized to be modest regarding academic achievements, whereas men are taught to be confident in this area (Berg et al, 1981; Gould & Slone, 1982; Phillips, 1987). Second, the societal stereotype is that women are less competent than men (Broverman, Vogel, Broverman, Clarkson, & Rosenkrantz, 1972; Deaux & Emswiller, 1974; Feather & Simon, 1975; FeldmanSummers & Kiesler, 1974). In fact, many parents have inaccurately low perceptions of their daughters' ability in such areas as

math. These perceptions eventually come to be shared by the daughters (Parsons, Adler, & Kaczala, 1982). Thus, girls learn from parents and society to underestimate their competence (i.e, to have low expectancies for success), whereas boys receive the message that they should be highly confident. According to self-consistency theory, "people interpret and judge their achievements and abilities in ways congruent with prior self-conceptions" (Jussim, Coleman, & Nassau, 1987, p. 95). Thus, a person's expectancy should bias how performance on a task is interpreted. This is especially true when there is some ambiguity regarding the quality of performance, such as, in the absence of feedback (Felson, 1981; Wells & Sweeney, 1986). Therefore, self-consistency should result in inaccurate self-evaluations except when expectancies coincide with performance. Self-consistent tendencies have been demonstrated in people with positive and negative self-schemas (Brown, 1986; Jussim et al, 1987; Markus, 1977; Shrauger, 1975; Swann, 1987; Swann, Griffin, Predmore, & Gaines, 1987; Swann & Hill, 1982, Swann & Read, 1981a, 1981b; Wells & Sweeney, 1986). Women are least and men most confident on masculinegender-typed tasks (Bridges, 1988; Deaux & Farris, 1977; Janman, 1987; Karabenick, Sweeney, & Penrose, 1983). Self-consistency theory predicts that men's high confidence on masculine tasks will positively bias their self-evaluations, resulting in overestimations of performance. Women's low expectancies on masculine tasks should bias their self-evaluations negatively, thereby resulting in inaccurately low self-evaluations. Because neither women nor men have inordinately high or low expectancies on feminine tasks (Deaux & Farris, 1977; Lenney, 1981; Stein, Pohly, & Mueller, 1971), their self-evaluations should be more moderate and accurate on such tasks. Thus, the gender difference in self-evaluation bias should be most pronounced for masculine tasks. The self-consistency hypothesis does not predict that actual performance will have no effect on self-evaluations. It is not contended that people completely ignore reality (performance). Instead, it is hypothesized that even though people take actual performance into account, their initial level of confidence (expectancy) has a biasing effect on self-evaluations. In summary, the present experiments tested the following hypotheses: (a) Gender differences in the accuracy of selfevaluations interact with the gender-typedness of the task, and (b) these gender differences can be explained in part by selfconsistency theory.

Experiment 1 Method
Overview. Subjects were presented with a feminine-gender-typed, a masculine-gender-typed, and three neutral-gender-typed tasks. Sub1 Even in cases in which feedback eventually becomes available, such as after receiving a C- on an essay, interpretation of that feedback is still necessary. Potentially, one can agree with the feedback, one can disagree with it by citing extraneous factors such as the unfairness of the grading, or one can continue to be unsure about the quality of one's performance. Thus, even when feedback is available, biases may operate.

962

SYLVIA BEYER

jects stated performance expectancies, performed the task, and then evaluated their own performance in the absence of performance feedback. Subjects. Subjects were 39 male and 34 female students enrolled in introductory psychology courses at the University ofOregon. Participation in the experiment was in partial fulfillment of a course requirement. Tasks. Five tasks of differing gender-typedness were presented to subjects. A quiz on political and sports figures (masculine-gendertyped task) and a quiz on stars and fashion (feminine-gender-typed task) each comprised 10 multiple-choice questions to be answered in 2 min. Three neutral tasks were used. One of the neutral tasks was a character-detection task, which was unfamiliar to most subjects. Subjects had 3 min to circle 100 special characters embedded in a page of nontarget characters. A second neutral task required the use of basic mathematical skills to answer questions framed in practical contexts (henceforth referred to as practiced questions). This task consisted of 6 multiple-choice problems to be solved in 9 min. A third neutral task contained 20five-letteranagrams to be solved in 4 min. Manipulation checks confirmed the gender-typedness assigned to thefivetasks (discussed later). The tasks were presented in two orders. Thus, a 2 (gender) X 5 (type of task) X 2 (task order) design was used. Type of task was a within-subjects factor. Procedure. A female experimenter tested subjects in mixed-gender groups ranging in size from 12 to 20 participants. Subjects were told that this was a study on the effect of different aspects of tasks on performance. The experimenter emphasized the anonymity of test results and the noncompetitiveness of the tasks. The experimenter provided the subjects with information on the number and type of questions, the amount of time available for the task, and an example of a typical question and its correct answer.2 Subjects indicated how many correct answers they expected to produce and rated on a 7-point scale how difficult they expected the task to be. After finishing the task, subjects estimated the number of correctly answered questions (selfevaluations) and reevaluated the difficulty of the task without receiving feedback regarding their actual performance. This procedure was repeated for all five tasks. After completion of the last task, subjects indicated on a 9-point gender-typedness-rating scale whether they thought that men or women perform better on each of thefivetasks. Subsequently, subjects were debriefed.

Results and Discussion

To facilitate comparison across tasks and experiments, the expectancy, performance, self-evaluation, and inaccuracy scores were transformed into percentage scores. The few significant task order effects that were obtained did not interact with gender and thus do not affect the conclusions. Because of space considerations, these effects are not reported.3 Manipulation checks. To assure that the tasks had been assigned the proper gender-typedness labels, several manipulation checks were used. A task was considered feminine- (masculine-) gender-typed (a) if there was a significant gender difference in performance favoring women (men), (b) if there was a corresponding gender difference in postperformance task difficulty ratings, and (c) if both genders rated the task as at least 6 (at most 4) on the gender-typedness scale, which ranged from 1 2 For the masculine task, the following example was provided as a (men do better) to 9fyvomendo better). Neutral-gender-typedness typical question: "Which two teams played in the last Super Bowl?" was assigned to a task (a) if there was no gender difference in Additional sample questions/problems may be obtained from Sylvia performance, (b) if there was no gender difference in postperBeyer. 3 formance task difficulty ratings, and (c) if both genders rated These results can be obtained from Sylvia Beyer.

