Task self-efficacy and task interest as a function of performance on a gender-neutral task

Task self-efficacy and task interest as a function of performance on a gender-neutral task

Journal of Vocational Behavior 30, 203-215 (1987) Task Self-Efficacy and Task Interest as a Function of Performance on a Gender-Neutral Task GAIL HA...

884KB Sizes 0 Downloads 191 Views

Journal of Vocational

Behavior 30, 203-215 (1987)

Task Self-Efficacy and Task Interest as a Function of Performance on a Gender-Neutral Task GAIL HACKETT University of California, Santa Barbara

AND NANCY K. CAMPBELL Ohio State University This study experimentally investigates several hypotheses about the relationships between performance on a gender-neutral task and gender, self-efficacy, performance attributions, and task interest. Ninety-two subjects were randomly assigned to a success or failure condition and attempted to solve a series of easy or difficult anagrams. Results indicated that changes in self-efficacy expectations as a result of task success or failure were in accordance with predictions from self-efficacy theory; 2 x 2 x 4 ANCOVAs, with the pretest as the covariate, were conducted on self-efficacy strength, level, and task interest. Subjects decreased their ratings of self-efficacy and task interest as a result of the failure experience, and the same ratings increased as a result of the success experience. Few gender differences were found, supporting the hypothesis that the sex linkage of the task significantly influences gender differences in self-efficacy. Analyses of global verbal and mathematical ability ratings resulted in the same trends. Finally, women in the success condition were signiIicantly more likely than men in that condition to attribute their performance to luck; women in the failure condition were signiticantly more likely than men or women in any other group to attribute their failure to their lack of ability. Implications of these results for future research on career self-efficacy were discussed. o 1987 Academic &SS, IN.

Emerging only recently as a research topic in the career area, selfefficacy theory has already garnered enough empirical support to be This research was partially supported by a grant from the Academic Senate of the University of California, Santa Barbara. An earlier draft of this article was presented at the annual meeting of the American Psychological Association, Toronto, Canada, August 1984. The authors gratefully acknowledge Joellen Popma for her assistance with the data collection and coding. Requests for reprints should be sent to Gail Hackett, Counseling Psychology Progtam, Graduate School of Education, University of California, Santa Barbara, CA 93106. 203 OOOl-8791/87 $3.00 Copyright 0 1987 by Academic Press, Inc. AU rights of reproduction in any form reserved.

204

HACKETT

AND

CAMPBELL

considered an important construct in furthering our understanding of career development, especially that of women in nontraditional careers (Osipow, 1986). Hackett and Betz’s (1981) application of Bandura’s (1977, 1982) self-efficacy theory to the career domain stimulated research on their basic proposal that the differential sex-role socialization of men and women in our society, ultimately resulting in gender differences in occupational choice and adjustment, might be mediated by cognitive appraisals of competence, that is, self-efficacy expectations, with regard to vocational pursuits (Betz & Hackett, 1986). Preliminary studies on career self-efficacy provided support for the basic hypothesis that gender differences in educational and career choices are related to gender differences in career self-efficacy, particularly in nontraditional career areas (Betz & Hackett, 1981, 1983; Hackett, 1985). More generally, occupational self-efficacy expectations have been found to be predictive of range of occupational alternatives considered (Betz & Hackett, 1981; Lent, Brown, & Larkin, 1986), choice of math and science-related college majors (Betz & Hackett, 1983; Hackett, 198% and academic performance and persistence (Lent, Brown, & Larkin, 1984, 1986). More recent research on gender differences in self-efficacy generally has been supportive of the earlier findings of Betz and Hackett (Post-Kammer & Smith, 1985, 1986). One shortcoming of the body of research on career self-efficacy has been the correlational nature of the studies. Experimental investigations of the effects of behavioral performance on career-related self-efficacy expectations have been identified as a critical research priority to test the usefulness of the self-efficacy construct in intervening to facilitate career development, in addition to simply predicting career choice behavior (Lent & Hackett, 1985). In the first experimental test of the self-efficacy approach to career development, Hackett and Betz (1984) investigated the effects of failure at a math or verbal task on general and specific measures of mathematics self-efficacy, and on global math and verbal ability ratings. Conflicting results emerged. Findings indicated that task failure influenced self-efficacy expectations, but not always in the expected direction. Gender x task interactions were observed, and, contrary to predictions, the effects of task failure in one domain, the math or verbal task, generalized, in some cases, to positively influence self-efficacy expectations in the other domain. Campbell and Hackett (1986), in a companion study to the present investigation, explored the effects of mathematical task success and failure over repeated trials on mathematical task self-efficacy, task interest, and performance attributions. The results were consistent with theoretical predictions: task success led to an increase in self-efficacy expectations, while task failure led to a corresponding drop in both level and strength of task self-efficacy; task performance similarly influenced ability ratings,

