'.(1|11!11bl, M " II',‘“,.' ,Ili,’..2mlll'1' WU‘I‘WWV‘W '1 (W HW'I'JIF I! . HQIHM‘I1I‘S1 . "1' . .II. If,“ Immlh,‘, M ”flu“. N‘ ”I W I ‘1' K g ....... I I. " .‘111‘1 I‘M, W I I _ , 5 11 1 1.411A.4! 1 1 1 11 1 1 I . 1 1 1 111u1::111111511\.‘1151'.¢1;1 fi",1‘":‘ t,‘ 51161, I' 5 .1 ,lfiIf'i‘, 1 1,, I1 I ‘- 111; .I1I- .11 .1 11‘ 1,11,.1 v.3 3:3" ..'I ‘.2 .73? 15:1,: Lji‘I 1, 1‘1E1I'. 1111,111,111111L111111111111111 a11yfi 1:197:1- “M11 I l I 1 111 ,11 11.11.1111 .;1II1 , ; . :1 .111;I1.111;11-1-I1.1,_ 311.5; ,.1, u. I, I I. I. 31.1,] 11.11111 11 ,I,19 -I .I- t,I,, II II III ...'1 .III. 'r: .‘Isr‘tifli ‘ .. " I .. . “1;" 'L \‘1‘ I'm i,‘ .- Hm, "5‘” i't‘IfI‘Il.‘LI L- ' HW'I ,I'“I'III-.I‘1'1i133‘11"!‘7“"“3 ‘..1 “”1“” 5.9.21. 1} I 7" I!“ w1'T‘IIt'hH . .,, .I . H. .H . . ..II .,I1,1 5. III! I... Inn . .~ I (.1. HI] .. |<.'l J. l I .\1 .1‘. . . .. . l\|.. ’I'." .1 J1 1:3!SFI‘1‘7'9.‘ ,35" ' " '.!1' ‘ I'I'I’I"I 1'1 lI II: WWW "'IL".!.‘11‘I3‘3‘11'."f"4! '1'” ~I» - IH‘IH l H ‘1'?“ "I II' - IIII‘II'I: . .1 . I “t'I’rI "" ' .~ :IIIn . . . I... .' III.~ I.‘.: ;:III~:'-‘I"'II‘: ‘2‘..II"I"~W I‘II'I'I‘II ’lw ~ .1'1"“I‘.;."." III'. I‘: 1 “'h ‘ .I 'I ,‘I' I y ‘:“.“.' g I. 1, I- ,‘1 1.1 .1. ,,. .1.11,1:.,11 .I,',,"‘I ' .1,,,,1.11I11I,I.,; "1 , '4 - ..;r ‘L 51%. 1.5 ,‘lffdvm: ...,t:t'1,“11'\1..5 “11;, 'HH, ‘ ' IIM‘I "I W'I': .. ' -“I ..I ‘ "I""u‘I"I'VL" ',I .. I" ' ..'.) '1“. 1““ 1[|1l‘ Um, 011;“ |, [I' 1I‘,{,"‘,|11:l1‘[,, 11111111111111.1'1 I .I31I1 I’1I111“.1'11...1|1l1 '11111 I 11116111111111 ”11 1,1111. ‘1 ,,|1Dl‘ '1 ‘1 ‘ W I, |. ' 81' ‘,'I'.“'.1.“"‘1'111 ‘1: II1' 'I'q. I‘VE." ‘ ,.. I. I‘ \ "11 111"1.1'.|. 11.1,. ,11'1-«1.1I,|,',’1 I1.,1!H'1I,111‘ (Huh ‘I‘U.’,1",,\1m ”£13; 1'1I1.1“]1”1"I1')111'1'.‘1'; ' '-.' I01?" “.‘I. ,',...1II'I. .I,1.,,'. I ;I.I_;.. . ;-.e‘11..!11,I1I:.1'-.II 3'. ‘I‘H'. '3 2 . 1..1,1,:',.1‘1I‘,1‘,111 ",1 ,I‘ .,I10 1"I!‘.11",' I; :1m1,11"1l31'1" . ". l1.11[\1'1.-,' '“ ."II N." "m' ""m 11‘“ ‘ I .1 15111151, 1,1:I \1l1,'1 1|I1,1. 1v-‘11.1[1,11111511k1,1l'!1\11‘11'111.1!11:1‘1D‘1|1T,‘1!|1'w“1‘11 1,”, ‘,:I,1;1,',5111i",“,f.;';f1‘5.1I111H1Wtu‘l‘11irI1m 1,11||'l‘h|”111l.‘ M“ ”1“ . , 5,1. ,, I N. II.I;.,IIII~ III, II 'I , \ , ;. ,' I “I”... ,I‘I" '1'1I.1\11 ". ' :I'1'“1',‘1I1I.,‘.'II"H\1“ 1““, '.I 'mlw‘ “‘l ml,” "1‘ 1,,” L5“ II III III I IIII. ~i'fl" 'I"I,I """'I'111.‘,1I,'. , II." I... III I II III .1 ll‘ "HUN" " '4' WM N I H IIIIIII‘HW 'H’VI ..' L'W'I ‘Ll‘h‘lwligfimmwlm ”‘55" " "BM RY Mich} 3311 State University This is to certify that the thesis entitled THE EFFECTS OF SEX LINKAGE OF AREA OF ACHIEVEMENT ON MEN'S AND WOMEN'S EXPECTATIONS FOR SUCCESS AND CAUSAL ATTRIBUTIONS FOR OUTCOME presented by Michelle Rae Klee has been accepted towards fulfillment of the requirements for M.A. Psychology degree in Major professor Date 11/4/82 0-7639 MS U is an Affirmative Action/Equal Opportunity Institution lllllllllllllllllllllllllIlllllllllllllllllllllllllllllllllll 31293 10527 2052 MSU LIBRARIES w » RETURNING MATERIALS: Place in book drop to remove this checkout from your record. FINES will be charged if book is returned after the date stamped below. Mpwfi . .7, J- ‘" I it". h ‘! q! THE EFFECTS OF SEX LINKAGE OF AREA OF ACHIEVEMENT ON MEN'S AND WOMEN'S EXPECTATIONS FOR SUCCESS AND CAUSAL ATTRIBUTIONS FOR OUTCOME BY Michelle Rae Klee A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of MASTER OF ART Department of Psychology 1982 ABSTRACT THE EFFECTS OF SEX LINKAGE OF AREA OF ACHIEVEMENT ON MEN'S AND WOMEN'S EXPECTATIONS FOR SUCCESS AND CAUSAL ATTRIBUTIONS FOR OUTCOME BY Michelle Rae Klee Previous research has led to the conclusion that women have achievement inhibiting patterns of expectancy and causal attribution relative to men. The purpose of this investigation was to determine whether this conclusion is valid outside traditionally masculine achievement situations. Eighty-one female and sixty-five male undergraduate volun- teers were tested in mixed-sex groups. Task-specific ex- pectancies and attributions were measured regarding tests in traditionally feminine and masculine ability areas, interpersonal and academic skill. General attributions for academic and interpersonal outcome were assessed using the Multidimensional-Multiattributiona1 Causality Scale (MMCS). The results revealed few sex differences in expectancy and attributions within area of achievement. Differences across sex in interpersonal and academic MMCS attribution patters were discussed in terms of the applicability of the conceptual models to an interpersonal context. The possible influence of methodological and situation variables on ex- pectancy and attribution patterns found among men and women were discussed. ACKNOWLEDGEMENTS I wish to acknowledge the people who assisted me in this research. Elaine Donelson, Ph.D., was invaluable as the chairperson of my thesis committee. Her consistent support, interest and belief in the value of the research were extremely important to me. Elaine's familiarity with the literature and her help in the conceptualization of issues involved in women and achievement were very useful. I wish to express my appreciation to Larry Messé, Ph.D., for his help in the design and statistical analysis for this study. Larry's expertise, availability and patience were great assets. I also wish to express my appreciation to Dozier Thornton, Ph.D., for his assistance with and support of my research efforts in all stages of evolution toward a completed thesis. I am especially grateful for Dozier's belief,despite many delays, that I would finish. Finally, I would like to acknowledge the undergraduate research assistants who were instrumental in the pilot and data collection phases of this research: Karen Camella, Mary Clissold, Lori Dezell, Brian Fillingham, Celeste May and Randy Russell. ii TABLE OF CONTENTS LIST OF TABLES. . . . . . . . . . . . . . . . . . . . CHAPTER I. II. III. INTRODUCTION. . . . . . . . . . . . . . . . . . Construct of Locus of Control . . . . . . . . . The Weiner Model of Causal Attribution. . . . . Sex Differences in Expectancy and the Perception of Causality . . . . . . . . . . . Consideration of Achievement from a Feminine Perspective. . . . . . . . . . . . . Attribution in Traditionally Feminine Domains . Need for Goal Specific Measures of Attribution. The Problem . . . . . . . . . . . . . . . . . . Hypotheses. . . . . . . . . . . . . . . . . . . METHOD 0 O O O O O O O O O O O O O O O O O O O 0 Overview of Study . . . . . . . . . . . . . . . Subjects. . . . . . . . Experimenters . . . . . Procedure . . . . . . . Instruments . . . . . . RESULTS 0 O I O O O C O O O O O O O O O O O O 0 Experimental Design and Statistical Analysis. . Design. . . . . . . . . . . . . . . . . . . Classification of Subjects According to Success or Failure on Tasks. . . . . . Overview of Statistical Analysis. . . . . . Hypothesis Testing Multivariate Analyses of Variance . . . . . . . . . . . . . . . . . Hypothesis One. . . . . . . . . . . . . . . Hypothesis Two. . . . . . . . . . . . . . . Hypothesis Three. . . . . . . . . . . . . . Hypothesis Four . . . . . . . . . . . . . . Unpredicted Effects Across Sex for Hypothesis Testing Analyses . . . . . . . . . Post Hoc Multivariate Analyses of Variance. . . iii PAGE vi 11 18 26 30 34 38 43 43 45 47 47 51 59 59 59 59 61 62 66 69 72 72 74 77 IV. V. Distribution of High Attribution Scores . . . . Within Sex Patterns . . . . . . . . . . . . Between Sex Patterns. . . . . . . . . . . . Differences Between Area of Achievement Across Sex. . . . . . . . . . . . . . . . Summary of Results. . . . . . . . . . . . . . . DISCUSSION 0 O O O O O O O O O O O O O O O O O 0 Differences as a Function of Area of Achievement . . . . . . . . . . . . . . . Hypotheses One and Two. . . . . . . . . . . Valuation of Achievement in Masculine and Feminine Areas. . . . Applicability of the Model of Causality to the Communal Context . . . . . . . . . Sex Differences: Hypotheses Three and Four . . Traditionally Masculine Area of Achievement . . . . . . . . . . . . Traditionally Feminine Area of Achievement. Ambiguity of the Data on Sex Differences in Masculine Areas. . . . . . . . . . . . . Influence of Situation Variables as Results . . . . . . . . . . . . . . Discrepancies Between the MMCS and the Behavioral Task Data. . . . . . . . . . . Implications for Women and Achievement. . . . . Possible Changes in Attribution and Expectancy. . . . . . . . . . . . . . . . Other Factors Influencing Achievement Among Women . . . . . . . . . . . . . . . Area of Achievement by Condition Interaction. . CONCLUSIONS . . . . . . . . . . . . . . . . . . APPENDICES A. B. C. D. E. F. G. H. I. J. Experimenter Instructions . . . . . . . . . . . Preliminary Questionnaire . . . . . . . . . . . Expectations Questionnaire. . . . . . . . . . . Interpersonal Sensitivity Test. . . . . . . . . Matching Parts and Figures Test . . . . . . . . Posttest Questionnaire. . . . . . . . . . . . Multidimensional-Multiattributional Causality Scale . . . . . . . . . . . . . . . Personal Characteristics Form . . . . . . . . Hypothesis Testing Multivariate Analysis of Variance Tables . . . . . . . . . . . . F Ratios for Area of Achievement Effect Summary Tables. . . . . . . . . . . . . . . . iv 80 82 85 86 90 94 95 97 100 103 104 108 110 111 116 121 121 122 125 128 131 142 143 144 151 152 155 160 162 165 K. Post Hoc Multivariate Analysis of Variance Tables. 0 O O O O O O O O O I O O O O O O O O O 174 L. Summary Tables for Chi-square Analysis of Sex x Category of High Attribution Score Relationship. . . . . . . . . . . . . . . . . . 177 BIBLIOGRAPHY.....................181 10. LIST OF TABLES PAGE Outline of Procedure in Condition 1 and Condition 2. . . . . . . . . . . . . . . . . . 44 Definition of Variables Derived from Each Experimental Measure . . . . . . . . . . . . . 46 Summary of Significant Multivariate Results for the 2 (Sex) x 2 (Area of Achievement) x 2 (Condition) Analysis of Expectancy, Ability and Effort Scores for the Behavioral Tasks . . 64 Summary of Significant F Ratios for the Expectancy, Ability and Effort Scores for the Behavioral Tasks for All Subjects. . . . . 65 Summary of Significant E Ratios for the Expectancy, Ability and Effort Scores for the Behavioral Tasks for Group 3, IPS Success-MPF Failure Subjects . . . . . . . . . 65 Mean Expectancy and Attribution Scores for Interpersonal and Academic Outcome within seXO O O O O O O O O O O O O O O O O I I O O O 67 Mean Expectancy and Attribution Scores for Men and Women Within Area of Achievement . . . 73 Percentage of Men and Women With Highest Attribution Score in Each Category Within Area of Achievement. . . . . . . . . . . . . . 81 Percentage of Men and Women With Higher Attribution Scores in the Internal or External Attribution Category Within Area of Achievement. . . . . . . . . . . . . . 83 Percentage of Subjects With Highest Attribution Score in Each Attribution Category Within Area of Achievement. . . . . . 87 vi TABLE 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. Percentage of Subjects With Higher Attribution Scores in the Internal or External Attribution Category Within Area of Achievement. . . . . . Mean Scores for Posttest Checks of Perceived Sex-linkage of IPS and MPF . . . . . . . . . . Summary of 2 x 2 x 2 MMCS MANOVA Results for Success Ability, Success Effort and Failure Effort Attribution Scores. . . . . . . . . . . Summary of 2 x 2 x 2 Behavioral Task MANOVA Results for Expectancy, Ability and Effort Scores of All Subjects . . . . . . . . . . . . Summary of 2 x 2 x 2 Behavioral Task MANOVA Results for Expectancy, Ability and Effort Scores of Group 1, IPS Failure-MPF Failure Subjects . . . . . . . . . . . . . . . . . . . Summary of 2 x 2 x 2 Behavioral Task MANOVA Results for Expectancy, Ability and Effort Scores of Group 2, IPS Failure-MPF Success Subjects . . . . . . . . . . . . . . . . . . . Summary of 2 x 2 x 2 Behavioral Task MANOVA Results for Expectancy, Ability and Effort Scores of Group 3, IPS Success-MPF Failure Subjects . . . . . . . . . . . . . . . . . . . Summary of 2 x 2 x 2 Behavioral Task MANOVA Results for Expectancy, Ability and Effort Scores of Group 4, IPS Success-MPF Success Subjects . . . . . . . . . . . . . . . . . . . Summary of F Ratios for Main Effect of Area of Achievement for MMCS Data . . . . . . . . . Summary of F Ratios for Main Effect of Area of Achievement for Behavioral Task Data of All smjects O O O O O O O O O O O O C O O O 0 Summary of F Ratios for Main Effect of Area of Achievement for Behavioral Task Data of IPS Failure-MPF Failure Subjects . . . . . . . Summary of F Ratios for Main Effect of Area of Achievement for Behavioral Task Data of IPS Failure-MPF Success Subjects . . . . . . . vii PAGE 89 114 162 162 163 163 164 164 165 165 166 166 TABLE PAGE 23. Summary of E Ratios for Main Effect of Area of Achievement for Behavioral Task Data of IPS Success-MPF Failure Subjects. . . . . . . . 167 24. Summary of F Ratios for Main Effect of Area of Achievement for Behavioral Task Data of IPS Success-MPF Success Subjects. . . . . . . . 167 25. Summary of F Ratios for Main Effect of Area of Achievement for MMCS Data of Females . . . . 168 26. Summary of F Ratios for Main Effect of Area of Achievement for Behavioral Task Data of All Felnales O O O O O O O O O O O O O O O O O O 168 27. Summary of F Ratios for Main Effect of Area of Achievement for Behavioral Task Data of IPS Failure-MPF Failure Females . . . . . . . . 169 28. Summary of F Ratios for Main Effect of Area of Achievement for Behavioral Task Data of IPS Failure-MPF Success Females . . . . . . . . 169 29. Summary of F Ratios for Main Effect of Area of Achievement for Behavioral Task Data of IPS Success-MPF Failure Females . . . . . . . . 170 30. Summary of F Ratios for Main Effect of Area of Achievement for Behavioral Task Data of IPS Success-MPF Success Females . . . . . . . . 170 31. Summary of E Ratios for Main Effect of Area of Achievement for MMCS Data of Males . . . . . 171 32. Summary of F Ratios for Main Effect of Area of Achievement for Behavioral Task Data of All Males C I O C O O O O O I O O O I O O O O O 171 33. Summary of §_Ratios for Main Effect of Area of Achievement for Behavioral Task Data of IPS Failure-MPF Failure Males . . . . . . . . . 172 34. Summary of F Ratios for Main Effect of Area of AchievemEnt for Behavioral Task Data of IPS Failure-MPF Success Males . . . . . . . . . 172 viii TAB LE PAGE 35. Summary of F Ratios for Main Effect of Area of Achievement for Behavioral Task Data of IPS Success-MPF Failure Males . . . . . . . . . 173 36. Summary of F Ratios for Main Effect of Area of Achievement for Behavioral Task Data of IPS Success-MPF Success Males . . . . . . . . . 