the task between 4.1 and S.9 on the gender-typedness scale. Analyses of variance (ANOVAS), with gender as the independent variable, were performed on these dependent variables. Table 1 contains the means for the manipulation checks. Men outperformed women on the politics and sports task and thought that this task was easier than women did, F(l, 71) = 7.21, p < .009; F(l, 71) = 19.26, p < .0001, respectively. Both genders rated the task below 4 on the gender-typedness scale. Thus, the politics and sports task satisfied the criteria for a masculine task. Women outperformed men and thought that the star and fashion task was easier than men did, both Fs(l, 71) > 4.20, ps < .05. Both genders rated the task above 6 on the gender-typedness scale. Thus, the star and fashion task satisfied the criteria for feminine gender-typedness. There were no differences in performance or rated task difficulty on the character-detection task, both i%(l, 71) < 1. Men and women rated the task in the neutral-gender-typedness range. Thus this task satisfied the requirements for a neutral task. Men and women rated the practical questions as neutral and performed equally well, F(l, 71) < 1. Men rated the task as easier than women did, F(l, 71) = 4.37, p < .04. This task thus satisfied only two of the three criteria for neutral gender-typedness and may be considered neutral to slightly masculine gender-typed. Men and women rated the anagram task as neutral and did not differ in performance or task difficulty ratings, both R(l, 71) < 1.60, ps > .21, thereby satisfying all the criteria for neutral gender-typedness. Expectancies and self-evaluations. To assess the effect of task gender-typedness on expectancies and self-evaluations, two repeated measures analyses were performed. Type of task was a within-subjects factor, gender was a between-subjects factor. Significant and marginally significant interactions between task gender-typedness and subject gender were followed up with ANOVAS for each task. Table 1 contains the means (in percentages) for expectancies and self-evaluations. Type of task and subject gender interacted in their effect on expectancies, F(4,62) = 8.10, p < .001. Men were more confident (i.e., had higher expectancies) than women on the masculine task, F(l, 68) = 20.52, p < .0001, and on the character-detection task, F(l, 71) = 6.16, p < .02. There were no gender differences in expectancies for the practical questions, anagrams, or the feminine task. This is in line with previous research showing that women do not hold higher expectancies than men even on feminine tasks (Carr et al, 1985; Cole, King, & Newcomb, 1977; Deaux & Earns, 1977; Garland & Smith, 1981; Gitelson et al, 1982; McMahan, 1982). Type of task and gender also had an interactive effect on self-evaluations, F(4, 66) = 7.68, p < .001. As predicted, men had higher self-evaluations than women on the masculine task, F(l, 71) = 22.03, p < .0001, and also on the practical questions, F(l, 71) = 4.35, p < .05. No significant gender differences in

ACCURACY OF SELF-EVALUATIONS

963

Table 1 Means for Gender Typedness, Task Difficulty, Expectancies, Actual Performance, Self-Evaluations, and Inaccuracy of Self-Evaluations Gender of subjects Masculine task Women Men Feminine task Women Men Character detection task Women Men Practical questions task Women Men Anagram task Women Men Gender typedness* 3.4 3.1 6.7 6.3 5.7* 5.2* 4.7 4.5 5.5 5.2 Task difficulty" 4.3 3.1 3.0* 3.6* 3.1 3.1 4.6* 3.9 5.1 5.5 Expectancy0 49.7** 71.9" 69.7 66.1 78.4* 88.3* 71.7 77.2 56.1 59.4 Performance 65.0** 74.2** 73.2* 63.8* 86.1 82.4 50.5 53.5 40.0 35.5 Self-evaluation 53.2*** 73.3"* 72.6 65.9 81.3 83.2 61.3* 73.0* 41.5 37.2 Inaccuracy of self-evaluations'1 -11.8 b ** -0.9** -0.6 2.1 -4.8. 0.8 10.8b 19.5C 1.5. 1.7.

Note. Asterisk superscripts indicate significant (two-tailed-test) gender differences. Letter subscripts denote whether an inaccuracy score is significantly different (two-tailed test) from zero (accuracy). * As rated on a 9-point scale ranging from men do better (1) to women do better (9) on this task. b As rated on a 7-point scale ranging from very easy (1) to very difficult (7). c To make expectancy, performance, self-evaluation, and inaccuracy scores comparable across tasks, I transformed them into percentage scores. d Negative (positive) inaccuracy scores indicate under- (over-) estimation. * p < . 0 5 . * * p < . 0 1 . ***/><.0001.

,p<.05.

b p<.001.

c p<.0001.

self-evaluations appeared on the feminine task, F(l, 71) = 1.86, p > . 17, or on the remaining twd neutral tasks. Self-consistency hypothesis. The self-consistency hypothesis predicts that expectancies affect self-evaluations above and beyond the effects of actual performance. This was tested by regressingfor each taskperformance, expectancy, gender, and the Performance X Gender, Expectancy X Gender, Performance X Expectancy, and Performance X Expectancy X Gender interactions on self-evaluations. If expectancies account for a significant increase in explained variance over and above the variance explained by actual performance, self-consistency tendencies are operating. If subject gender can significantly increase the amount of explained variance in self-evaluations over and above actual performance and expectancies, this would indicate that performance and self-consistency tendencies cannot completely explain self-evaluations. Note that a test of the increase in variance accounted for (AR2) that is contributed by gender, when performance and expectancies are partialed out, is identical to an analysis of covariance (ANCOVA) with performance and expectancies as the covariates. A significant interaction between gender and performance or gender and expectancy would indicate that the amount of variance in self-evaluations explained by performance or expectancy differs for men and women. No significant Performance X Expectancy or Performance X Expectancy X Gender interactions were expected, and with one exception, none were found to be significant.4 The standardized regression coefficients and the amount of explained variance added by a variable to variance already explained by previously entered variables (AR2) are shown in Table 2. As expected, actual performance was a significant predictor