TASK

SELF-EFFICACY

AND

PERFORMANCE

205

task interest, and performance attributions. Gender differences and gender x task interactions in self-efficacy and task interest appeared. A trend for women to be more strongly affected than men by both success and failure was also observed. The task employed by Campbell and Hackett (1986) was a sex-linked task in the sense that mathematics is often viewed as a “male” domain, and gender differences are therefore routinely observed when this domain is investigated (Fennema & Sherman, 1977; Lenney, 1977). Because of the possibility that the results were a function of the sex linkage of the task, it was deemed important to study the effects of successful and unsuccessful performance on a different, nonsex-linked (i.e., gender-neutral) task in any future research. The purpose of the present study, then, is to continue the line of experimental research begun by Hackett and Betz (1984) and continued by Campbell and Hackett (1986). In this study, one gender-neutral task was employed, a verbal anagram task, and both a contrived success and a contrived failure condition were included. Subjects received three exposures to the experimental manipulation, the effects of task success or failure on task self-efficacy expectations and task interest were observed over time, and the generality of task performance effects examined. Finally, attributions of task performance were also assessed. The choice of task domain was influenced by previous research in the vocational area examining the effects of task performance on task interests (Locke, 1965; Osipow, 1972; Osipow & Scheid, 1971), and the research design is based on the assumption that career self-efficacy and vocational interests are aggregations of task self-efficacy expectations and task preferences. Consequently a controlled, microanalytic examination of the factors influencing task self-efficacy and task interests or preferences is seen as contributing to our understanding of more general vocational processes. The major hypotheses of the study are as follows: (a) task success will result in an increase in task self-efficacy and task interest, and task failure will produce a decrease in task self-efficacy and task interest; (b) gender differences in response to success or failure on this verbal (genderneutral) task will not occur; (c) the success and failure conditions will result in differential performance attributions; and (d) gender differences in performance attributions will be observed. METHOD

Subjects Subjects were 92 undergraduates (46 females, 46 males) enrolled in introductory psychology courses at a small, private liberal arts college in the Midwest. Participation was voluntary; subjects received course credit for their involvement in the study.