173 37. Summary of 2 x 2 x 2 MMCS MANOVA Results for Failure Ability, Success Context, Failure Context, Success Luck and Failure Luck Attribution Scores. . . . . . . . . . . . . . . 174 38. Summary of 2 x 2 x 2 Behavioral Task MANOVA Results for Task Difficulty and Luck Scores of All Subjects . . . . . . . . . . . . . . ... 174 39. Summary of 2 x 2 x 2 Behavioral Task MANOVA Results for Task Difficulty and Luck Scores of IPS Failure-MPF Failure Subjects . . . . . . 175 40. Summary of 2 x 2 x 2 Behavioral Task MANOVA Results for Task Difficulty and Luck Scores of IPS Failure-MPF Success Subjects . . . . . . 175 41. Summary of 2 x 2 x 2 Behavioral Task MANOVA Results for Task Difficulty and Luck Scores of IPS Success-MPF Failure Subjects . . . . . . 176 42. Summary of 2 x 2 x 2 Behavioral Task MANOVA Results for Task Difficulty and Luck Scores of IPS Success—MPF Success Subjects . . . . . . 176 43. 2 x 4 Contingency Table and Chi-square Values for Tests of the Relationship Between Sex and Category of Highest Attribution Score for the Interpersonal Area. . . . . . . . . . . . . . . 177 44. 2 x 2 Contingency Table and Chi-square Values for Tests of the Relationship Between Sex and Category of Higher Internal vs. External Attribution Score for the Interpersonal Area. . 178 45. 2 x 4 Contingency Table and Chi-square Values for Tests of the Relationship Between Sex and Category of Highest Attribution Score for the ACademic Area . . . . . . . . . . . . . . . . . 179 ix TABLE 46. PAGE 2 x 2 Contingency Table and Chi-square Values for Tests of the Relationship Between Sex and Category of Higher Internal vs. External Attribution Score for the Academic Area. . . . 180 INTRODUCTION A growing body of research indicates that cognitive variables are important in understanding a wide range of human behavior, including achievement behavior. People's beliefs about why they succeed or fail have been shown to have significant implications for their affective responses to outcome, future expectations and behavior (Weiner, Frieze, Kukla, Reed, Rest & Rosenbaum, 1971). While attri- bution theory has contributed to our understanding of achievement-related behavior, it has also raised some serious concerns. Though there have been some ambiguities in the data, self-derogatory patterns of attribution and expectancy that have negative implications for the fulfill- ment of intellectual and occupational potential have fre- quently been found among women. Such dysfunctional patterns have not generally been reported among men. Though these sex differences in the perception of causality have been found among subjects of many ages on a variety of tasks (Frieze, 1975), our picture of patterns of attribution and expectancy among males and females is in- complete. Attribution research has primarily been con- ducted in traditionally masculine areas of achievement. We may have accurate data on the typical attributions and l expectancies of men and women in areas in which men are encouraged to develop independent mastery, but the picture could be unrepresentative of cognitions about achievement in domains culturally defined as feminine. The purpose of the present study was to contribute to a more broadly based understanding of achievement attribu— tions and expectancy of success by comparing patterns among males and females in an area culturally defined as feminine to patterns in an area culturally defined as masculine. The study was intended to help assess the degree to which the debilitating attribution patterns and low expectancies of success reported among females gen- eralize to areas of behavior outside traditionally mascu- line areas of achievement. Obtaining a more inclusive picture of the process of assigning causality is seen as a preliminary step in developing attributional retraining programs to alter self-derogatory and achievement-inhibiting patterns of attribution and expectancy. REVIEW OF THE LITERATURE Construct of Locus of Control The study of the process of assigning causality developed from a social learning theory framework. Rotter (1966) conceptualized locus of control as the individual's generalized perception of the causal connection between personal characteristics or personal behavior and rein- forcement. He theorized that people considered outcomes to be due to causes originating within them, such as ability or effort, or to causes external to them, such as luck or task difficulty. The development by Rotter (1966) of a measure of generalized expectancy of internal or external control gave rise to a proliferation of research on the construct, primarily considering locus of control as an individual difference variable. Summaries of a variety of experimental, survey and life situation studies of internal- external locus of control (Lefcourt, 1976; Phares, 1976) indicate that individuals who characteristically make in- ternal attributions are more likely to pursue and use infor- mation related to their own well-being, show a higher level of coping and activity in a variety of situations and are more likely to express affects indicative of life satisfac- tion than individuals who characteristically make external 3 attributions. The perception of internal control has also been reported by a number of researchers to be positively correlated with academic achievement (Crandall, Katkovsky & Crandall, 1965; McGhee & Crandall, 1968; Messer, 1972), though there have been some contradictory findings (Butterfield, 1964; Hjelle, 1970). The importance of locus of control as a mediator of life situations is docu- mented in such research as the work of Langer and Rodin (1976) in which perceived control of availability of rein- forcements was found to be positively related to activity, happiness and length of life among nursing home residents. Despite the generally more positive consequences of an internal orientation, these favorable implications of internal perception of control are modulated by contextual variables. While internal attributions generally produce more positive feelings in response to success than external attributions, they also produce more negative affect in response to failure (Weiner, Heckhausen, Meyer & Cook, 1972). Among individuals who experience frequent success, internal locus of control would lead to positive feelings about self and a high level of activity in most situations. For individuals in milieus in which they actually possess little control, or individuals who experience frequent failure, perceptions of internal control could lead to negative feelings or resignation (Lefcourt, 1980). Separate consideration of success and failure situations, which was not included in Rotter's (1966) measure of locus of control, is necessary to understand the specific implica- tions of locus of control. The Weiner Model of Causal Attribution While the Rotter (1966) concept of the perception of causality deals only with whether causes originate within the individual or in the environment, Weiner and his associates (1971) have developed a more differentiated model. The schema formulated by Weiner incorporates the dimension of stability versus instability in addition to the internal-external dimension, and considers attributions for success and failure separately. The four causal cate- gories of the model are (a) internal stable, (b) external stable, (c) internal unstable, and (d) external unstable. Though a diversity of causes may be employed to explain outcome, the attributions usually cited which correspond to the categories above are (a) ability or lack of ability, (b) effort or lack of effort, (c) contextual variables such as task ease or task difficulty, and (d) good or bad luck. While the one dimensional locus of control model has import for affect in response to outcome, the addition of the stability dimension generates implica- tions for expectancy of future outcome as well. Stable attributions for success and failure produce expectations of the same outcome in future attempts at a given task, while unstable attributions present the possibility of different outcomes in the future (Weiner, et al., 1972). weiner's two-dimensional model for classifying attri- butions for success and failure affords a much clearer understanding and prediction of behavior based on the perception of causality than the one-dimensional model. Within Weiner's conceptual framework, various causal attri- butions have specific implications for affect, perceived control, expectancy and future behavior. Attributions of success to the stable, internal factor of ability would maximize pride, certainty of future success and feelings of personal control. Success ascribed to the unstable cause of effort would evoke pride, but no security about future success. The effort attribution would carry the implication of control over future outcome through con- tinued effort. Imputed to external factors, success produces little pride and no perception of personal control. While success due to task ease would lead to expectancy of continuing success at the same task, the external unstable attribution of luck would provide no certainty of future outcome. In response to a failure experience, internal causal attributions would be hypothesized to generate more nega- tive affect than external attributions. The stable attri- bution of lack of ability, in specific, would be most negative because the individual would perceive her/himself as having no control over the situation and would expect continued failure. Ascription of failure to the unstable factor of lack of effort would produce less shame. It implies control over future outcome and the possibility of success despite current failure. Though failure due to external factors of bad luck or task difficulty would produce little negative affect because of the absence of personal responsibility, the attributions assume no con- trol over the situation on the part of the individual. While the expectancy of different outcome on a task in the future is precluded by the stable task difficulty attri- bution, expectancy of success on a task at another time may accompany a luck attribution for failure, since luck could change (Frieze, 1975). Causal attributions also have consequences for the individual's tendency to reapproach an activity or similar activities in the future. The ascription of failure to lack of ability would be most likely to preclude future attempts at an activity, due to the shame associated with the failure and the expectancy of future failure. However, any attribution minimizing the perceived control would be expected to inhibit reapproaching an activity after failure, to some degree. The degree of constraint would depend upon the stability of the cause. While an individual attri- buting failure to task difficulty would be unlikely to attempt the task again, another attempt might be considered if the failure was imputed to luck. It seems that the probability of an individual reapproaching an activity would be maximized by attributions of ability and effort for success, which enhance pride and carry the implica- tion of future control and the expectancy of success. The attribution of failure to lack of effort would also be likely to promote reapproaching a task. Despite associated negative affect, the effort attribution assumes control of future outcome, and the expectancy of success if effort is increased. Empirical support for the Weiner model. Hypotheses about the relationships of expectancy, affect and future behavior to attributions have been widely investigated and generally supported. Research by Weiner, Heckenhaus, Meyer and Cook (1972) substantiated the theoretical rela- tionship defined above between stability of causal attri- bution and expectancy in achievement situations. Supple- menting evidence of greater intensity of affect associated with internal than external attributions (Weiner, 1972) was a study conducted by Weiner, Russel and Lerman (1976). Subjects were asked to rate the intensity of a variety of emotional responses of a story character to success or failure, given various attributions. Support was provided for the relationship predicted by Weiner's two-dimensional model of causality between attribution and the duration, intensity and type of affect. In an earlier study, Fitch (1970) also found a relationship between enduring affect and attributional tendency consistent with Weiner's theory. Fitch reported that males low in self-esteem were more likely to attribute failure internally, a causal attribu- tion maximizing shame, while males with high self-esteem generally ascribed success to internal causes. The achievement motivation research with males gen- erally corroborates the predictions of Weiner's model for the relationship between attribution and the tendency to reapproach an activity or to approach a new activity (Weiner, 1970; 1972). The application of the two-dimensional model to the achievement literature clarified the apparently contradictory findings regarding the relationship of locus of control to achievement. Kukla (1972) reported that men with high achievement motivation tend to attribute success to both high ability and effort, and to ascribe failure to lack of effort. Failure appears to increase motivation to succeed in high achievement males (Weiner, 1972). Thus motivation to achieve excellence in challenging tasks is facilitated by attributions that maximize pride for success and personal control over future outcome following failure. Low achievement males are less likely to attribute success to internal causes and they frequently ascribe failure to low ability (Weiner & Kukla, 1970; Weiner & Potepan, 1970). Such attributions produce little positive affect with success and no expectancy of change after failure, creating 10 minimal motivation to approach challenging tasks. Thus internal attributions can have very different implications for affect, expectancy and achievement behavior, depending upon whether they are used to explain success or failure and whether they are stable or unstable. In studies with children doing math problems, Dweck and her colleagues (Deiner & Dweck, 1973; Dweck & Reppuci, 1973) observed relationships between causal ascription and response to failure fairly consistent with the findings among adult males. Children whose performance deteriorated most or who were less persistent in the face of failure were inclined to blame their failure primarily on uncon- trollable, external factors, or on lack of ability. In contrast, children who persisted and maintained or