of subjects' self-evaluations for all five tasks, indicating that subjects do rely heavily on objective reality when evaluating performance, all Fs > 18.29, ps < .0001. Expectancies accounted for a significant proportion of variance on four of the five tasks, indicating that self-consistency effects are present when unambiguous feedback is not provided, all /%(1, 68) > 6.55, ps < .02. The anagram task was the sole case in which expectancies did not explain additional variance, F(l, 69) < 1. The reason for the absence of self-consistency effects for anagrams is probably that anagrams provide unambiguous feedback regarding performance. Most subjects know whether an unscrambled anagram represents a real word or not. Thus, when clear performance feedback is provided, self-consistent tendencies appear to be reduced. For the practical questions, the additional variance in selfevaluations explained by gender was significant, F(l, 66) = 8.15, p < .006. On the masculine task, the additional variance approached significance, F(l, 66) = 3.09, p < .09. When performance and expectancies were not controlled, the gender difference in self-evaluations was highly significant on these two tasks. With performance and expectancies controlled, the ef4 On the character-detection task, the interaction between performance, expectancy, and gender was significant, F(l, 63) = 9.17, p < .004. To understand the nature of this interaction, I trichotomized performance and expectancy scores. The self-evaluations of subjects with high, intermediate, or low performance and either high, intermediate, or low expectancies were compared. Men's self-evaluations were higher than women's. Thus, the triple interaction was ordinal, or order preserving, with respect to gender, the variable of interest in this study. For this reason, the triple interaction did not affect the conclusions.

964

SYLVIA BEYER

Table 2 Standardized Coefficients (Coeff) and Amount ofAdditional Variance Accounted for (AR2) Stepl performance Task Experiment 1 Masculine task Feminine task Character-detection task Practical-questions taskc Anagrams Experiment 2 Masculine task Feminine task Neutral task Coeff. .456 .502 .725 .245 .968 .555 .731 .754 AR
2

Step 2 expectancy Coeff. .425 .372 .243 .269 .000 .290 .205 .140 AR
2

Step 3 gender" Coeff. -.143 .000 -.074 -.581 -.003 -.201 .084 .030 AR .016* .000 .005 .033* .000 .029***** .006** .001
2

Gender" Coeff. -.487 .160 -.058 -.240 .089 -.609 .196 .021 AR2 .237***** .012 .000 .058** .008 .371***** .038*** .000

.463***** .451***** .470***** .292**** .931***** .725***** .716***** .615*****

.189***** .111***** .068*** .065*** .017 .054***** .015***** .001***

Note. All tests are two-tailed. " Women are coded 1; men are coded 0.b Performance and expectancy not controlled.c The Performance X Gender interaction was significant for this task. AR2 = .042**; standardized coefficient = .471. *p<.10. **p<.05. ***/><.01. */?<.001. *****p<.0001.

feet of gender was greatly reduced. Gender did not even approach significance after controlling performance and expectancy on the feminine, character detection, and anagram tasks, all Fs < 1. The last two columns of Table 2 compare the standardized coefficients and ARh of gender when performance and expectancy were statistically controlled versus not controlled. For the practical questions, the interaction between gender and performance was significant, F(l, 66) = 4.84, p < .04. Table 2 illustrates that the reason for this interaction is that the amount of variance in self-evaluations explained by performance was higher for women than men. Accuracy of self-evaluations. To assess the effect of type of task on the accuracy of self-evaluations, a repeated measures analysis was performed. Type of task was a within-subjects factor; gender was a between-subjects factor. Inaccuracy of selfevaluations was calculated by subtracting performance from self-evaluation scores (see Table 1). A marginally significant interaction between type of task and gender was found, F(4, 66) = 2.28, p < .07. To elucidate the nature of this interaction, repeated measures analyses were computed, using performance and self-evaluations as the repeated factors (which is similar to an analysis of difference scores) and gender as between-subjects factor. This analysis determines whether an inaccuracy score is significantly different from 0 (i., whether it represents over- or underestimation) and whether there exists a significant gender difference in accuracy. When significant gender differences in the accuracy of self-evaluations were found, ANCOVAS were performed with expectancies as the covariate and gender as the between-subjects variable. This analysis permitted a test of the self-consistency hypothesis that gender differences in the accuracy of self-evaluations would be mediated by expectancies. The gender difference in accuracy of self-evaluations was significant on the masculine task, F(l, 71) = 9.50, p < .003. On

this task, men's self-evaluations were accurate (i.e, their inaccuracy score was not significantly different from 0), F(l, 37) < 1, whereas women underestimated performance (i, their inaccuracy score was significantly different from 0), F(l, 32) = 15.42, p < .0004. However, the gender difference in accuracy of selfevaluations was no longer significant when expectancies were statistically controlled, F(l, 67) = 1.54, p > .22. The gender difference in accuracy was marginally significant on the character-detection task, F(l, 68) = 3.28, p < .08. Men's self-evaluations were accurate, F(l, 35) < 1, whereas women's were inaccurately low, F(l, 32) = 6.04, p < .02. Again, when expectancies were statistically controlled, the gender difference in accuracy of self-evaluations did not approach significance, F(l, 68) < 1. On the remaining three tasks, for which there had been no gender difference in expectancies, no gender differences in accuracy of self-evaluations were found: Women and men were equally accurate in evaluating performance on the feminine task, F(l, 69) < 1, and they equally overestimated performance on the practical questions, F(l, 71) = 2.04, p > .15, and the anagrams, F(l, 69) < 1. Thus, on the two tasks in which significant or marginally significant gender differences in accuracy of self-evaluations were found (the masculine task and the character-detection task), the gender difference became nonsignificant after I controlled for expectancies. On the three tasks on which no gender difference in the accuracy of self-evaluations was found, there had been no gender difference in expectancies. This suggests that the gender difference in accuracy of self-evaluations was mediated by expectancies (self-consistency). Men performed as well on the masculine task as women did on the feminine task. The performance of men on the feminine task was comparable to the performance of women on the masculine task. Thus, task difficulty could be ruled out as a possible confound.

ACCURACY OF SELF-EVALUATIONS

965

Experiment 2
In Experiment 1, several findings were of marginal significance. For example, it is unclear whether gender would account for a clearly significant proportion of variance in self-evaluations, over and above the effects of performance and expectancies, if power were increased. Because of the theoretical and practical implications of the findings, I decided to replicate Experiment 1 on a larger sample before drawing firm conclusions. Some methodological changes were made. Method
Subjects. Subjects were 85 male and 92 female students enrolled in introductory psychology courses at the University ofOregon. Participation in the experiment was in partial fulfillment of a course requirement. Tasks. To increase comparability of scores across tasks, the number of items on all tasks was increased to 40. The three tasks that were presented were of a higher difficulty level than in Experiment 1 to avoid a potential ceiling effect. The masculine and feminine tasks were a quiz on sportsfiguresand a quiz on movie and TV stars, respectively. Both tasks contained 40 multiple-choice questions to be answered in 8 min. The character-detection task was used as the neutral task. Subjects had 2 min to circle 40 special characters embedded in a list of nontarget characters. All six permutations of task orders were presented. A 2 (gender) X 3 (gender-typedness of task) X 6 (task order) design was used. Task gender-typedness was a within-subjects factor. Procedure. The procedure was similar to the one used in Experiment 1 except for the following additions. A rating scale assessed how important it was for each subject to do well on the task, to ensure that both genders were equally motivated. In addition to stating expectancies in quantitative form, subjects gave qualitative ratings of expectancies: "On the following scale please indicate how good you are at this type of task." The scale ranged from very bad (1) to very good(J). After performing the task, in addition to stating how many questions they believed to have answered correctly (quantitative self-evaluations), subjects rated how well they thought they had done (qualitative selfevaluations) and how well they expected to do on a similar task in the future.