206

HACKETT

AND

CAMPBELL

Instruments Educational survey. A brief survey containing a series of questions eliciting demographic information, for example, age and gender, as well as information regarding educational background and major and career plans, was administered. This survey was a modified version of the instrument employed by Hackett and Betz (1984). In addition to the demographic information, two items assessing global ratings of subjects’ perceptions of their verbal and mathematical skills were included. For both verbal and mathematical ability, subjects were asked to rate themselves on a scale from extremely low ability (0) to extremely high ability (9), in comparison to other college students. Sel$eficacy/interesr ratings. Ratings of strength and level of task selfefficacy and interest in the task were obtained via a 3-item scale administered as a pretest, and then after each task attempt. Self-efficacy strength was assessed by asking subjects to rate their confidence in passing an anagram “test” on a scale from not confident at all (0) to very con.dent (9). Level of task self-efficacy was assessed by asking subjects to estimate the number of anagrams they expected to solve successfully (O-12); and task interest was rated from no interest at all (0) to a high degree ofinterest (9). Hackett and O’Halloran (1985) conducted a study ex amining the reliability of various self-efficacy measures, including the self-efficacy and interest ratings employed herein. They found the test-retest reliabilities of the self-efficacy level and strength ratings over a l-week period to be .55 and .70, respectively. The l-week test-retest reliability for the interest scores was reported as .67. A brief, eight-item questionnaire, Postexperimental questionnaire. adapted from K&la (1972), eliciting subjects’ reactions to the experimental task was administered after the completion of the study. Four questions were concerned with self-evaluations related to performance attributions, that is, ratings of potential ability, effort, task difficulty, and luck in solving the task. These are all factors found in the attribution literature to be important in performance self-assessment and future expectations of performance (e.g., Feather, 1966, 1%9; Feather & Simon, 1971; Kukla, 1972). Two other questions on this instrument required subjects to rate their success at the experimental task and their satisfaction with their performance. Global ratings of math and verbal abilities were assessed as on the educational survey. All other ratings were obtained on a O-9 scale from not at all to extremely; e.g., for the task luck item subjects responded to the question, “How lucky would you say you were in solving these tasks” on a scale of 0, not at all lucky to 9, extremely lucky.

TASK

SELF-EFFICACY

AND

PERFORMANCE

207

Procedure All subjects were met by one of two female experimenters, a doctoral candidate in counselor education and a master’s degree student in student personnel work and counseling. Subjects received instructions, completed the educational survey, and were then escorted to individual testing rooms where they were given a written description of the experimental task and were asked to complete the Self-Efficacy/Interest Rating Scale. Subjects were informed that the anagram task was a test of their verbal abilities, and that they had to successfully solve at least six anagrams to “pass” the test. The Self-Efficacy/Interest Rating Scale was collected, and subjects were again instructed that they had 10 min to finish the anagram task. Three different anagram tasks were then administered, with subjects completing the Self-Efficacy/Interest scale after each task attempt. Finally, the postexperimental instrument was administered, and subjects were thoroughly debriefed. Experimental task. Each verbal task employed in this study consisted of a set of 12 disarranged words, or anagrams, that subjects were asked to unscramble. A pilot study was conducted to assess the level of ditlicuhy of anagrams adapted from previous expectancy research (e.g., Feather, 1966; Feather & Simon, 1971), from the Hackett and Betz (1984) study, as well as additional anagrams developed by the authors. These anagrams were administered to counseling psychology students as a check on their level of difficulty. Ultimately, a pool of 72 anagrams, half extremely difficult, half relatively easy, were generated for use in the current study. Three sets of 12 anagrams for the “failure” condition, and three different sets of 12 anagrams for the “success” condition resulted. Specifically, subjects in the “success” group received a list of 12 anagrams for each task attempt, 10 of which were relatively easy to solve, 2 of which were somewhat more difficult. The subjects in the “failure” group received 10 difficult and 2 easy anagrams. Subjects were randomly assigned, by gender, to either the success or failure condition. Data Analysis First, a 2 x 2 x 3 (Experimental Group x Gender x Repeated Measures) ANOVA was performed on the actual number of correct items for each task attempt to check for group differences on this variable. Then, three 2 x 2 x 3 (Experimental Group x Gender x Repeated Measures) analyses of covariance, with the pretest as a covariate, were conducted on self-efficacy strength (i.e., confidence in passing the test), self-efficacy level (i.e., number of anagrams subjects expected to successfully solve), and task interest; the global ratings of verbal and mathematical skill were analyzed via 2 x 2 ANCOVAs (Experimental Group x Gender), with pretests as covariates. Additionally, two-way ANOVAs