sur- passed their previous levels of sophistication in problem solving following failure emphasized internal, controllable factors such as effort. Clearly, there is substantial empirical support for the theoretical relationships between attribution and affect, expectancy and future goal-directed behavior de- rived from the two-dimensional model of causality developed by Weiner. While Weiner (1972) hypothesizes that achieve- ment motivation results from learned patterns of attribu- tion, it should be noted that the direction of the relation- ship has not yet been established (Frieze, 1975). ll Sex_Differences in Expectancy and the Perception of Causality The study of the process of assigning causality has proven a particularly useful approach to understanding the well-documented lower level of achievement among women than among men. Though causal directions have not been estab- lished between achievement and attribution, there is evi- dence of attributional and expectancy patterns among women that would have very negative implications for achievement and for feelings of self-worth. Such patterns are not as typically found among men. Sex differences in three ele- ments of the attributional process contribute to the under- standing of the relatively low achievement found among women: (a) personal expectancies for success, (b) evalua- tions and attributions regarding the performance of others based on sex, and (c) categories of causal attribution typically employed (Frieze, 1975). Among subjects of a wide range of ages in studies employing a variety of achievement tasks and methodologies, males have been shown to hold higher expectancies for success than females. Crandall (1969) found females to make lower estimates of their future performance than males among grade school children regarding novel intel- lectual task, among college students regarding grades and among young adults regarding recall tasks. Such sex dif- ferences on academic and intellectual tasks have been 12 substantiated by other researchers (Frieze, 1975; Lenney, 1977). Higher anticipation of success in males than in females has also been reported on manual dexterity tasks (Montarelli & Hill, 1969; Rychlak & Eacker, 1962; Rychlak & Lerner, 1965) and in imagining success in seven different professions (Feldman-Summers & Kiesler, 1974). Relative to their actual level of ability or performance history, females tend to underestimate their future success while males tend to overestimate their future success. The under- estimation of their future performance by females seems par- ticularly inappropriate in academic situations, when girls on the average are far more successful than boys in ele- mentary school grades and reading achievement, receive less criticism from teachers and believe that they are more highly regarded by teachers than boys (Dweck, Goetz & Strauss, 1977). From a reanalysis of the achievement literature, Lenney (1977) concluded that females do not display lower self-confidence than males in all achievement settings. There is evidence that at least three situational variables determine women's self-confidence in achievement situations: (a) the nature of the task, (b) the availability of per- formance feedback, and (c) the emphasis on social compar- ison. The expectancies of females for their performance have been found to be equal to the expectancies of males, or higher, on tasks that are labeled feminine. Girls and 13 women also display higher self-confidence in an activity when performance is presented as being unimportant than when it is presented as being important (Nicholls, 1975). Though women consistently expect to do less well than men if minimal or ambiguous feedback on a task is given, Lenney points out that sex differences have not been re- ported when unambiguous feedback is given. Females' expectancies for their performance appear to be lower when the situation emphasizes the presence of the social environ- ment, such as when a socially defined standard of perfor- mance is given, or when there is direct competition with another individual (Lenney, 1977). While Lenney's reeval- uation of the literature suggests that sex differences in expectancy do not exist in all situations, the preponderance of the achievement research indicates that females hold lower expectancies for success than males. Differences between males and females in personal expectancies for success could easily derive from more general expectations based on prevalent sex-role stereo- types. People appear to hold expectancies founded on male and female stereotypes that influence their perception and evaluation of the performance of others. WOmen are perceived by a large variety of people as less competent, active, self-confident, independent, logical and objective than men (Rosenkrantz, Vogel, Bee, Broverman & Broverman, 1968). In apparent acceptance of this stereotype, female 14 subjects have been found to evaluate written articles and paintings lower when they are attributed to women than when the same articles and paintings are attributed to men (Goldberg, 1968; Pheterson, Kiesler & Goldberg, 1971). Deaux and Emswiller (1974) found that subjects of both sexes attributed a man's success on a masculine task to skill, while a woman's success on the same task was ascribed to luck. Frieze (1975) suggested that some evidence of women giving higher ratings to competent women than to competent men (Deaux & Taynor, 1971) and higher ratings to female-written articles than to male-written articles (Morris, 1970) may reflect increased supportive- ness of women for women fostered by the Women's Movement. At this point, there is little evidence in the literature of change in male perception of women's competence (Frieze, 1975). The stereotype-based devaluation of female abil- ities appears to be internalized by females and reflected in low success expectancies on many tasks. A third area of sex difference is the categories of causality that males and females habitually assign to explain their successses and failures. Several studies have indicated that women use luck as a causal ascription for both success and failure in traditionally masculine achievement areas more than men do (Bar-Tal & Frieze, 1973; Feather, 1969; McMahan, 1972, cited in Frieze, 1975; Simon & Feather, 1973). WOmen have been found to be less 15 likely than men to attribute success in academic achievement to their ability (Frieze, 1973; McMahan, 1971, cited in Frieze, 1975). When considering only the internal-external dimension of causality, Rotter and Hochreich (1973, cited in Frieze, 1975) reported that college women, unlike college men, are consistently more external than internal in their perception of causality. There also are data contradicting the generalization that women are inclined to make external attributions. Crandall and associates (1965) found women to be more internal than external for failure. McMahan (1971; 1972, cited in Frieze, 1975) and Lefcourt, Von Baeyer, Ware and Cox (1979) have also reported that women are more likely than men to attribute failure in academic achievement to lack of ability. Feather (1969) and Frieze (1973) failed to replicate findings of greater internality for failure among females. Evidence from work by Dweck and Reppucci (1973) that grade school girls are more likely than boys to attribute failure both to lack of ability and to external factors suggests that the contradictory findings may merely reflect the fact that females use both types of attribution to explain failure. It seems probable that the situational variables specified by Lenney (1977) as determinants of expectancy may also influence whether lack of ability or external ascriptions for failure are made. In contrasts to females, there is evidence that males most 16 frequently attribute failure to task difficulty (Deaux, 1976). Such patterns of expectancy and causal attribution appear to be tailored to interfere with women's fulfill- ment of their intellectual and professional potential. Research indicates that individuals who expect to do better on achievement tasks actually evidence superior performance to individuals who hold lower expectations (Battle, 1965; Feather, 1966; Tyler, 1958). Thus females' lower self- expectations based on stereotypes are likely to be fulfilled. Frieze (1978) and Deaux (1976) hypothesize that expectancy influences the individual's interpretation of the causes for outcome. Deaux cites a number of studies in which subjects with high expectancies for success ascribed success inter- nally and failure externally, while subjects with low ex- pectations for success did the opposite. The assumption of a relationship between expectancy and the perception of causality leads to a number of predictions about the in- fluence of the characteristically low expectancies of females on the attribution process and achievement behavior (Deaux, 1976; Frieze, 1978). If a woman has low expec- tations for success based on internalized stereotypes of female incompetence, success would probably be attributed to temporary causes because it is inconsistent with expec- tations. Therefore the success experience would not in- crease expectancy of future success, nor would it enhance 17 the tendency to reapproach the task to the degree that it would if success were ascribed to stable or controllable factors. If the expectancy of success is high, as has generally been reported with males, the pride and expectancy- enhancing attribution of success to ability is likely to be made, because it is consistent with expectations. When a woman with low expectancy of success fails on a task, it confirms her expectations and reaffirms her belief in low ability or lack of control over outcome. The lack of ability ascription for failure frequently employed by females would give rise to negative feelings about self and low expectancy of future success. Though luck, the other category of attribution reportedly used by women to account for failure, would not produce much negative affect, it would not enhance the expectancy of future success either. Despite the possibility that luck will change, failure ascribed to luck would not greatly motivate reapproaching the activity because there is no perception of control over future outcome, unlike when failure is attributed to lack of effort. Thus patterns found among women of assigning low ability or luck as the cause of failure could be expected to result in resigna- tion rather than accelerated effort in response to failure. Given the difference in evaluations and attributions about the performance of others that people tend to make on the basis of sex, women's low expectancies for success 18 and their self-derogatory attributions, the negative feelings of women about themselves and their competence are not surprising. These internal barriers to achieve- ment are reinforced by the frequent adverse social con- sequences of achievement for females (Donelson & Gullahorn, 1977). The low level of achievement of most women compared to men is not difficult to understand. Consideration of Achievement from a Feminine Perspective These conclusions about the negative implications of the findings of attribution theory research for achieve- ment among women are limited in that they consider achieve- ment only from a masculine perspective. Achievement has conventionally been defined in the psychological literature as, "...the competition with a standard of excellence, or the desire to accomplish something difficult, to master objects, people or ideas, and to do so as rapidly and independently as possible.” (Donelson & Gullahorn, 1977). However, this has been translated to mean the pursuit of excellence in areas such as academics and career. While males are encouraged and rewarded for mastery in such areas, independent striving for competence in such domains is frequently discouraged in females. Achievement motiva- tion has been operationalized as a masculine characteristic which, despite the impact of the Women's Movement, has limited applicability for many women (Stein & Bailey, 1975). 19 Theories and research which attempt to define the essential psychological dimensions distinguishing between women and men suggest that achievement may be more appropriately assessed in women within the domain of interpersonal behavior. Dominance of the interpersonal orientation among females. Bakan (1966) proposed that two fundamental orientations characterize all living organisms: a sense of communion and a sense of agency. Communion reflects selflessness, a concern with others and a desire to be at one with others. Agency describes a sense of self, mani- fest in self-assertion, self-protection and self-expansion. While Bakan (1966) postulates that both modalities are in all human beings, he identifies agency with masculinity and communion with femininity. A number of other writers have proposed similar principles of masculinity and femininity (Block, 1973; Carlson, 1971; Donelson & Gullahorn, 1977; Spence & Helmreich, 1978). Empirical investigations indicate that despite marked changes in the past 15 years in traditional sex role patterns, there continues to exist across sex, religion, marital status and educational level, a strong consensus about the psychological attributes distinguishing between males and females that is very consistent with Bakan's formulation (Bem, 1974; Ellman, Press & Rosenkrantz, 1974; Rosenkrantz, Vogel, Bee, Broverman & Broverman, 1968; 20 Spence, Helmreich & Stapp, 1974). The ideal man is perceived as high in competence, independence, activity and dominance, characteristics subsumed under Bakan's sense of agency. The attributes of the ideal woman, emotional warmth, expressiveness and sensitivity to the feelings of others, are quite congruent with Bakan's communal orientation. The anthropological literature indicates that the majority of cultures produce women who evidence more communal characteristics than agential characteristics, and men who evidence more agential attributes (Spence & Helmreich, 1978). The encouragement of females to develop a communal orientation, frequently at the expense of the development of agential attributes, begins early in the socialization process. Parents appear to have clearly defined notions of what is typical and appropriate behavior for boys and girls from when their child is very young. For example, interviews of the parents of four-year-old boys and girls revealed that parents expected girls to have a greater interest in people than boys (Lambert, Yackley & Hein, 1971; Parsons, Frieze & Ruble, 1976). When mothers were asked when they would expect or permit certain behaviors of their child, mothers of girls responded with signifi- cantly later ages for independence granting than mothers of boys, and showed a similar nonsignificant trend for achievement induction (Collard, 1974; Hoffman, 1975). 