Results and Discussion


To facilitate comparison with Experiment 1, expectancy, performance, self-evaluation, and inaccuracy scores were transformed into percentage scores. Manipulation checks. Men and women considered it equally important to do well on the masculine, feminine, and neutral tasks, all Fs(l", 175) < 1.62, ps > .21. The gender-typedness scale was omitted, so the manipulation check for gender-typedness was reduced to ensuring that on the feminine (masculine) task women (men) outperformed and rated the task difficulty lower than men (women). For the neutral task, no performance or rated task difficulty difference was expected. Manipulation checks are depicted in Table 3. Men performed better on the masculine task and thought that it was easier than women did, both Fs(l, 175) > 64.35, p < .0001, confirming that this task was masculine gender-typed. Women performed better on the feminine task and found it to be easier than did men, F(l, 175) = 8.06, p < .006; F(l, 175) = 14.55, p < .0002, respectively. Thus, the star quiz was feminine

gender-typed. The genders did not differ in performance, F(l, 175) < 1, or rated task difficulty, F(l, 174)= 2.44, p>. 11, on the character-detection task, confirming that the task was neutral. Expectancies, self-evaluations, and future expectancies. To assess the effect of task gender-typedness on expectancies, selfevaluations, and future expectancies, repeated measures analyses were performed. Gender-typedness of task was a withinsubjects factor; gender was a between-subjects factor. Significant interactions were followed up with ANOVAS for each task. Table 3 depicts the means (in percentages) for expectancies, self-evaluations, and future expectancies. Task gender typedness and gender interacted in their effect on expectancies, F(2,158) = 11.99, p < .001. As in Experiment 1, men had higher expectancies than women on the masculine task, F(\, 174) = 34.92, p < .0001, and the character-detection task, F(l, 172) = 8.58, p < .004. There was again no gender difference in expectancies for the feminine task, F(l, 167) = 1.22, p>.27. Task gender typedness and gender interacted in their effect on self-evaluations, F(2,166) = 51.33, p < .0001. As expected, men had higher self-evaluations than did women on the masculine task, F(l, 174) = 102.68, p < .0001, whereas women had higher self-evaluations than did men on the feminine task, F(l, 174) = 6.96, p < .009. On the neutral task, there was no significant gender difference in self-evaluations, F(l, 173) < 1. Subjects' ratings on a 7-point scaleregardinghow well they thought they had performed (qualitative self-evaluations) mirrored their self-evaluations.5 Gender-typedness of task and gender had an interactive effect on expectancies for future performance, F(2,162) = 32.67, p < .0001. Even when actual performance was statistically controlled, men expected to do better on future masculine and neutral tasks than women did, F(l, 172) = 12.08, p < .001; F(l, 171) = 7.35, p < .007, respectively. On the feminine task, women had higher future expectancies than men, F(l, 175) = 6.14, p<.02. Selfconsistency effects. For all three tasks, performance, expectancy, gender, and the interaction terms wereregressedon self-evaluations. The standardized regression coefficients and ARh are presented in Table 2. None of the interactions between gender and performance and between gender and expectancies were significant.6 As expected, actual performance was

These results can be obtained from Sylvia Beyer. However, on the masculine task, the Performance X Expectancy X Gender interaction was significant, F(l, 167) = 10.45, p < .002. To understand the nature of this interaction, performance and expectancy scores were trichotimized. The self-evaluations of subjects with high, intermediate, or low performance and either high, intermediate, or low expectancies were compared. This procedure indicated that men's self-evaluations were consistently higher than women's. Because the interaction was ordinal, it did not affect the conclusions. The interaction between performance and expectancy was significant for the feminine task, F(\, 161) = 4.65, p < .04, and neutral tasks, F(l, 165) = 6.11, p < .02. After I trichotimized the data, it became clear that the interaction was due to subjects in the high-performance conditions, in which expectancy did not increase self-evaluations. These results are most likely due to a ceiling effect.
6

966

SYLVIA BEYER

Table 3 Means for Task Difficulty, Expectancies, Performance, Self-Evaluations, Inaccuracy of Self-Evaluations, and Future Expectancies
Gender of subjects Masculine task Women Men Feminine task Women Men Neutral task Women Men Task difficulty* 5.3**** 3.6**** 3.3*** 4.1*** 4.3 4.6 Inaccuracy Selfof selfFuture evaluations evaluations0 expectancies'1 33**** 65**** 67* 58* 55 54 -5 b **** 0 -1 -5b 2.8**** 4.5**** 4.6* 4.1* 4.2 4.6

Expectancy 45**** 65**** 65 68 76** 82**

Performance 38**** 61**** 67** 59** 61 59

A ***

-6,,

Note. Asterisk superscripts indicate significant (two-tailed-test) gender differences. Letter subscripts denote whether an inaccuracy score is significantly different (two-tailed test) from zero (accuracy). 'As rated on a 7-point scale ranging from very easy (1) to very difficult (7). "To make expectancy, performance, self-evaluation, and inaccuracy scores comparable across tasks, I transformed then into percentage scores. c Negative (positive) inaccuracy scores indicate under- (over-) estimation. d As rated on a 7-point scale ranging from expect to do poorly (1) to expect to do very well (7). * p < . 0 5 . **p<.0\. ***p<.001. ****/><.0001. , p < . 0 1 . b p<.0001.