208

HACKETT

AND

CAMPBELL

(Experimental Group x Gender) were performed on the ability, effort, difficulty, luck, satisfaction, and perceptions of success ratings from the postexperimental questionnaire. Newman-Keuls post hoc comparisons (Games, 1971) were employed to examine all significant main and interaction effects. Finally, correlation coefficients were computed between the four levels of the major dependent variables, that is, self-efficacy strength and level and task interest, to explore the relationships between task self-efficacy expectations and the degree of interest in the task. RESULTS

In order to check the experimental manipulation, the number of problems correctly solved by each subject for each separate anagram task was analyzed via a three-way (Experimental Group x Gender x Repeated Measures) ANOVA. Significant main effects for the Experimental Group [F(l, 88) = 2222.9, p < .OOOl] and Repeated Measures factors [F(2, 176) = 4.62, p < .Ol] provided support for the success of the experimental manipulation. No other significant main or interaction effects were found. Significant main effects for the covariate (i.e., pretest self-efficacy strength, F(1, 87) = 22.0, p < .OOl), and for the Experimental Group and Repeated Measures factors [F( 1, 87) = 409, p < .OOl, and F(2, 176) = 3.76, p < .05, respectively] emerged on the three-way (experimental group x gender x repeated measures) analysis of covariance on strength of self-efficacy. Significant two-way interactions for Repeated Measures x Experimental Group [F(2, 176) = 60.21, p < .OOl] and Repeated Measures x Gender [F(2, 176) = 3.77, p < .05] were also found. The cell means and standard deviations for the covariance analysis on self-efficacy strength are presented in Table 1. The Newman-Keuls post hoc analysis (Games, 1971) on the adjusted cell means for the repeated measures x experimental group interaction yielded significant differences between all cell means (p < .05). Thus, the success group rated their confidence in passing the anagram test as higher than the failure group on every rating; the success group’s ratings went up significantly over time, and the failure group’s ratings declined significantly over time. The post hoc analyses on the repeated measures x gender interaction (collapsed over the Experimental Group factor) revealed no significant differences for women’s ratings across time, but men’s self-efficacy strength ratings declined significantly between Post-test 1 and Post-test 3 @ < .05). Results for the three-way covariance analysis on level of self-efficacy are very similar to the findings for the covariance analysis of self-efficacy strength, except that for self-efficacy level no significant repeated measures X gender interaction emerged. For the analysis on self-efficacy level, significant main effects for the covariate [F( 1, 87) = 27.5, p < .OOl] and the Experimental Group and Repeated Measures factors [F(l, 87) =

TASK

SELF-EFFICACY

209

AND PERFORMANCE

TABLE 1 Cell Means and Standard Deviations for Self-Efficacy Strength (Confidence) Experimental Group and Gender Experimental

group

Success Self-efficacy strength Pretest (covariate) Post-test 1 Post-test 2 Post-test 3

by

Failure

Female

Male

Female

Male

M

SD

M

SD

M

SD

M

SD

6.78

1.9

6.26

1.9

5.70

1.9

7.13

1.9

7.74 8.30 8.%

1.6 1.4 1.3

7.17 7.91 8.13

2.0 2.1 2.2

2.78 1.96 1.65

1.4 1.2 0.9

4.30 2.61 2.00

1.8 1.5 1.3

Note. N = 92, 23 subjects per group. * Confidence in ability to “pass” the verbal task rated on a scale from Nor conjident at all (0) to Very confident (9).