21 Research findings that girls are more obedient, cooperative, responsible and generally better socialized than boys of the same age (Donelson & Gullahorn, 1977) reflect the greater emphasis on the development of socially pleasing, other-oriented behavior in girls. Bronfenbrenner (1961) theorizes that girls may receive more love-oriented dis- cipline that would discourage attempts at independence and mastery and lead to an interpersonal orientation. Hoffman (1975) postulates that while little boys are developing mastery and confidence in coping with the environment, and learning instrumental independence, little girls are learning to use their interpersonal relationships in an instrumental way. Parental stereotype-based expec- tations may result in greater protection and insufficient independence training, significantly limiting the girl's development of a sense of personal competence. Findings of Crandall and Rabson (1960) that, in free play, grade school girls are more likely than boys to withdraw from threatening situations and to seek help from adults or peers, supports the hypothesis that girls learn to be effective through eliciting the help and protection of others. The previously discussed evidence that females underestimate their ability and hold much lower expecta- tions for success in conventional achievement areas than males (Frieze, 1975) is congruent with the proposition 22 that girls develop a limited sense of personal competence in many areas of activity. Research indicates that, from grade school on, females become increasingly communally oriented. By adolescence, their focal concerns are very different from the focal concerns of boys (Carlson, 1965; Rosenberg, 1965). Adolescent girls are more peOple-oriented and assign top value to traits of being well-liked and maintaining inter- personal harmony, while boys are more concerned with per- sonal achievement and competence (Douvan & Adelson, 1966; Rosenberg, 1965; Rosenberg & Simmons, 1975). Research indicates that while male ego-identity formation can generally be explained in terms of crisis and committment in the areas of occupation and ideology, female ego-identity formation focuses on issues of interpersonal relationships (Constantinople, 1969; Hodgson, 1971; Marcia, 1980). The self-images of college women are primarily interpersonal and the self-images of college men are largely impersonal and individualistic (Carlson, 1971). In a cross-cultural study, Block (1973) found that women valued communal traits more highly for themselves while men valued agential traits more highly. Application of male-based concepts of achievement to females. Despite such a contrast in the general importance of agential and communal domains for males and females, constructs of achievement motivation derived from the 23 observation of males performing agential tasks have been extended to females. This has resulted in some confusing findings. While most data indicates that achievement motivation is positively correlated with academic per- formance, competitiveness and independence in white middle class males (McClelland, Atkinson, Clark & Lowell, 1953), traditional measures of achievement motivation do not correlate well with actual achievement or competitiveness in females in academic settings (Entwhistle, 1972). Arousal techniques which stress intelligence and leadership do not result in increased achievement imagery with women (Veroff, Wilcox & Crandall, 1953). The inapplicability of tradi- tional achievement motivation measures to females has led some theorists to conclude that women are not motivated to achieve. Veroff (1969) proposed a developmental theory of achievement motivation, attempting to account for the ultimately lower level of achievement by women than by men. He postulated that although both sexes are externally motivated to achieve when young, males progress to a stage of internal motivation, but females do not. Similarly, Crandall (1963) hypothesized that the apparent achievement behavior that females do manifest is motivated by the social approval it will bring, not by intrinsic pleasure in mastery. Interpersonal skill as the expression of achievement. Stein and Bailey (1975) question the validity of such 24 conclusions about achievement. They postulate that the attempt to apply male-based constructs and measures of achievement motivation to females has led to misinterpre- tations of female behavior. Stein and Bailey (1975) hypothesize that because of cultural definitions of ”feminine," women and girls express their motivation to achieve in different activities than males express their motivation, in activities socially prescribed for females. Hoffman (1975) assumes that female use of interpersonal behavior in an instrumental manner interferes with the development of skills for independent mastery of the environment which characterizes achievement behavior. However, Stein and Bailey (1975) define interpersonal interaction as the primary traditional area of feminine skill and achievement. A number of other authors have postulated that women frequently translate their achieve- ment strivings into conventionally feminine areas such as social skill or homemaking because they are discouraged from achieving in traditionally masculine domains (Frieze, 1975; Hoffman, 1972; Horner, 1972; Tangri, 1972). There is some empirical support for the hypothesis that females express achievement motivation in areas culturally defined as feminine. Social skill arousal has been shown to increase motivation in women with traditional values (French & Lesser, 1964). Friedrich and Harding (1968) reported feminine role achievement to be positively 25 correlated with grade average and measures of efforts on a test of social skill. In addition, Stein and Bailey's perspective on the expression of achievement motivation offers a viable reinterpretation of some earlier work on achievement. Stein and Bailey postulate that because the primary area of feminine skill has traditionally been social skill, what Crandall (1963) and others have labeled affiliation motivation or affiliative needs may be more accurately conceptualized as achievement within the feminine role. The authors interpret findings that social arousal techniques with women lead to achievement imagery as evidence that achievement, not affiliation, is the motivating factor. If female achievement efforts were based on the desire for social approval rather than the desire for the attainment of a standard of excellence, one would expect social reinforcement or the absence of social reinforcement to influence sex differences on experimental task performance, but it does not (Lenney, 1977; Stein & Bailey, 1975). Evidence that nursery school girls who attempt to obtain recognition for achievement also make more attempts to gain love and affiliation has been interpreted by some to mean that girls use achievement as a means of obtaining love and affection (Tyler, Rafferty & Tyler, 1962). An alternate explanation is that both the efforts to obtain recognition for achievement and to obtain affiliation were 26 manifestations of achievement motivation. With age, the more masculine expressions of achievement motivation by girls may be eliminated by socialization, while the expression of achievement through interpersonal skill continues. Though increasing freedom to participate in fields of endeavor previously restricted primarily to men makes it possible for women and girls to manifest achieve- ment motivation by altering or abandoning the traditional concepts of femininity, achievement through conventionally feminine channels gives rise to much less conflict. While Stein and Bailey's (1975) conceptualization of interpersonal interaction as the traditional primary area of feminine skill and achievement is very useful, the con-- struct of achievement must be applied to the interpersonal domain with care. Though the definition previously cited specified accomplishments with people as an aspect of achievement, achievement as it is conventionally defined and interpreted would carry agential implications of mastery or power over people. Interpersonal achievement from a communal orientation, in contrast, would suggest a striving for excellence in interpersonal skill without implications of dominance over others. Attribution and Expectancy in Traditionally Feminine Areas In light of Stein and Bailey's reinterpretation of the achievement data as indicating that motivational 27 patterns for females and males are similar, but are expressed in different contexts, it seems likely that both men and women might evidence different attributions and expectancies in traditionally feminine domains than they have shown in masculine domains. The conclusions drawn by Lenney (1977) in her reanalysis of the achieve- ment literature cited earlier support this hypothesis with regard to expectancy. Sex-linking of the task is one of the situational variables determining the self-expectations of women for their performance in achievement situations. Stein, Pohly and Mueller (1971) reported that sixth grade girls expected to perform best on tests that would predict future performance in "girl's school subjects," while boys held the opposite expectation. Deaux and Farris (1974, cited in Lenney, 1977) found no sex difference in expec- tancy when an anagram task was labeled feminine, though women's expectations were lower than men's when the anagram task was sex-linked as masculine. Thus there is some evidence that when tasks are labeled feminine, expectancies of females for performance seem to be as high or higher than the expectancies of males. The results of the few studies in which attributions have been assessed in traditionally feminine areas lend support to the hypothesis that males and females tend to have different beliefs about causality in feminine domains than have been reported in masculine domains. Goetz and 28 Dweck (1979) examined the attributions of grade school children in the face of social rejection. They failed to find a significant sex difference in attributions. Lefcourt and his associates (1979) investigated attribu- tions of college students with regard to affiliation out- come and academic achievement outcome, using a scale on which subjects endorsed causes for outcomes in hypothetical situations. In contrast with the sex differences typically reported in masculine domains, Lefcourt found males sig- nificantly more likely than females to attribute affilia- tion outcome to luck and to ascribe success in affiliation to external causes. Lefcourt noted that his findings must be replicated before reliability of male and female patterns on his scales can be assumed because of the large number of comparisons made in the analysis from which these results were obtained, and the small number of items in the sub- scales representing each category of attribution. Lefcourt (et al., 1979) also examined the relationship of attributional patterns for affiliation and academic achievement in college students to specific behavior in interpersonal situations and in an academic achievement related situation. He found significant correlations for college males between attributional patterns regarding academic achievement and behavior during the academic task. Fidget-like motions indicating distress while doing anagrams of intermediate difficulty were negatively correlated with 29 externality for academic achievement. With more difficult anagrams, the negative correlations were stronger. Sig- nificant correlations between these variables were not found among females. In interpersonal situations, Lefcourt et a1. (1979) found different behaviors and self-ratings to be correlated with patterns of affiliation attribution for men and women. Among females, attribution of affiliation outcome in general to effort, and attribution of success in affilia- tion situations to internal causes, were positively cor- related with self-disclosure in dyadic discussions between same-sexed strangers. There were no significant relation- ships between attributions and behavior in the dyadic interaction among males, though the results were in the same direction as with females. In another study using dyads of same-sexed subjects, Lefcourt and his colleagues (1979) measured head nodding while listening, speech dura- tion when speaking and a number of ratings of comfort and intimacy made by subjects after the interaction. Among males, nodding while listening was positively correlated with attributions of affiliation outcome to ability. Speech duration of the other member of the dyad was posi- tively associated with the listener's endorsement of affil- iation effort attribution and negatively related to affili- ation context attributions. The speaker's rating of intimacy of interaction was positively related to the 30 listener's endorsement of effort attributions for affiliation. No significant relationships between affiliation attributions and the behaviors and self- ratings assessed in the second study were found among females. The work by Lefcourt reveals some interesting sex differences both in typical perceptions of causality and in behaviors that might be predicted for males and females from their attributional patterns. The findings suggest that attributions may have differing relevance for the behavior of men and women. It is particularly interesting that no relationship was found between attributional pat- terns and behaviors measured in the academic situation for females. It would be useful to know whether the percep- tions of causality indicated by subjects in the hypo- thetical situations on the paper and pencil measure would be replicated in actual academic and affiliation achieve- ment situations. Need for Goal-specific-Measures of Attribution In approaching research in the area of locus of control/attribution theory, a methodological problem in the way the concept has been operationalized in the literature needs to be considered. A good deal of the research is based on a generalized measure of locus of control, Rotter's I-E Scale (1966), with an assumption 31 that the perception of personal control over outcome is a unitary factor rather than multidimensional. Rotter's measure is a 24-item scale which samples beliefs about personal control in achievement, affiliation, business and world affairs. Factor analyses of the scale have failed to reveal clusters of items around reinforcement concerns (Lefcourt et al., 1979). Though the Rotter I-E Scale was designed as a general measure of locus of control, investi- gators have attempted to predict a diversity of specific behaviors and personality attributes in both males and females based on this measure. The Rotter measure has been found to have little pre- dictive utility for females (Lefcourt et al., 1979). Per- haps items tapping such areas as business, world affairs and traditional achievement were not meaningful for most women, so that internality or externality on these items had few implications for women's generalized expectancy of personal control. The scale does appear to be fairly robust in reflecting a generalized sense of personal con- trol in males. The majority of the items of Rotter's scale appear to be concerned with agential values. Despite the difficulty in predicting women's behavior from the scale, such conclusions regarding women as the statement that college women are consistently more external than college men (Rotter & Hochreich, 1973, cited in Frieze, 1975) have been drawn from research using the Rotter I-E Scale. 