a significant predictor of subjects' self-evaluations on all three tasks, all Fs(l, 173) > 134.40, ps < .0001. Over and above the effect of performance, expectancies accounted for a significant proportion of variance on all three tasks, confirming that selfconsistency tendencies exist, all Fs(l, 167) > 8.00, p < .006. On the masculine and feminine tasks, gender accounted for a significant amount of variance in self-evaluations over and above the effects of performance and expectancies, F{\, 171) = 26.11, p < .0001 for the masculine task; F(l, 164) = 4.01, p < .05, for the feminine task. This indicates that when the effects of performance and expectancy were statistically controlled, the effect of gender on self-evaluations was still significant on the masculine and feminine tasks. Table 2 compares the standardized coefficients and additional variance of gender when expectancy and performance were statistically controlled versus not controlled. On the neutral task, gender did not even approach significance when performance and expectancy were statistically controlled, F(l, 168) < 1. Accuracy of self-evaluations. To assess the effect of task gender typedness on the accuracy of self-evaluations, a repeated measures analysis was performed. Task gender-typedness was a within-subjects factor; gender was a between-subjects factor. The Task Gender Typedness X Gender interaction was significant, F(2,171) = 10.66, p < .0001. Repeated measures analyses were computed, using performance and selfevaluations as the repeated factors, to elucidate whether an inaccuracy score would be significantly different from 0 and whether there would exist a significant gender difference in accuracy. Inaccuracy scores can be found in Table 3. When significant gender differences in the accuracy of self-evaluations were found, ANCOVAS were performed with expectancies as the covariate and gender as the between-subjects variable. As in Experiment 1, men and women evaluated performance accurately on the feminine task, F(\, 164) < 1. Men and women

equally underestimated performance on the neutral task, F{\, 163) = 30.90, p < .0001. This latterfindingwas somewhat unexpected. Both genders had had very high expectancies for this task and therefore should have overestimated their performance, if self-consistent tendencies were operating. This neutral task may represent a boundary condition for self-consistency effects. Conceivably, subjects' overly optimistic confidence backfired by making subjects think they failed. The frustration with an unexpectedly low performance may have then manifested itself in underestimations. The Gender Typedness of Task X Gender interaction was due to the masculine task, for which the gender difference in accuracy of self-evaluations was significant, F(l, 174) = 21.21, p < .0001. As hypothesized, men overestimated, F(\, 78) = 9.79, p < .003, whereas women underestimated, F(l, 86) = 15.81, p < .0001, performance on this task. When expectancies were statistically controlled, the gender difference in accuracy of selfevaluations was reduced but still significant, F(\, 172) = 8.29, p < .004. This indicates that self-consistency can partially, but not entirely, explain the gender difference in accuracy of selfevaluations. General Discussion Across both experiments, men overestimated performance on three (one masculine and two neutral) tasks, were accurate on four (one neutral, one masculine, and both feminine) tasks, and underestimated performance on only one (neutral) task. In contrast, women overestimated only twice (two neutral tasks), were accurate on both feminine tasks, and underestimated on four (both masculine and both character detection) tasks.7 Lenney (1981) found that both genders equally overestimated performance on verbal ability (feminine gender typed?) and spatial-me7

ACCURACY OF SELF-EVALUATIONS

967

Thus, men tend either to be accurate or to overestimate, whereas women tend either to be accurate or to underestimate. However, when expectancies were statistically controlled, these gender differences in the accuracy of self-evaluations became nonsignificant on most tasks. Thus, gender differences in the accuracy of self-evaluations are mediated by expectancies. On the masculine task in Experiment 2, the gender difference in accuracy was greatly reduced but still significant when expectancies were statistically controlled. This suggests that the gender difference in accuracy of self-evaluations was partially mediated by expectancies but that another as yet unidentified mediating variable must exist (cf. Judd & Kenny, 1981). As expected, actual performance had a strong effect, accounting for the majority of variance in self-evaluations. This indicates that subjects do base their self-evaluations to a great extent on reality (actual performance). Interestingly, the proportion of variance explained by performance was greater in Experiment 2. This probably indicates that making tasks longer enables subjects to get a better feel for their performance, which concomitantly may decrease the effect of self-consistency (cf. additional accounted for variance for performance and expectancy in Experiments 1 and 2). Still, even when the effect of performance was partialed out, expectancies had a significant effect on self-evaluations in both experiments. Thus, men's high confidence positively biased their self-evaluations. On the feminine tasks, in which initial confidence was somewhat lower, men evaluated performance more accurately. The opposite pattern was obtained for women. On masculine tasks, women's low expectancies negatively biased their self-evaluations. On feminine tasks, women had realistic expectancies and consequently also interpreted performance accurately. Men's overestimation on the masculine task in Experiment 2 appears to be inconsistent with their accurate evaluations in Experiment 1. But men's expectancies on the masculine task in Experiment 1 were not inordinately high; therefore, overestimation would not be expected. In Experiment 2, in which expectancies were higher than performance, men did overestimate. There may be a limit to the application of self-consistency theory to gender differences in accuracy of self-evaluations, because it is unlikely that a very large discrepancy between expectancies and actual performance (as for the neutral task in Experiment 2) can be altogether ignored by an individual. It could be argued that the observed self-consistency tendencies increase accuracy; if expectancies are positively correlated with performance, any effect of expectancies on self-evaluations increases the correlation between performance and selfchanical (masculine gender typed?) subtests of an intelligence test. This finding seems inconsistent with the present findings of women's underestimation on masculine tasks. However, in Lenney's (1981) study, the gender typedness of these subtests was not established. In fact, she found no gender difference in performance, therefore the subtests may have been of neutral gender typedness. In addition, subjects' expectancies were not assessed. According to self-consistency theory, if both genders had high expectancies for these subtests, we would expect subjects to overestimate their performance. Because of these differences in her design and the present one, it is not clear whether her results are actually inconsistent with the present findings.