38.9, p < .OOl, and F(2, 176) = 8.44, p < .OOl, respectively] and for the Repeated Measures x Experimental Group interaction [F(2, 176) = 28.79, p < .OOl] were found. Newman-Keuls post hoc comparisons on the significant two-way interaction from the ANCOVA on self-efficacy level were performed. Table 2 presents the cell means and the standard deviations by Experimental Group and Gender. Level of self-efficacy was found, once again, to be higher overall for the success than for the failure group (p < .05); furTABLE 2 Cell Means and Standard Deviations for Self-Efficacy Level by Experimental Group and Gender Experimental

group Failure

Success Male

Female Self-efficacy lever Pretest

Female

Male

M

SD

M

SD

M

SD

M

SD

6.74

1.6

7.04

1.7

6.48

1.8

6.93

1.9

8.00 8.17

1.6 1.3

7.70

2.1

1.8

1.2

8.56

1.2

8.13

1.7

3.52 2.87 2.56

1.0

7.91

4.74 3.13 3.09

1.7 1.7 1.3

(covariate)

Post-test 1 Post-test 2 Post-test 3

Note. N = 92, 23 subjects per group. * Self-estimated predictions of the number of anagrams subjects (o-12).

1.0

would correctly solve

210

HACKETT

AND CAMPBELL

TABLE 3 Cell Means and Standard Deviations for Task Interest by Experimental Group and Gender Experimental

group Failure

Success Female

Task interest” Pretest (covariate) Post-test 1 Post-test 2 Post-test 3

Female

Male

M

SD

M

SD

M

SD

M

SD

6.00

1.4

5.56

2.5

5.04

2.3

5.65

1.5

6.70 6.57 6.26

1.4 1.6 1.7

6.17 6.13 5.91

2.5 2.6 2.4

3.91 3.39 2.91

2.4 2.6 2.1

4.39 3.56 3.09

1.8 2.2 2.3

Note. N = 92, 23 subjects per group. ’ Degree of interest (liking) for task, rated from no of interest

Male

interest

at all (0)

to a high degree

(9).

thermore, subjects in the success condition significantly increased their self-efficacy level ratings over time, while the opposite trend existed for subjects in the failure condition Gr, < .05). No significant gender differences emerged. The repeated measures analysis of covariance (Experimental Group x Gender x Repeated Measures, with the pretest as the covariate) on task interest ratings also yielded significant main effects for the experimental condition [F(l) 87) = 70.6, p < .OOl], and the Repeated Measures factor [F(2, 176) = 23.25, p < .OOl], and a significant two-way interaction [Repeated Measures x Experimental Group, F(2, 176) = 7.16, p < .Oll. No gender differences or interactions involving gender occurred. The post hoc analysis of the significant two-way interaction revealed that the task interest ratings of subjects in the success group did not differ significantly over time whereas the ratings of subjects in the failure group declined significantly between the first and last post-test (all ps < .05). Unadjusted cell means and standard deviations from this analysis are presented in Table 3. Results from the two-way ANCOVAs on global verbal and global mathematical ability ratings also support the major experimental hypothesis. Significant main effects for experimental group emerged [F( 1,87) = 14. I and 6.7, ps < .OOl and .05, for verbal and math, respectively], indicating higher ratings for subjects in the success condition. No significant effects for gender were found, nor were there any significant interactions. Cell means and standard deviations for the global verbal and math ability ratings are displayed in Table 4. The results of the 2 x 2 ANOVAs (Experimental Group x Gender)

TASK SELF-EFFICACY

211

AND PERFORMANCE

TABLE 4 Means and Standard Deviations for Global Verbal and Math Ability Ratings by Experimental Group and Gender Experimental

group

Success

Failure

Female Ability ratings Verbal rating (covariate) Verbal rating (post-test) Math rating (covariate) Math rating (post-test)

M

Male SD

Female

Male

M

SD

M

SD

M

SD

1

6.78

1.4

8.83

1.7

6.43

1.5

6.74

1.4

2

7.17

1.2

6.87

1.6

5.91

1.6

6.84

2.0

1

5.70

1.9

6.39

1.8

4.74

1.8

6.13

1.9

2

5.%

1.7

6.52

1.8

4.65

1.8

5.78

2.1

Note. N = 92, 23 subjects per group. Verbal skills/abilities were rated on a scale from low ability (0) to Extremely high ability (9); mathematical scores were rated on the same scale. Subjects were asked to rate themselves in comparison to other college students. Extremely