32 In his early work, Rotter (1954) made it explicit that his formula for behavior potential referred to specifics. When he discussed behavior potential as a function of expectancy and value of reinforcement, he defined expectancy as, "...the probability held by the individual that a particular reinforcement will occur as a function of a single specific behavior on his part in a specific situation or situations," (p. 165). It is only in situations in which the individual has not had some prior experience that more generalized expectancies will take precedence. Following Rotter's original formulation, Lefcourt and his colleagues (1979) make the point that if one wishes to investigate the relationship of locus of control to a specific criterion, the optimal strategy would be to develop a locus of control measure for the particular criterion. Some strong support for the use of goal-specific measures of locus of control derives from research with the attribution measure developed by Lefcourt et al. (1979) and employed in the studies by Lefcourt cited earlier, the Multidimensional-Multiattributional Causality Scale (MMCS). The MMCS incorporates the dimensions of internal-external and stable-unstable, assessing attribu- tions for both success and failure. There are two sub- scales, one pertaining to academic achievement and the other to affiliation. 33 The discriminant validity of Lefcourt's (et al., 1979) goal-specific measure has been established. Items from the academic scale have an average correlation of r=.31 (§;241) with the total academic scale, but an average correlation of r=.10 with the total affiliation scale. The average correlation between affiliation items and the total affili- ation scale was r=.31. The same items correlated r=.09 on the average with the total academic scale. The academic achievement locus of control scale was found to be related to behavior indicating discomfort during an achievement task, while the affiliation scale was unrelated. The affiliation locus of control scale made possible the pre- diction of a number of social interaction criteria. There was no significant relationship of the academic scale to the criteria. A few other investigators have employed goal-specific measures of locus of control. Crandall, Katkovsky and Crandall (1965) developed the first scale designed to assess the perception of control in a specific area. Their research with the Intellectual Achievement Responsibility scale for children demonstrated the usefulness of a goal- specific measure in predicting academic achievement behavior. It also pointed to the need to distinguish attributions for success from attributions for failure. The discriminant validity of a goal-specific locus of control measure in contrast to Rotter's general I-E scale 34 was supported by work on a health-related locus of control measure devised by Wallston, Maides and Wallston (1976) and Wallston, Wallston, Kaplan and Maides (1976). Indi- viduals expressing some measure of personal control over health were found to be more likely to seek out health related information and to show more preference for self- directed weight reduction programs over group plans than individuals with the perception of less personal control over health. Rotter's scale was unrelated to the criterion. Reid and Ziegler (1977) found that the perception of per- sonal control over daily events which elderly people be- lieved were important determinants of their happiness, contentment and adjustment was, in interaction with value of these satisfactions, highly related to measures of life satisfaction and ratings made by others of zest for life and assertiveness. Though the majority of work on locus of control has been carried out with a generalized measure, the available literature demonstrates the value of using goal-specific measures of the personal perception of causality to make predictions about behavior associated with a specific goal area. The Problem Weiner's two-dimensional model of causal attribution would predict that high expectancies of success and 35 attributions of failure to lack of effort and success to ability and effort would maximize achievement. The attri- butional patterns and expectancies of women typically re- ported in the literature have negative implications for achievement. Frequently reported among women are low ex- pectancies of success, attributions of success to external factors, and attributions of failure to lack of ability or external factors. However, generalizations in the liter- ature about the perception of causality in males and fe- males seem to have been made with little consideration of their different socialization toward agential and communal values. The attribution theory research has primarily been conducted in traditionally masculine domains, such as academics. While conclusions about typical patterns of attribution among females and males may be valid for tradi- tionally masculine areas of achievement, the patterns may be unrepresentative of the attributional process for the full range of human behaviors. Research on socialization and identity formation among females and the theoretical work of Stein and Bailey (1975) indicate that interpersonal behavior is the primary area in which women learn to express their personal competence and motivation to achieve in our society, while achievement in masculine domains is frequently inhibited. Given the differences in the socialization of females and males, the findings that males hold higher expectancies for success 36 and make more achievement promoting attributions than females in the conventionally masculine domains in which most of the attribution theory research has been conducted are predictable. The perception of causality in tradi- tionally feminine areas has not been well assessed (Stein & Bailey, 1975), but one might predict that women would be more likely to make achievement promoting attributions in domains culturally sanctioned for women. The small amount of research available suggests that the patterns of attri- bution and expectancy for both males and females may be quite different in the context of feminine activities and interests than the patterns reported in masculine domains. In order to obtain a more representative picture of the process of assigning causality among males and females, and of the negative or positive implications of this pro- cess for them, expectations and causal attributions for success and failure must be assessed in conventionally feminine areas of mastery. The objectives of the present study were: (1) To compare the expectancies and the use of achieve- ment-related attributions among females in a traditionally masculine area to the expectancies and the use of achievement-related attributions among females in a traditionally feminine area. (2) To compare the expectancies and the use of achievement-related attributions among males in a 37 traditionally masculine area to the expectancies and the use of achievement-related attributions among males in a traditionally feminine area. (3) To determine whether past findings of sex differ- ences in expectancy and attribution in tradi- tionally masculine areas of achievement will be replicated. (4) To determine whether males and females differ significantly in their expectancies and attribu- tional patterns in a traditionally feminine area of achievement. Attributions of primary interest in this research were attributions theorized to facilitate achievement: attribu- tions of success to effort and ability and attribution of failure to effort. On the basis of the work of Stein and Bailey (1975), interpersonal behavior was chosen as the conventionally feminine area of achievement in which to assess the perception of causality. Academic achievement was chosen as the conventionally masculine area in which to measure attributions and expectancy, since the pursuit of excellence in academics fits the common operationalization of achievement in agential terms, and because the patterns of sex differences in the attribution process reported in the literature have frequently been based on research with academic or intellectual tasks. 38 To accomplish the specified objectives, the attribu- tions of male and female college students regarding outcome in hypothetical academic and affiliative situations were assessed, using Lefcourt's Multidimensional-Multiattribu- tional Causality Scale (Lefcourt et al., 1979). This phase of the experiment was designed to replicate Lefcourt's research on patterns of attribution among males and females using the MMCS. The attributions and expectancies of sub- jects with regard to two tasks in which they participated were also measured. One task was linked to interpersonal success. The other task was linked to academic success. Because of the possibility of order effects, the order of administration of the two experimental tasks was counter- balanced. In Condition 1, the interpersonal task occurred first. In Condition 2, the academic task occurred first. Hypotheses H1: Females will hold higher expectancies for success and employ achievement promoting attributions more fre- quently in a traditionally feminine area of achieve- ment than in a traditionally masculine area of achievement. Specific predictions: (1) Females will endorse promoting attributions more on the interpersonal scale of the MMCS than on the academic scale of the MMCS. 39 (2) Females will hold higher expectancies for success and employ achievement promoting attributions to account for a greater percentage of outcome on the interpersonal task (Interpersonal Sensitivity Test) than on the academic task (Matching Parts and Figures Test). Males will hold higher expectancies for success and employ achievement promoting attributions more fre- quently in a traditionally masculine area of achieve- ment than in a traditionally feminine area of achieve- ment. . Specific predictions: (1) Males will endorse achievement promoting attribu- tions more on the academic scale of the MMCS than on the interpersonal scale of the MMCS. (2) Males will hold higher expectancies for success and employ achievement promoting attributions to account for a greater percentage of outcome on the academic task than on the interpersonal task. Males will hold higher expectancies for success and employ achievement promoting attributions more fre- quently than females in a traditionally masculine area of achievement. Specific predictions: 40 (l) Males will endorse achievement promoting attribu- tions more than females on the academic scale of the MMCS. (2) Males will hold higher expectancies for success and employ achievement promoting attributions to account for a greater percentage of outcome on the academic task than females. Females will hold as high, or higher, expectancies .5 for success and employ achievement promoting attribu- tions as frequently, or more frequently, than males in a traditionally feminine area of achievement. (1) Females will endorse achievement promoting attri- butions as much as, or more than, males on the interpersonal scale of the MMCS. (2) Females will hold as high, or higher, expectancies for success and employ achievement promoting attributions to account for as great, or a greater, percentage of outcome on the interpersonal task than males. Definition of Terms Achievement promoting attributions: (l) The attribution of success to effort or ability (2) The attribution of failure to effort 41 Independent Variables (1) Sex (2) Area of achievement (a) (b) Traditionally feminine, interpersonal Traditionally masculine, academic (3) Condition: order of administration of the experi- mental tasks (a) (b) Condition 1: interpersonal task (Interpersonal Sensitivity Test) first Condition 2: academic task (Matching Parts and Figures Test) first Dependent Variables (l) Expectancy for success (a) (b) Expectancy for success on the Interpersonal Sensitivity Test: response to the Expectations Questionnaire item, "What percent of the items on the following test do you expect to get correct?" on a scale from 0 to 100% with lo-point intervals. Expectancy for success on the Matching Parts and Figures Test: operationalized as for the Inter- persOnal Sensitivity Test. (2) Use of achievement promoting attributions for outcome (a) In the feminine area of achievement (b) In 1. 42 Scores on the success ability, success effort and failure effort subscales of the inter- personal scale of the MMCS. Percent of outcome on the Interpersonal Sensitivity Test attributed to ability and percent of outcome attributed to effort, as indicated by response to item 4 on the Posttest Questionnaire (Appendix E). the masculine area of achievement Scores on the success ability, success effort and failure effort subscales of the academic scale of the MMCS. Percent of outcome on the Matching Parts and Figures Test attributed to ability and percent of outcome attributed to effort, as indicated by the response to item 4 on the Posttest Questionnaire (Appendix E). METHOD Overview of Study A 2 (Sex of Subject) x 2 (Area of Achievement, a repeated measure) x 2 (Condition, order of tests) factorial design was employed in this study. Table 1 presents an outline of the experimental procedure. Data were collected from mixed-sex groups of undergraduate volunteers. Expec- tancy for success and task-specific and general causal attributions for outcome in a traditionally masculine and a traditionally feminine area were assessed. Academics was designated as the traditionally masculine area. Inter- personal skills was designated as the traditionally feminine area. In order to assess task-specific attributions and expectancies, subjects were asked to participate in a test of interpersonal sensitivity (Interpersonal Sensitivity Test, IPS) and in a test of understanding of spatial rela- tions (Matching Parts and Figures Test, MPF). The order of administration of the two tests was counterbalanced to rule out order effects. Subjects in Condition 1 engaged in the IPS before doing the MPF. In Condition 2, subjects par- ticipated in the MPF first. Conditions 1 and 2 differed only in the order of administration of the interpersonal and academic tasks. 