evaluations. According to this interpretation, accuracy is defined as the correlation between performance and self-evaluations. This interpretation addresses the legitimate question of whether high- (low-) performance people also have high (low) self-evaluations. Because the determination of whether men and women over- or underestimate their performance was of primary interest, accuracy is defined in this article as the discrepancy between performance and self-evaluations. The correlational definition of accuracy cannot address this particular question. The hypothetical case of perfect positive correlations between performance and self-evaluations for both men and women illustrates this point. The correlational definition would consider men and women equally accurate. However, this pattern of correlations could be obtained in a number of waysfor example, if all men equally overestimated performance, whereas all women accurately evaluated performance. Thus, the correlational approach to accuracy can show whether a relation between performance and self-evaluations exists, but it cannot show whether self-evaluations are accurate in absolute termsnamely, are congruent with performance or reflect over- or underestimations. In addition to performance and expectancies, gender accounted for some of the variance in self-evaluationsalbeit a small proportion never topping 4%on the masculine tasks, the practical questions, and to a smaller degree on the second feminine task. Unfortunately, exactly what it is about gender that predicts self-evaluations cannot be determined from the present research. Gender is a rather crude division and "serves as only a gross marker in predicting individual differences in behavior" (Deaux, 1984, p. 108). What is known, however, is that to explain the effect of gender, one has to look at variables beyond performance and expectancies. A fruitful next step would be to look for moderating variables that can elucidate what distinguishes people who have low expectancies and underestimate performance from people who have high expectancies and overestimate performance. An intriguing lead comes from the literature on learned helplessness. Helpless children underestimate performance, whereas mastery-oriented children are accurate in their perceptions of success (Diener & Dweck, 1980). Girls are more likely to be learned helpless than boys (Dweck & Bush, 1976; Dweck, Davidson, Nelson, & Enna, 1978; Dweck, Goetz, & Strauss, 1980; I. W Miller & Norman, 1979). One may speculate that women who underestimate performance on masculine tasks may be adopting a learned helpless pattern. There is a possible alternative explanation for women's low self-evaluations on masculine and unfamiliar tasks. Women might try to appear modest by publicly stating low self-evaluations even though their real self-evaluations are higher (Berg et al, 1981; Gould & Slone, 1982). This explanation cannot account for the fact that women's underestimation was a task-specificrather than generalizedphenomenon, present only in those tasks for which they had initially low confidence (masculine and unfamiliar neutral tasks). Furthermore, it has been found that women are modest only when making public statements or in anticipation that their statements will be revealed to other subjects (Berg et al, 1981; Gould & Slone, 1982). Because it was emphasized that the task results were anonymous

968

SYLVIA BEYER Alloy, L. B, & Abramson, L. Y. (1979). Judgment of contingency in depressed and nondepressed students: Sadder but wiser? Journal of Experimental Psychology: General, 108, 441-485. Alloy, L. B, Abramson, L. Y, & Viscusi, D. (1981). Induced mood and the illusion of control. Journal ofPersonality and Social Psychology, 41,1129-1140. Alloy, L. B, & Ahrens, A. H. (1987). Depression and pessimism for the future: Biased use of statistically relevant information in predictions for self versus others. Journal of Personality and Social Psychology, 52, 366-378. Bar-Tal, D , & Frieze, I. H. (1977). Achievement motivation for males and females as a determinant of attributions for success and failure. Sex Roles, 3,301-113. Battle, E. S. (1965). Motivational determinants of academic task persistence. Journal of Personality and Social Psychology, 2, 209-218. Baumgardner, A. H, & Levy, P. E. (1988). Role of self-esteem in perceptions of ability and effort: Illogic or insight? Personality and Social Psychology Bulletin, 14,429-438. Beck, A. T. (1976). Cognitive therapy and the emotional disorders. New York: International Universities Press. Berg, J. H., Stephan, W G, & Dodson, M. (1981). Attributional modesty in women. Psychology of Women Quarterly, 5, 711-727. Bridges, J. S. (1988). Sex differences in occupational performance expectations. Psychology of Women Quarterly, 12, 75-90. Broverman, I. K., Vogel, S. R, Brovennan, D. M, Clarkson, F. E, & Rosenkrantz, P. S. (1972). Sex-role stereotypes: A current appraisal. Journal of Social Issues, 28, 59-78. Brown, J. D. (1986). Evaluations of self and others: Self-enhancement biases in social judgments. Social Cognition, 4,353-376. Carr, P. G, Thomas, V G, & Mednick, M. T. (1985). Evaluation of sex-typed tasks by Black men and women. Sex Roles, 13, 311-316. Cole, D, King, K., & Newcomb, A. (1977). Grade expectations as a function of sex, academic discipline, and sex of instructor. Psychology of Women Quarterly, 1, 380-385. Crandall, V. C. (1969). Sex differences in expectancy of intellectual and academic reinforcement. In C. P. Smith (Ed.), Achievement-related motives in children. New York: Russell Sage Foundation. Deaux, K. (1984). From individual differences to social categories: Analysis of a decade's research on gender. American Psychologist, 39,105-116. Deaux, K, & Emswiller, T. (1974). Explanation of successful performance on sex-linked tasks. Journal of Personality and Social Psychology, 29, 80-85. Deaux, K, & Farris, E. (1977). Attributing causes for oneTs own performance: The effects of sex, norms, and outcome. Journal of Research in Personality, 11, 59-72. Diener, C. I, & Dweck, C. S. (1980). An analysis of learned helplessness: II. The processing of success. Journal ofPersonality and Social Psychology, 39, 940-952. Dweck, C. S., & Bush, E. S. (1976). Sex differences in learned helplessness: I. Differential debilitation with peer and adult evaluators. Developmental Psychology, 12,147-156. Dweck, C. S, Davidson, W, Nelson, S, & Enna, B. (1978). Sex differences in learned helplessness: II. The contingencies of evaluative feedback in the classroom and III. An experimental analysis. Developmental Psychology, 14, 268-276. Dweck, C. S, Goetz, T. E, & Strauss, N. L. (1980). Sex differences in learned helplessness: IV An experimental and naturalistic study of failure generalization and its mediators. Journal of Personality and Social Psychology, 38, 441-452. Erkut, S. (1983). Exploring sex differences in expectancy, attribution, and academic achievement. Sex Roles, 9, 217-231.