on the post-test attribution and evaluation ratings yielded significant main effects for experimental groups for all items (all ps < .05). That is, subjects in the success condition rated themselves as significantly more able, as trying harder, as more lucky, as more successful, as more satisfied, and rated the task as less difficult, than subjects in the failure condition. Furthermore, in addition to the main effect for experimental group, significant Experimental Group x Gender interactions were found on two items: ability and luck [F(l, 88) = 4.8 and 4.4, ps < .05]. In response to the question “How good are you, potentially, at this kind of task,” subjects in the success condition rated themselves significantly higher than subjects in the failure condition. Newman-Keuls post hoc comparisons further demonstrated that no gender differences existed between subjects in the success condition, but a significant gender difference in favor of men (p < .OS) emerged between subjects in the failure condition. Post hoc comparisons of the experimental group x gender interaction for the item asking subjects to rate their degree of luck in solving the anagrams yielded significant differences between women in the successcondition and the three other groups (p < .05). No significant differences emerged between men in the success condition and men and women in the failure condition. Therefore, while men and women who failed at the task rated their ability as significantly lower than men and women who were successful, women who failed seemed to attribute their

212

HACKETT

AND CAMPBELL

TABLE 5 Correlations between Strength and Level of Self-Efficacy and interests Over Time Variables Self-efficacy strength 1. Pretest 2. Post-test 1 3. Post-test 2 4. Post test 3 Self-efficacy level 5. Pretest 6. Post-test 1 7. Post-test 2 8. Post-test 3 Task interest 9. Pretest 10. Post-test 1 11. Post-test 2 12. Post-test 3

1

2

3

4

5

6

7

.46 .22 .16

.a7 .85

.%

.80 .40 .24 .19

.36 .92 .83 .83

.19 .83 .93 .91

.I3 .84 .92 .95

.32 .20 .13

35 35

.94

.58 .49 .39 .29

.36 .67 .64 .66

.25 ..59 .61 .63

.24 58 .60 .65

.59 .45 .36 .29

.30 .63 58 S9

.28 .63 .64 .63

8

9

10

11

.22 150 .62 .65

.78 .69 .62

.94 .88

.94

Note. N = 92, 46 males and 46 females. Values of r of .20 and .27 correspond to the .05 and .Ol levels of statistical significance.

failure more often than men to their lack of ability. Conversely, women who were successful at the task were more likely to attribute their success to luck than successful men, or women or men who had failed at the task. Finally, to explore the interrelationships among the two self-efficacy variables across time and the task interest variable across time, Pearson product-moment correlation coefficients between the major dependent variables were computed. As can be seen from the correlation matrix presented in Table 5, pretest scores correlate most highly with other pretest scores and post-test scores correlate most highly with other posttest scores, reflecting the changes in all ratings due to task success or failure. The task interest intercorrelations between pretest and the posttests are more stable than the pretest/post-test correlations for the selfefficacy variables, indicating that the self-efficacy variables are more sensitive to the experimental manipulations than the interest variable. Correlations for task interest ratings range from .62 to .94; self-efficacy strength intercorrelations range from .16 to .96, while the correlations on the various self-efficacy level ratings range from .13 to .94. All correlations between post-test scores for the same variable are high (TS of .85 to .94 for the self-efficacy strength post-test intercorrelations, .85 to .96 for self-efficacy level, and .88 to .94 for task interest). However, between-variable correlations for post-test ratings are higher for the selfefficacy variables (TS of .83 to .95 for post-test strength/level correlations) than for either self-efficacy variable and task interest (YS of .58 to .67

TASK

SELF-EFFICACY

AND

PERFORMANCE

213

for post-test strength/interest correlations, rs of .58 to .65 for post-test level/interest correlations). This pattern of results is very similar in nature and magnitude to the results reported by Campbell and Hackett (1986). The findings of both in-;cstigations support the hypothesis that, while significant correlations between self-efficacy and interests exist, the magnitude of these correlations is moderate. DISCUSSION