43 44 Table 1 Outline of Procedure in Condition 1 and Condition 2 Condition 1 1. Preliminary Questionnaire 2. Interpersonal task Introduction to IPSa Expectations Question- naire IPS performance Posttest Questionnaire 3. Academic task Introduction to MPF Expectations Question- naire MPF performance Posttest Questionnaire 4. Personal Beliefs MeasureC 5. Personal Characteristics Form 6. Debriefing Condition 2 Preliminary Questionnaire Academic task Introduction to MPFb Expectations Question- naire MPF performance Posttest Questionnaire Interpersonal task Introduction to IPS Expectations Question- ' naire IPS performance Posttest Questionnaire Personal Beliefs Measure Personal Characteristics Form Debriefing aInterpersonal Sensitivity Test, a test in the traditionally feminine ability area of interpersonal skill bMatching Parts and Figures Test, a test in the tradi- tionally masculine ability area of intellectual skill cThe title under which the Multidimensional-Multiattribu- tional Causality Scale was presented to subjects: a measure of general attributions for academic and inter- personal outcome 45 Expectancies for success for each task were measured on the Expectations Questionnaire before subjects began the task. Subsequent to the administration of each task, data on self-estimates of success, attribution for outcome, and effectiveness of the sex linkage manipulation were gathered using the Posttest Questionnaire. In order to assess gen- eral attributions regarding outcome in academic and inter- personal situations, subjects were asked to respond to the Multidimensional-Multiattributional Causality Scale (MMCS, Lefcourt et al., 1979). The MMCS was presented to subjects as the "Personal Beliefs Measure." Table 2 presents defini- tions of the variables derived from each experimental measure which was used in the present study. As a first step in both conditions, subjects completed a Preliminary Questionnaire about their academic record and social interactions. The Personal Characteristics Form, designed to obtain data on parents' educational and occupa- tional backgrounds, parents' achievement expectations for the subject and the subject's career aspirations, was ad- ministered last. Data from the Preliminary Questionnaire and the Personal Characteristics Form were not used in this phase of the research. Subjects The subjects were 146 undergraduate student volunteers, 81 females and 65 males, recruited in spring quarter of 46 Table 2 Definition of Variables Derived from Each Experimental Measure EXPECTATIONS QUESTIONNAIRE Expectancy of success Percentage of items which sub- ject expected to answer correctly, on a scale from 1 to 100 with 10 point intervals. POSTTEST QUESTIONNAIRE Task attribution scores Percentage of outcome which subject attributed to ability, effort, task difficulty and luck for each task. Task success score Sum of scores on three items: a) self-ratings of success compared to expectations on a scale of l to 7, b) self-ratings of success compared to other college students on a scale of l to 7, and c) subject's estimate of percent of items answered correctly. Composite success score was the criteria for success or failure classifi- cation for each task. Manipulation checks a) Estimate of the percent of task items which female college students answered correctly on the average. b) Estimate of the percent of task items which male college students answered correctly on the average. c) Rating of the task as masculine or feminine on a scale of l to 7. PERSONAL BELIEFS MEASURE (Multidimensional- Multiattributional Causality Scale, MMCS) MMCS attribution scale scores Scores on eight Likert scales concerning generaI attributions for academic outcome and eight Likert scales concerning general attributions for interpersonal outcome. There were separate scales for success and failure for four categories of attribution: success ability, failure ability, success effort, failure effort, success context, failure context, success luck and failure luck. 47 1980 from an introductory psychology course at Michigan State University. Though 157 subjects were run, the responses of 11 people were discarded because of missing data. Subjects were called and asked to participate in the experiment under the title, "Personal Beliefs and Success." Extra course credit was given for participation. Experimenters Each session was jointly conducted by one female and one male experimenter. The experimenters alternated in reading the instructions within each session. Procedure A description of each instrument employed in the study will be provided following the discussion of experimental procedure. Copies of the instruments can be found in Appendices B through J. The subjects were tested in mixed sex-groups. They were randomly assigned to Condition 1 or 2. There were five sessions of each condition. Upon their arrival, sub- jects were told that the experiment was designed to examine the relationship of personality characteristics and per- sonal beliefs to the likelihood of success in two different kinds of skills (see Instructions, Appendix A). After the activities involved in the experiment were briefly de- scribed, subjects were asked to read and sign a statement of informed consent. 48 The first experimental task in both conditions was to complete the Preliminary Questionnaire (see Appendix B). In Condition 1, subjects were then prepared for the Interpersonal Sensitivity Test (see Appendix D). The IPS was defined as being related to probable success in the area of interpersonal skills. Subjects were told that there is some reason to believe that women as a group are generally better than men in some interpersonal skills, such as sensitivity to the feelings of others. Subjects were informed that the purpose of the IPS was to assess interpersonal sensitivity in identifying the feelings of other people on the basis of facial cues. The Inter- personal Sensitivity Test consisted of 32 pictures of faces, one of which was a sample. For each picture, sub- jects were asked to respond by writing the name of one, or more than one, emotion that they though the person was feeling. After reading the instructions for the test aloud, one of the experimenters went through the example aloud, suggesting a possible answer. After being told that they would have five minutes to respond to the 31 items once the test began, subjects were asked to fill out the Expectations Questionnaire (see Appendix C) before taking the IPS. When everyone had completed the Expectations Questionnaire, subjects were instructed to begin the IPS. As designed, few subjects were able to complete the IPS in 49 the time allotted. When time was up, the Posttest Questionnaire, Form 1 (see Appendix F), was administered. In Condition 2, the Matching Parts and Figures Test (MPF, see Appendix E), was administered following the Preliminary Questionnaire. The MPF was presented as being related to probable success in some areas of academic skill. Subjects were told that there is some reason to believe that men as a group are generally better than women in academic subjects which require an understanding of spatial relations, such as mechanical engineering. Subjects were informed that the purpose of the MPF was to assess their understanding of spatial relations. The Matching Parts and Figures Test consisted of 21 multiple-choice problems, the solution of which required understanding of spatial rela- tions. As one of the experimenters read the instructions aloud, subjects were given time to consider four example problems. The solution to one of the examples was given. After being told that they would have five minutes to solve the 21 problems once the test began, subjects were asked to fill out an Expectation Questionnaire identical to the questionnaire administered in Condition 1 prior to the IPS. When everyone had completed the Expectations Question- naire, subjects were instructed to begin the MPF. As designed, few subjects were able to complete the task in the time allotted. When time was up, the Posttest Question- naire, Form 2 (see Appendix F), was administered. 50 Procedures for Conditions 1 and 2 were identical, with the exception of order of administration of the IPS and MPF and accompanying Expectations and Posttest Questionnaires. After completing the Posttest Questionnaire following the first behavioral task, subjects in both conditions were prepared for the second behavioral task. In Condition 1, the MPF was the second task. In Condition 2, the IPS was the second task. In both conditions, after the Posttest Questionnaire following the second task had been completed, the Personal Beliefs Measure (see Appendix G) was administered. "Per- sonal Beliefs Measure" was the name under which the Multidimensional-Multiattributional Causality Scale (Lefcourt et al., 1979) was presented to subjects in this study. The MMCS is a 48-item questionnaire on which sub- jects are asked to give Likert scale ratings to a variety of attributions about success and failure in situations of academic achievement and in situations of affiliation. Finally, the Personal Characteristics Form, a l3-item questionnaire designed to gather background information (see Appendix H), was administered. Subjects were allowed as much time as they wished to complete the Personal Beliefs Measure and the Personal Characteristics Form. After all instruments had been completed, subjects were debriefed and the experimenters answered any questions. 51 Instruments The Preliminaryguestionnaire (see Appendix B) is a four-item questionnaire designed as a lead in to the present study, to establish a focus on the academic and interper- sonal domains. Subjects were asked to indicate the number of people they consider close friends and their preferences for one-to-one, small group or large group interactions. Subjects were also asked for their high school and college grade point averages. Data from the Preliminary Question- naire were not used in this phase of the research. The Expectations Questionnaire (see Appendix C) is a four-item questionnaire designed for this study. On a scale from.0'Uo 100% of items correct, with 10% intervals, subjects were asked to indicate their criteria for themselves for success, their expectancy for success, and their expecta- tions for the average performance of other college students. Subjects were also asked to indicate on a Likert scale from 1 to 7 the importance to them of success on the test. The Expectations Questionnaire was administered before subjects took the Interpersonal Sensitivity Test and before subjects took the Matching Parts and Figures Test. The measure of expectancy for success is the only item on the Expectations Questionnaire to be used in the analysis of data for the present study. The Posttest Questionnaire (see Appendix F) is a 10- item questionnaire designed for this study. Of the items on 52 the questionnaire, the item assessing attributions for outcome was of primary interest to this research. Subjects were asked to indicate what percent of their performance they thought was due to each of four causes: effort, ability, luck and task difficulty. Test instructions indicated that the sum of the percentages assigned to the four causes should total 100%. The four causes are the categories of attribution most frequently employed in Weiner's two-dimensional model of causality. Effort is considered an internal, unstable cause, ability an internal, stable cause, luck an external, unstable cause and task difficulty an external, stable cause of outcome on the specific task. The four categories of causality parallel the four categories of causality included in the MMCSIV, though the equivalent to task difficulty on the MMCSIV is the broader category of "context." Other items on the Posttest Questionnaire of particular interest to this study were three self-evaluations of per-~ formance. There was one seven-point scale on which to indicate self-evaluation of success relative to one's ex- pectations. On another seven-point scale, subjects were to indicate their perception of their success relative to the success of other college students. A third item assessed subjects' estimates of the percent of items they answered correctly, on a scale from 0 to 100%, with 10% intervals. 