and self-evaluations never had to be publicly stated, it seems unlikely that women's biased self-evaluations were due to a modesty effect. Women's inaccurately low self-evaluations on masculine and unfamiliar neutral tasks may have potentially damaging consequences. As Experiment 2 has shown, on masculine and neutral tasks women underestimated performance and subsequently stated low expectancies for similar future tasks. Women's inaccurately low assessment of their performance on such tasks might have led to their low future expectancies. This may at least partially account for the underrepresentation of women in certain male-dominated areas; people with low expectancies for future tasks do not usually seek out opportunities for performing similar tasks (Weiner et al, 1972). For example, despite adequate past performance, women with low future expectancies for their math grades have been shown to choose not to enroll in advanced math courses (Lantz & Smith, 1981). To put it even more bluntly, a woman's misperceptions regarding her competence may lead to low expectancies for future performances and dissuade her from pursuing a career in certain masculine-gender-typed domains, whereas a man of equal ability might not doubt his competence for a minute.8 Women's misperceptions are probably more likely to occur when performance feedback is unavailable or ambiguous. However, as mentioned earlier, even in the face of clear performance feedback such as school grades, there is room for biased interpretation. For example, a high grade in math would probably not lead to positive self-evaluations if it were attributed to luck or easiness of the test, which are all too common attributional patterns in women (Feather, 1969; Meehan & Overton, 1986; Pasquella et al., 1981; Simon & Feather, 1973; Sohn, 1982; Zuckerman, 1979). The present experiments have shown that initial expectancies affect the accuracy of self-evaluations. Because of the serious implications of underestimations for self-confidence and psychological health (cf. Taylor et al, 1989) more attention should be devoted to the investigation of gender differences in the accuracy of self-evaluations. Such research will not only elucidate the underlying processes of self-evaluation biases and therefore be of theoretical interest but will also be of practical value by suggesting ways of eliminating women's underestimations of performance.
8 Shouldn't women, who, as just discussed, often misjudge their math abilities, also underestimate their performance on the practical questions in Experiment 1 because this task required mathematical skills? Not necessarily. The practical questions required the use of very basic math concepts (multiplication, division, calculation of percentages) that most college students regardless of gender are familiar with. Subjects were aware of this fact because they had been presented with a sample question. Thus, this task was different from the typical situation in which women underestimate their math abilitiesnamely, a math class in which new and difficult concepts are introduced.

References
Abramson, L. Y, & Alloy, L. B. (1981). Depression, nondepression, and cognitive illusions: A reply to Schwartz. Journal of Experimental Psychology: General, 110, 435-447.

ACCURACY OF SELF-EVALUATIONS Feather, N. T. (1966). Effects of prior success and failure on expectations of success and subsequent performance. Journal of Personality and Social Psychology, 3, 287-298. Feather, N. T. (1969). Attribution of responsibility and valence of success and failure in relation to initial confidence and perceived locus of control. Journal of Personality and Social Psychology, 13,129-144. Feather, N. X, & Simon, J. G. (197S). Reactions to male and female success and failure in sex-linked occupations: Impressions of personality, causal attributions, and perceived likelihood of different consequences. Journal of Personality and Social Psychology, 31, 20-31. Feldman-Summers, S, & Kiesler, S. B. (1974). Those who are Number Two try harder. Journal of Personality and Social Psychology, 30, 846-855. Felson, R. B. (1981). Ambiguity and the self-concept. Social Psychology Quarterly, 44, 64-69. Fitch, G. (1970). Effects of self-esteem, perceived performance, and choice on causal attributions. Journal of Personality and Social Psychology, 76,311-315. Frankel, A , & Snyder, M. L. (1978). Poor performance following unsolvable problems: Learned helplessness or egotism? Journal ofPersonality and Social Psychology, 36,1415-1423. Garland, H, & Smith, G. B. (1981). Occupational achievement motivation as a function of biological sex, sex-linked personality, and occupation stereotype. Psychology of Women Quarterly, 5, 568-585. Gitelson, I. B, Petersen, A. C, &Tobin-Richards, M. H. (1982). Adolescents' expectancies of success, self-evaluations, and attributions about performance on spatial and verbal tasks. Sex Roles, 8, 4 1 1 419. Gould, R. J, & Slone, C. G. (1982). The "feminine modesty" effect. A self-presentational interpretation of sex differences in causal attribution. Personality and Social Psychology Bulletin, 8, 477-485. Greenwald, A. G. (1980). The totalitarian ego: Fabrication and revision of personal history. American Psychologist, 35, 603-618. Heilman, M. E, & Kram, K. E. (1978). Self-derogating behavior in womenFixed or flexible: The effects of co-workert sex. Organizational Behavior and Human Performance, 22, 497-507. Janman, K. (1987). Achievement motivation theory and occupational choice. European Journal of Social Psychology, 17, 327-346. Janoff-Bulman, R. (1989). The benefits of illusions, the threat of disillusionment, and the limitations of inaccuracy. Journal of Social and Clinical Psychology, 8,158-175. Judd, C. M, & Kenny, D. A. (1981). Process analysis: Estimating mediation in treatment evaluations. Evaluation Review 5, 602-619. Jussim, L, Coleman, L, & Nassau, S. (1987). The influence of self-esteem on perceptions of performance and feedback. Social Psychology Quarterly 50, 95-99. Karabenick, S. A , Sweeney, C, & Penrose, G. (1983). Preferences for skill versus chance-determined activities: The influence of gender and task sex-typing. Journal of Research in Personality, 17,125-142. LaNoue, J. B, & Curtis, R. C. (1985). Improving women's performance in mixed-sex situations by effort attributions. Psychology of Women Quarterly, 9, 337-356. Lantz, A. E, & Smith, G. P. (1981). Factors influencing the choice of nonrequired mathematics courses. Journal of Educational Psychology, 73, 825-837. Lenney, E. (1977). Women's self-confidence in achievement settings. Psychological Bulletin, 84,1-13. Lenney, E. (1981). What's fine for the gander isn't always good for the goose: Sex differences in self-confidence as a function of ability area and comparison with others. Sex Roles, 7, 905-923. Levine, R, Gillman, M.-J, & Reis, H. (1982). Individual differences for sex differences in achievement attributions? Sex Roles, 8, 455466.