The present investigation, in combination with the results of the companion study conducted by Campbell and Hackett (1986), provides support for the Hackett and Betz (1981) extension of Bandura’s self-efficacy theory (1977, 1986) to the career domain. As hypothesized, task performance influenced ratings of task self-efficacy and global verbal ability, with successful subjects increasing their self-efficacy and ability ratings with regard to the experimental task and unsuccessful subjects decreasing these same ratings. Global mathematical ability ratings were similarly affected by task success or failure, although to a lesser degree than were verbal ability estimates. These findings provide support for Bandura’s (1977, 1982) contentions concerning the generality of self-efficacy expectations. Task performance was also found to affect task interest in the same ways in which task self-efficacy was affected, albeit to a lesser degree. These findings also replicate those reported by Campbell and Hackett (1986). In contrast to the consistent and strong gender differences in selfefficacy and task interests observed by Campbell and Hackett (1986) in their very similar study employing success and failure on a mathematical task, the absence of gender differences in this study is particularly striking. The failure to find clear-cut differences in men’s and women’s responses to a gender-neutral task underscores the importance of the nature of the task to understanding the effects of performance on self-efficacy expectations. And, although the present study investigated tasks that can only be considered an analog to the kinds of variables that contribute to the development of vocational interests and career self-efficacy, the results are consistent with those of investigations on more general phenomena relevant to occupational choice. For example, the studies performed on career self-efficacy and occupational choice have consistently supported the hypothesis that gender differences in self-efficacy are likely to occur where gender-role stereotyping of occupations is greatest. That is, the more nontraditional the occupation or career domain is for women, the more likely that gender differences in occupational self-efficacy will occur (cf. Betz & Hackett, 1981; Hackett, 1985; Post-Kammer & Smith, 1985). The relationships obtained between the self-efficacy and task interest variables are also very similar to the patterns of relationships between these variables reported by Campbell and Hackett (1986). The results

214

HACKETT

AND CAMPBELL

from both studies provide correlational support for Betz and Hackett’s (1981) contention that self-efficacy, though related to interests, contributes independently to the prediction of performance and other vocationally related behaviors. And finally, the patterns in the results of the attribution and selfevaluation measures reported in this study are remarkably similar to those reported by Campbell and Hackett (1986). Task performance strongly and significantly affected performance evaluations; all subjects in the success group scored more positively on all items than subjects in the failure group. Fewer gender differences on self-evaluations were found in the present study than in Campbell and Hackett (1986); however, it was found that women in the failure group rated their potential ability lower than men in that same group and women in the success group rated their luck in completing the task higher than men in the same group. It remains to be seen how these differential attributions to ability and luck interact with self-efficacy expectations to affect performance. In summary, one of the main purposes of this research was to experimentally test the hypothesis, derived from Bandura’s (1977) selfefficacy theory, that efficacy expectations relevant to vocational task behaviors can be modified via behavioral performance and that the nature of the task will determine whether gender differences in responding will occur. The results of this study, in combination with previous and concurrent findings (Campbell & Hackett, 1986; Hackett & Betz, 1984), provide support for this hypothesis. The findings also suggest that selfefficacy interacts complexly with situational factors, type of task, vocational interests, gender-role influences, and with performance attributions. Further investigations are needed to disentangle the effects of these reciprocal influences on career decision making. REFERENCES Bandura, A. (1977). Self-efficacy: Toward a unifying theory of behavioral change. Z’sychological Review, 84, 191-215. Bandura, A. (1982). Self-efficacy mechanism in human agency. American Psychologist, 31, 122-147. Bandura, A. (1986). Social foundations of &ought and action. Engtewood Cfis, NJ: Prentice-Hall. Betz, N. E., & Hackett, G. (1981). The relationship of career-related self-efficacy expectations to perceived career options in college women and men. Journal of Counseling Psychology, 28, 399-410. Betz, N. E., & Hackett, G. (1982, August). Behavioral competence and ielf-efficacy expectations with respecr to career facilitation skills. Paper presented at the Annual Meeting of the American Psychological Association, Washington, DC. Betz, N. E., & Hackett, G. (1983). The relationship of mathematics self-efficacy expectations to the selection of science-based college majors. Journal of Vocational Behavior, 23, 329-345. Betz, N. E., L Hackett, G. (1986). Applications of self-efficacy theory to understanding career choice behavior. Journal of Social and Clinical Psychology, 4, 279-289.