969

Levine, R, Reis, H. T, Sue, E , & Turner, G. (1976). Fear of failure in males: A more salient factor than fear of success in females? Sex Roles, 2, 389-398. Lewinsohn, P. M, Mischel, W, Chaplin, W C, & Barton, R. (1980). Social competence and depression: The role of illusory self-perceptions. Journal of Abnormal Psychology, 89, 203-212. Markus, H. (1977). Self-schemata and processing information about the self. Journal of Personality and Social Psychology 35,63-78. McMahan, I. D. (1982). Expectancy of success on sex-linked tasks. Sex Roles, 8, 949-958. Meehan, A. M, & Overton, W F. (1986). Gender differences in expectancies for success and performance on Piagetian spatial tasks. Merrill-Palmer Quarterly, 32, 427-441. Miller, A. (1985). A developmental study of the cognitive basis of performance impairment after failure. Journal of Personality and Social Psychology, 49, 529-538. Miller, I. W, & Norman, W H. (1979). Learned helplessness in humans: A review and attribution-theory model. Psychological Review, 56,93-118. Mura, R. (1987). Sex-related differences in expectations of success in undergraduate mathematics. Journal for Research in Mathematics Education, 18,15-24. Parsons, J. E, Adler, T. F, & Kaczala, C. M. (1982). Socialization of achievement attitudes and beliefs: Parental influences. Child Development, 53, 310-321. Parsons, J. E, Meece, J. L, Adler, T. F, & Kaczala, C. M. (1982). Sex differences in attributions and learned helplessness. Sex Roles, 8, 421-432. Pasquella, M. H., Mednick, M. T. S, & Murray, S. R. (1981). Causal attributions for achievement outcomes: Sex-role identity, sex and outcome comparisons. Psychology of Women Quarterly, 5, 586-589. Phillips, D. (1987). Socialization of perceived academic competence among highly competent children. Child Development, 58, 13081320. Shrauger, J. S. (1975). Responses to evaluations as a function of initial self-perceptions. Psychological Bulletin, 82, 581-596. Shrauger, J. S, & Kelly, R. J. (1988). Global self-evaluation and changes in self-description as a function of information discrepancy and favorability. Journal of Personality, 56, 709-728. Shrauger, J. S, & Terbovic, M. L. (1976). Self-evaluation and assessments of performance by self and others. Journal of Personality and Social Psychology, 44, 564-572. Simon, J. G, & Feather, N. T. (1973). Causal attributions for success and failure at university examinations. Journal of Educational Psychology, 64, 46-56. Sleeper, L. A., & Nigro, G. N. (1987). ItTs not who you are but who you're with: Self-confidence in achievement settings. Sex Roles, 16, 57-69. Snyder, C. R. (1989). Reality negotiation: From excuses to hope and beyond. Journal of Social and Clinical Psychology, 8,130-157. Sohn, D. (1982). Sex differences in achievement self-attributions: An effect-size analysis. Sex Roles, 8, 345-357. Stein, A. H , Pohly, S. R, & Mueller, E. (1971). The influence of masculine, feminine, and neutral tasks on children's achievement behavior, expectancies of success, and attainment values. Child Development, 42,195-207. Swann, W B, Jr. (1987). Identity negotiation: Where two roads meet. Journal of Personality and Social Psychology, 53,1038-1051. Swann, W B, Jr, Griffin, J. J, Jr, Predmore, S. G, & Gaines, B. (1987). The cognitive-affective crossfire: When self-consistency confronts self-enhancement. Journal of Personality and Social Psychology, 52, 881-889. Swann, W B, Jr, & Hill, C. A. (1982). When our identities are mistaken: Reaffirming self-conceptions through social interaction. Journal of Personality and Social Psychology, 43, 59-66.

970

SYLVIA BEYER

Swann, W B, Jr., & Read, S. J. (1981a). Acquiring self-knowledge: The Weiner, B, Frieze, I. H, Kukla, A, Reed, L, Rest, S, & Rosenbaum, search for feedback that fits. Journal of Personality and Social PsyR. M. (1972). Perceiving the causes of success and failure. In E. E. chology, 41,1119-1128. Jones, E. E. Kanouse, H. H. Kelley, R. E. Nisbett, S. Valins, & B. Swann, W B, Jr., & Read, S. J. (1981 b). Self-verification processes: How Weiner (Eds.), Attribution: Perceiving the causes ofbehavior. Morriswe sustain our self-conceptions. Journal ofExperimental Social Psytown, NJ: General Learning Press. chology, 77,351-372. Wells, L. E, & Sweeney, P. D. (1986). A test of three models of bias in Taylor, S. E, & Brown, J. D. (1988). Illusion and well-being: A social self-assessment. Social Psychology Quarterly, 49,1-10. psychological perspective on mental health. Psychological Bulletin, Wiegers, R. M, & Frieze, I. H. (1977). Gender female traditionally, 705,193-210. achievement level, and cognitions of success and failure. Psychology Taylor, S. E., Collins, R. L., Skokan, L. A, & Aspinwall, L. G. (1989). of Women Quarterly, 2,125-137. Maintaining positive illusions in the face of negative information: Zuckerman, M. (1979). Attribution of success and failure revisited, or: Getting the facts without letting them get to you. Journal of Social The motivational bias is alive and well in attribution theory. Journal and Clinical Psychology, 8,14-129. of Personality, 47, 245-287. Viaene, N. (1979). Sex difference? in explanations of success and failure. In O. A. Hartnett, G. Boden, & M. Fuller (Eds.), Sex-role stereoReceived April 13,1989 typing. London: Tavistock. Revision received June 1,1990 Vollmer, F. (1986). Why do men have higher expectancy than women? Accepted June 8,1990 Sex Roles, 14, 351-362.

Butcher, Geen, Hulse, and Salthouse Appointed New Editors, 1992-1997


The Publications and Communications Board of the American Psychological Association announces the appointments of James N. Butcher, University of Minnesota; Russell G. Geen, University of Missouri; Stewart H. Hulse, Johns Hopkins University; and Timothy Salthouse, Georgia Institute of Technology as editors of Psychological Assessment: A Journal of Consulting and Clinical Psychology, the Personality Processes and Individual Differences section of the Journal of Personality and Social Psychology, the Journal of Experimental Psychology: Animal Behavior Processes, and Psychology and Aging, respectively. As of January 1,1991, manuscripts should be directed as follows: For Psychological Assessment send manuscripts to James N. Butcher, Department of Psychology, Elliott Hall, University of Minnesota, 75 East River Road, Minneapolis, Minnesota 55455. For JPSP: Personality send manuscripts to Russell G. Geen, Department of Psychology, University of Missouri, Columbia, Missouri 65211. For JEP: Animal send manuscripts to Stewart H. Hulse, Johns Hopkins University, Department of Psychology, Ames Hall, Baltimore, Maryland 21218. For Psychology and Aging send manuscripts to Timothy Salthouse, Georgia Institute of Technology, School of Psychology, Atlanta, Georgia 30332. Manuscript submission patterns make the precise date of completion of 1991 volumes uncertain. Current editors will receive and consider manuscripts through December 1990. Should any 1991 volume be completed before that date, manuscripts will be redirected to the newly appointed editor-elect for consideration in the 1992 volume.

S-ar putea să vă placă și