TASK SELF-EFFICACY

AND PERFORMANCE

215

Campbell, N. E., & Hackett, G. (1986). The effects of mathematics task performance on math self-efficacy and task interest. Journal of Vocational Behavior, 28, 149-162. 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. (1%9). Attribution of responsibility and valence of success and failure in relation to initial confidence and task performance. Journal of Personality and Social Psychology, 13, 129-144. Feather, N. T., & Simon, J. G. (1971). Attribution of responsibility and valence of outcome in relation to initial confidence and success and failure of self and other. Journal of Personality and Social Psychology, 18, 173-188. Fennema, E., & Sherman, J. A. (1977). Sex-related differences in mathematics achievement, spatial visualization, and affective factors. American Educational Research Journal, 14, 51-71. Games, P. A. (1971). Multiple comparisons of means. American Educational Research Journal, 8, 531-565. Hackett, G. (1985). The role of mathematics self-efficacy in the choice of math-related majors of college women and men: A path analysis. Journal of Counseling Psychology, 32, 47-56. Hackett, G., & Betz, N. E. (1981). A self-efficacy approach to the career development of women. Journal of Vocational Behavior, 18, 326-339. Hackett, G., L Betz, N. E. (1984, April). Gender difirences in the effects of relevanr and irrelevant task failure on mathematics self-eficacy expectations. Paper presented at the annual meeting of the American Educational Research Association, New Orleans. Hackett, G., & O’Halloran, S. (1985). Test-retest reliabilities of career self-efficacy measures. Unpublished manuscript. Kukla, A. (1972). Attributional determinants of achievement-related behavior. Journal of Personality and Social Psychology, 21, 166-174. Lenney, E. (1977). Women’s self-contidence in achievement settings. Psychological Bulletin, 84, 1-13. Lent, R. W., Brown, S. D., & Larkin, K. C. (1984). Relation of self-efficacy expectations to academic achievement and persistence. Journal of Counseling Psychology, 31,356362. Lent, R. W., Brown, S. D., & Larkin, K. C. (1986). Self-efficacy in the prediction of academic performance and perceived career options. Journal of Counseling Psychology, 33, 265-269. Lent, R. W., & Hackett, G. (1985, August). Methodological and conceptual considerations in studying career self-eficacy. Paper presented at the annual meeting of the American Psychological Association, Los Angeles. Locke, E. A. (1%5). The relationship of task success to task liking and satisfaction. Journal of Applied

Psychology,

49, 379-385.

Osipow, S. H. (1972). Success and preference: A replication and extension. Journal of Applied Psychology, 56, 179-180. Osipow, S. H. (1986). Career issues through the life span. In M. S. Pallak & R. 0. Perloff (Eds.), Psychology and work: Productivity, change, and employment (APA Master Lecture Series). Washington, DC: American Psychological Association. Osipow, S. H., & Scheid, A. B. (1971). The effect of manipulated success ratios on task performance. Journal of Vocational Behavior, 1, 93-98. Post-Kammer, P., & Smith, P. L. (1985). Sex differences in career self-efficacy, consideration, and interests of eighth and ninth graders. Journal of Counseling Psychology, 32, 551559. Post-Kammer, P., & Smith, P. L. (1986). Sex differences in math and science career selfefficacy among disadvantaged students. Journal of Vocational Behavior, 29, 89-101. Received: August 18, 1986.