53 Other items on the Posttest Questionnaire assessed perceptions of the sex-linkage of the behavioral tasks. Estimates of the performance of female college students and male college students on the task were measured on a scale of 0 to 100% of items answered correctly. Subjects' perception of the task as masculine or feminine and per- ception of the relatedness of performance on the task to success in the designated area of achievement were also assessed, using 7-point scales. Two items on the Posttest Questionnaire paralleled items on the Expectations Question- naire. Subjects were asked to indicate the expectancies of success they had held before the task and the importance of success on 7-point scales. The data from these two items were not analyzed in this phase of the research. Two parallel forms of the Posttest Questionnaire were administered, one after the Interpersonal Sensitivity Test and one after the Matching Parts and Figures Test. The two forms were identical except for item 9. On the IPS Posttest Questionnaire, item 9 measured subjects' percep- tion of the relatedness of their performance on the IPS to probable success in interpersonal relationships. Item 9 on the MPF Posttest Questionnaire measured perceptions of the relatedness of performance on the MPF to probable success in academic subjects related to engineering. The Interpersonal Sensitivity Test (IPS; see Appendix D) was designed for this study as a task around which to 54 measure expectancy and attribution about performance in a traditionally feminine area of achievement. The IPS con- sists of 16 pictures of the face of one man and 16 pictures of the face of one woman. The pictures are photocopied photographs (Coleman, 1949). In each picture of the two sets, the model has a different facial expression. For each picture, the task was to indicate what emotion or emotions the model's facial expression was revealing. Sub- jects were asked to identify the dominant emotion expressed by the model and any secondary emotions expressed. There were lines for three responses for each picture on the answer sheet. A list of 60 emotions was provided on the answer form, though subjects were told that they were free to use any emotion they wished. One of the experimenters went through one sample item aloud before the test was administered, suggesting possible responses for the first picture. Five minues were allowed for completing the 31 test items. Subjects were informed that many of them would not be able to complete the test in the time alloted. There were no criteria for correctness of responses on the IPS. Scores were obtained for the number of pictures for which there was at least one response, and for the total number of responses made. For the purpose of this study, responses regarding the IPS made on the Expectations Questionnaire and the Posttest Questionnaire were of interest rather than performance on the IPS, per se. 55 The Matching Parts and Figures Test (MPF; see Appendix E) was employed in this study as a task around which to measure expectancy and attributions about performance in a traditionally masculine area of achievement. The MPF con- sists of 21 paper and pencil items taken from a test of understanding of spatial relations, such as might be employed in an aptitude test for engineering (Arco Edi- torial Board, 1947). For 13 of the items, there were two or three-dimensional drawings of some parts and three drawings of figures. The task was to decide which figure the parts could be assembled to make. For each of the remaining eight items, there was one two-dimensional figure and a choice of three sets of parts to choose from, one set of which could be assembled to make the figure. In addition to the 21 test items, there were four sample items which subjects were allowed to consider before beginning the test. The solution was given for only one sample item. Five minutes were allowed for the test. Subjects were informed that many of them would not be able to complete the test in the time alloted. Though scores of the number of items answered and the number of items answered correctly on the MPF were avail- able, responses to the Expectations Questionnaire and the Posttest Questionnaire regarding the MPF were of interest, rather than performance on the MPF, per se. 56 The Multidimensional-Multiattributional Causality Scale (Lefcourt et al., 1979; see Appendix G) was presented to subjects as the "Personal Beliefs Measure" in the present study. The MMCS consists of two 24-item Likert scales, one concerning academic achievement and one concerning affilia- tion. As the instrument was employed, the Likert scale for each item was from 1 to 5. Each scale is composed of 12 items pertaining to success and 12 items pertaining to failure. The items of both scales are divided into four categories of attribution, based on Weiner's (1971) model of causal attribution with internal-external and stable— unstable dimensions. Thus for each scale there are six stable external items related to contextual characteristics, six unstable external items relating to luck, six stable internal items relating to ability and six unstable in- ternal items relating to effort and motivation. A variety of measures can be derived from each scale, including total internality and externality scores, internality and externality for success and failure separately and scores for six or three item scales concerning specific attribu- tions. With a population of male and female college under- graduate students, Lefcourt et a1. (1979) found Cronbach's alpha to range from .58 to .80 for the academic scale and from .58 to .81 for the affiliation scale. Correlated Spearman-Brown split-half correlations range from .67 to 57 .76 for academics and from .61 to .68 for affiliation. Test-retest correlations for the affiliation locus of control scale range from .50 to .70 and for the academic scale from .51 to .62. The average correlation between the achievement and affiliation scales for the groups of students whom Lefcourt et a1. tested was significant but low, in the mid 20's. Adequate discriminant validity of the items for each scale was indicated by an average cor- relation of affiliation items with the affiliation scale of r=.31, but an average correlation of r=.09 with the academic scale. Similarly, academic items correlated .31 on the average with the achievement locus of control scale, while academic items correlated only .10 on the average with the affiliation scale. Lefcourt and his associates (1979) evaluated the influence of social desirability on MMCS scores using the Crowne-Marlowe (1964) need for approval measure. The need for approval was significantly correlated with the academic scale (r=.33, p < .01), but not with the affiliation scale. The negative relationship between academic achievement externality and need approval (r=-.30, p < .02) primarily accounted for the academic-need approval correlation. The MMCS affiliation scale has been found to correlate from .37 to .55 with the Rotter I-E scale (1966). The academic scale has been found to correlate from .23 to .60 with the Rotter measure (Lefcourt et al., 1978). 58 The Personal Characteristics Form (see Appendix H) is a 14-item questionnaire designed for this study. Some of the items on the questionnaire pertain to demographics such as the age, sex and educational level of the subject. Other items were designed to obtain information about the subject's parents. Respondents were asked to indicate the occupations and educational background of their parents. One item asked what level of academic achievement the subject perceived her/his parents to expect of her/him. The questionnaire also asked whether subjects thought that their parents' achievement expectations for them were related to their sex. Responses to the remaining items on the questionnaire provided data on the subject's college major and occupational and marriage plans. Data from the Personal Characteristics Form were not used in this phase of the research. RESULTS Experimental Design and Statistical Analysis Design The design of this experiment was a 2 x (Sex of Subject) x 2 (Area of Achievement, a repeated measure within each subject) x 2 (Condition, order of administra- tion of the IPS and MPF) factorial design. After the responses of all subjects had been discarded because of missing data, there were 34 males and 46 females in Condition 1 and 31 males and 35 females in Condition 2. Classification of Subjects According to Success or Failure on Tasks Subjects were classified for success and failure on the IPS and MPF on the basis of self-evaluations of per- formance on the Posttest Questionnaire. A composite success score for each task was computed by summing scores on three Posttest items: (a) self-ratings of success compared to expectations (success expectations score), (b) self-ratings of success compared to other college students (success comparison score), and (c) self-estimate of the percent of items answered correctly (percent correct score). The three items contributing to the success score were highly 59 60 correlated for each task. For the Interpersonal Sensitivity Test, the correlation of success expectations and success comparison was .62 (g'= 146; p < .001). For success compar- ison and percent correct, r = .69 (g = 146; p < .001). IPS success expectations and IPS percent correct scores corre- lated .51 (g = 146; p < .001). For the Matching Parts and Figures Test, the correlation between success expectations and success comparison was .62 (2.: 146; p.< .001). Success comparison and percent correct correlated .71 (2.: 146; p_< .001). The correlation between success expectations and percent correct was .63 (2.: 146; p < .001). Medians were computed for the IPS success score and the MPF success score. Subjects who scored above the median on the success score for each task were classified as success subjects for the area of achievement. Subjects whose scores fell below the median were classified as failure subjects. Subjects were divided into four groups on the basis of their IPS and MPF success scores. Group 1 included sub- jects who were classified as failures on both tasks. Sub- jects of Group 2 scored below the median on the IPS success score and above the median on the MPF success score. Group 3 comprised subjects classified for success on the IPS and failure on the MPF. Subjects with success scores above the median for both tasks made up Group 4. 61 Overview of Statistical Analysis Expectancy and attribution data were subjected to a 2 x 2 x 2 (Sex x Area of Achievement x Condition) multi- variate analysis of variance. Multivariate tests of the main effect of the repeated measure, area of achievement, were not available because of limitations of the computer statistical package. Univariate comparisons were obtained. When multivariate F ratios exceeded a confidence level of .05, associated univariate results were examined. Findings were reported for univariate results which were significant at the .05 level. Tests of simple effects were calculated where appropriate. There were two phases of data analysis, the hypothesis testing phase and the post hoc phase. The results of the hypothesis testing analyses were of primary interest. The MMCS data were analyzed independently of the IPS and MPF data. Dependent variables for the 2 x 2 x 2 hypothesis testing analysis of MMCS data were success ability, success effort and failure effort attribution scores. Dependent variables for the 2 x 2 x 2 hypothesis testing analysis of behavioral task data were expectancy, ability and effort scores. The data for all subjects were pooled for one analysis of IPS and MPF data. The data of each of the four success-failure subgroups were also subjected to an inde- pendent analysis. 62 Of secondary interest were the results of the post hoc 2 x 2 x 2 (Sex x Area of Achievement x Condition) multi- variate analyses of scores for attributions not defined as achievement promoting. The post hoc analyses were conducted to assess the effects of sex, area of achievement and con- dition on the use of all categories of attribution generated by the Weiner two-dimensional model of causality. Dependent variables for post hoc analysis of the MMCS data were the failure ability, success context, failure context, success luck and failure luck scale scores. Dependent variables for the post hoc analyses of the IPS and MPF data were task difficulty and luck attribution scores. As with the hy- pothesis testing analyses, a post hoc anlaysis of the behavioral task data was conducted on the pooled data for all subjects and on the data for each of the four subgroups. In order to obtain descriptive information about attribution patterns, the category of highest attribution score per subject was cross-tabulated with sex within each area of achievement and across sex with area of achievement. Chi square analyses were computed to determine whether the use of particular categories of attribution within each area of achievement were associated with sex. Hypothesis Testing Multivariate Analyses of Variance A summary of the significant multivariate results of hypothesis testing analyses will be presented, followed by 63 a discussion of the results as they apply to each of the four hypotheses. Unpredicted findings will then be dis- cussed. All multivariate F ratios for the hypothesis testing analyses of MMCS data and behavioral task data are presented in Tables 13 through 18 in Appendix I. No significant multivariate effects were obtained in the MANOVA on the MMCS data. Table 3 presents significant multivariate results for the hypothesis testing analyses of the task data. The analyses of the task data from Group 1 and Group 4 produced no significant multivariate results. Signifi- cant sex x area of achievement and area of achievement x condition interactions were obtained in the MANOVA on the IPS and MPF data for all subjects. The analysis of Group 2 data produced a significant condition effect. An area of achievement x condition interaction was obtained in the analysis of data from Group 3 subjects. As noted earlier, the 2 x 2 x 2 multivariate analysis did not assess main effects of area of achievement. Uni- variate 5 tests produced significant area of achievement effects for success ability score (F (1,142) = 4.93; p < .05) and failure effort score (F (1,142) = 9.24; p < .005) on the MMCS. Tests for simple effects of area of achieve- ment for achievement promoting attribution scores on the behavioral task yielded no significant results. Table 19 through 36 in Appendix J present all F ratios for tests of simple effects of area of achievement on MMCS and task data. 64 Table 3 Summary of Significant Multivariate Results for the 2 (Sex) x 2 (Area of Achievement) x 2 (Condition) Analysis of Expectancy, Ability and Effort Scores for the Behavioral Tasks Source of Variance df Multivariate F P. All Sub'ects (df for Error = 140) Sex x Area of Achievement Area of Achievement x Condition Group 2: IPS Failure- MPF Success or Error = 16) Condition Grou 3: IPS Success- MPF‘Failure d or Error = 31) Area of Achievement x Condition 4.46 4.29 3.33 4.11 .005** .006** .046* .015* * ** ‘p < .05; p < .01 65 Tables 4 and 5 present the significant univariate E ratios associated with multivariate effects for the hypothesis testing MANOVAS. Following is a discussion of the results as they apply to each hypothesis, including simple effects analyses and individual comparisons where appropriate. Table 4 Summary of Significant F ratios for the Expectancy, Ability and Effort Scores for the Behavioral Tasks for All Subjects Source F p 1. Sex x Area of Achievement (d: = 1/142) Expectancy 9.04 .003** 2. Area of Achievement x Condition (d: = 1/142) Expectancy 4.95 .028* Effort 6.05 .015* ** * p '<.05; p < .01 Table 5 Summary of Significant F Ratios for the Expectancy, Ability and Effort Scores for the Behavioral Tasks of Group 3, IPS Success-MPF Failure Subjects Source F_ p 1. Area of Achievement x Condition (df = 1/33) Expectancy 9.29 .005** Ability 4.77 .036* ** p < .05; p ‘<.Ol 66 Hypothesis One Hypothesis One stated that females would hold higher expectancies of success and more frequently employ achieve- ment promoting attributions in the traditionally feminine area of achievement, interpersonal relations, than in the traditionally masculine area of achievement, academics. This hypothesis was not supported. Tests for a main effect of area of achievement for the MMCS data within females produced a significant effect for failure effort attribu- '~tion (F )1.l42) = 7.49; p < .01), but in the opposite direc- tion predicted. Women were more likely to attribute aca- demic failure to lack of effort than interpersonal failure. Mean expectancy and attribution scores for women for each area of achievement are presented in Table 6. The predicted main effect of area of achievement for expectancy, ability attribution and effort attribution was found neither for the IPS and MPF data for all subjects, nor for any of the four subgroups. Tables 25 through 30 in Appendix J present the F ratios for the tests of simple effects of area of achievement for MMCS scores and behav- ioral task scores within women subjects. Area of achieve- ment did produce some significant results in interaction with other variables. The significant multivariate sex by area of achievement interaction for the pooled task data reflects a univariate effect on expectancy (F (1,142) = 9.04; p <:.004). Tests of simple effects indicated that 67 Table 6 Mean Expectancy and Attribution Scores for Inter- personal and Academic Outcome Within Sex Female Subjects Male Subjects Area of Achievement Area of Achievement Inter- Inter- Var1ab1e personal Academic personal Academic Success, MMCS Ability 8.05 9.53 7.66 9.45 Effort 8.56 9.47 7.61 9.28 Context 8.15 5.28 7.92 4.92 Luck 2.62 6.14 2.38 5.63 Failure, MMCS Ability 5.89 5.68 5.57 4.08 Effort 6.37 9.39 6.23 9.77 Context 5.74 5.97 5.35 6.32 Luck 4.00 4.99 4.23 4.72 Across Success and Failure, Behavioral Tasks Expectancy 70.74 71.23 62.00 72.15 Ability 26.91 36.62 28.54 34.54 Effort 31.73 33.31 29.91 32.69 Task Difficulty 31.36 23.72 28.17 26.40 Luck 10.30 6.21 12.81 7.28 F ratios for tests of area of achievement effects are presented in Tables 19, 20, 25, 26, 31 and 32 in Appendix J. 68 individual comparisons were appropriate. Contrary to the hypothesis, the comparison of mean expectancy scores for the IPS and for the MPF within females produced no signifi- cant differences. WOmen did not have higher expectancies of success for the interpersonal task. The area of achieve- ment by condition effect in the analysis of task data for all subjects reflects univariate effects for expectancy (F (1,142) = 4.95; p < .028) and effort (F (1,142) = 6.05; p < .016). In the absence of significant simple effects, individual comparisons of mean expectancy scores were com- puted. The comparisons yielded no significant results relevant to Hypothesis One. The analysis of task data for the subjects classified as success subjects on the interpersonal task and as failure subjects on the academic task, Group 3, produced a signifi- cant effect of area of achievement by condition on expec- tancy (E (1,33) = 9.29; p < .005) and ability (2 (1,33) = 4.77; p ‘<.037). Tests for main effects indicated that individual comparisons were appropriate. The comparisons yielded no significant differences for expectancy of ability mean Scores among females. The results indicate that the use of achievement- promoting attributions by females did not vary as a func- tion of area of achievement, except the use of lack of effort to account for failure on the MMCS. Contrary to predictions, effort was endorsed more as cause for academic 69 than interpersonal failure. Expectancy for success among women did not vary as a function of area of achievement. Hypothesis Two Hypothesis Two stated that males will hold higher ex- pectancies of success and more frequently employ achieve- ment promoting attributions in a traditionally masculine area of achievement, academics, than in a traditionally feminine area of achievement, interpersonal relations. Tests of simple effects and individual means were computed for the same MMCS and behavioral task data as with females. Results produced some support for Hypothesis Two. Mean expectancy and attribution scores for men for each area of achievement are presented in Table 6. Tests of a main effect of area of achievement on the MMCS data within males yielded a significant effect on failure effort attribution (F (1,142) = 6.71; p.< .01). As predicted, males were more likely to attribute failure in academics to lack of effort than failure in interpersonal relations. There were trends for an area of achievement effect for success ability (F (1,142) = 3.88; p_< .10) and success effort (F (1,142) = 3.23; p<= .10) in the predicted direction. When the pooled behavioral task data was analyzed, the results also pro-' vided support for the predictions of Hypothesis Two regard- ing effort attributions, though there were no significant effects for ability attribution. Males were significantly 70 more likely to attribute outcome on the academic task to effort than outcome on the interpersonal task (F (1,142) = 7.49; p < .01). The analyses of task data for each of the four success-failure groups produced no main effects for area of achievement. F ratios for area of achievement effects for the MMCS data and task data for males are summarized in Tables 31 through 36 in Appendix J. The results for the analysis of task data for all sub- jects also supported the predictions of Hypothesis Two re- garding expectancy. Expectancy showed a significant effect for sex x area of achievement (F (1,142) = 9.04; 2.: .004). Individual comparisons of mean expectancy scores indicated that males held higher expectancies for success on the academic task than on the interpersonal task (F (1,142) = 9.01; p < .005). For the pooled task data, a significant area of achievement x condition effect for expectancy and effort in the absence of simple effects indicated that individual comparisons were appropriate. Within male sub- jects, comparisons between the mean effort scores for the two areas of achievement within each condition produced no significant results. Parallel comparisons of expectancy mean scores yielded a significant difference as a function of area of achievement within Condition 2'(F (1,142) = 6.59; ' p < .025). Males who participated in the academic task; first held significantly higher expectancies for success on 71 the academic task than on the interpersonal task. There was no significant difference for males in Condition 1. None of the analyses of task data of the four success- failure groups produced support for Hypothesis Two. Though the analysis of Group 3 data yielded a significant effect of area of achievement x condition for expectancy and ability, individual comparisons produced no significant differences as a function of area of achievement in either condition. The results provide some support for the hypothesis that men are more likely to have achievement promoting cognitions in a traditionally masculine area of achievement than in a traditionally feminine area of achievement. The MMCS data indicated that men were more likely to attribute academic failure to lack of effort than interpersonal failure. Men were also more likely to attribute outcome across success and failure to effort on the MPF than on the IPS. There was a trend in the MMCS data for male sub- jects to endorse ability and effort more as causes for academic success than as causes for interpersonal success. The predicted higher expectancies for success for the academic task were found only when the academic task was done firSt. 72 Hypothesis Three Hypothesis Three stated that in the traditionally masculine area of achievement, academics, males would hold higher expectancies for success than females and employ achievement promoting attributions more frequently than females. No support was obtained for this hypothesis. Table 7 presents a comparison between men and women's mean attribution and expectancy scores for academic outcome. A significant sex x area of achievement interaction re- flecting a significant univariate effect on expectancy (F (l,l42)==9.04; p<< .004) was obtained for the analysis of the behavioral task data for all subjects. Individual comparisons between sexes of mean expectancy scores yielded no significant differences. No other significant inter- action of sex x area of achievement was obtained in any of the MANOVAS. Hypothesis Four Hypothesis Four stated that in the traditionally feminine area of achievement, interpersonal relations, females would hold as high or higher expectancies for success than males and employ achievement related attri- butions as frequently or more frequently than males. Results of the analyses support Hypothesis Four. Means are presented in Table 7. None of the analyses yielded a significant multivariate effect of sex or sex x area of 73 Table 7 Mean Expectancy and Attribution Scores for Men and Wbmen within Area of Achievement Area of Achievement Interpersonal Academic Variable Females Males Females Males Success, MMCS Ability 8.05 7.66 9.35 9.45 Effort 8.56 7.61 9.47 9.28 Context 8.15 7.92 5.28 4.92 Luck 2.62 2.38 6.14 5.63 Failure, MMCS Ability 5.89 9.45 5.68 4.08 Effort 6.37 9.28 9.39 9.77 Context 5.74 4.92 5.97 6.32 Luck 4.00 5.63 4.99 4.72 Across Success and Failure, Behavioral Tasks Expectancy 70.74 62.00 71.23 72.15 Ability 26.91 28.54 36.62 34.54 Effort 31.73 29.91 33.31 26.40 Task Difficulty 31.36 28.17 23.72 32.69 Luck 10.30 12.81 6.21 7.28 Results of multivariate analyses of variance are presented in Tables 14 and 15 in Appendix I and in Tables 37 and 38 in Appendix K. 74 achievement, except the analysis of pooled behavioral task data. This analysis yielded a significant multivariate effect of sex x area of achievement, reflecting a signifi- cant univariate effect for expectancy. Individual compari- sons of male and female expectancy scores for the pooled task data indicated, as predicted, that women held signifi- cantly higher expectancies of success than men on the inter- personal task (§_(l,284) = 8.38; pg< .005). Thus, in sup- port of Hypothesis Four, women held higher expectancies of success than men and employed achievement promoting attri- butions as frequently as men in the interpersonal domain. Though not associated with a significant multivariate effect, there was a tendency for women to be more likely to attribute success to effort across area of achievement for the MMCS data (F (1,142) = 4.00; p< .05). Unpredicted EffegES'Aoross SEX for the Hypothesis Testing Analyses For the MMCS data, an unpredicted main effect of area of achievement across sex for success ability attribution (F (1,142) = 4.93; pf: .05) and failure effort attribution (F (1,142) = 9.29; p<1 .005) was obtained. Scores on the academic success ability and failure effort subscales were significantly higher than scores on the parallel affilia- tion subscales. There was a trend for subjects to attribute success in academics more to effort than success in inter- personal relations (3 (1,142) = 3.63; E < .10). Thus, 75 contrary to predictions, subjects of both sexes were more likely to make achievement promoting attributions regarding academic outcome than regarding interpersonal outcome. There was no effect of area of achievement across sex for any of the five hypothesis testing analyses of the behavioral task data. Some of the analyses did produce unpredicted condition or area of achievement x condition effects. The analysis of task data for all subjects pro- duced an effect for area of achievement x condition on expectancy (F (1,142) = 4.95; p < .028) and effort (F (1,142) = 6.05; p < .016). Tests of simple effects for each of the variables indicated that individual comparisons were appropriate. Comparisons between mean expectancy scores for each area of achievement within condition yielded a significant difference within Condition 2 (F (1,142) = 7.08; p < .01). Subjects held higher expectan- cies of success on the academic task than on the inter- personal task when they engaged in the academic task first. The difference between the mean expectancy score for the academic task in Condition 1 and Condition 2 reached sig- nificance (F (1,284) = 7.64; p < .01). Individual compari- sons between mean effort scores in the two areas of achievement were also significant only within Condition 2 (F (1,142) = 4.02; p < .05). Regardless of sex, subjects were more likely to attribute academic task outcome to effort than interpersonal task outcome, if the academic 76 task was done first. Among subjects who engaged in the academic task secondly, there was no significant difference in expectancy or effort as a function of area of achieve- ment. When the task data was broken down into the four success-failure groups, a significant multivariate effect for area of achievement x condition was obtained only for Group 3 (F (3,31) = 4.11; p<= .015). Expectancy (§_(l,33) = 9.29; p < .005) and ability (F (1,33) = 4.77; p < .037) showed a significant effect for area of achievement x condition. While tests of simple effects indicated that individual comparisons were appropriate, the comparisons produced no significant area of achievement differences in either condition. However, subjects did hold signifi- cantly higher expectancies for the academic task when they engaged in the academic task first than when they engaged in the academic task after doing the IPS (F (1,66) = 6.60; p.< .025). This finding for the subjects rated as success- ful on the interpersonal task and failing on the academic task parallels the results for the analysis of the task data for all subjects pooled. While a significant multi- variate effect for condition was produced in the analysis of Group 2 data, it reflected no significant univariate comparisons. The hypothesis testing analyses produced an unpre- dicted area of achievement main effect for the MMCS data 77 and an unpredicted area of achievement x condition effect for the behavioral task data. Subjects of both sexes were more likely to endorse achievement promoting attributions for academic outcome than for interpersonal outcome for the MMCS. Area of achievement and order of administration of the behavioral tasks interacted to produce higher ex- pectancies for success and more attributions of outcome to effort for the MPF than for the IPS when subjects partici- pated in the MPF first. Post Hoc Multivariate Analyses of Variance The 2 x 2 x 2 (Sex x Area of Achievement x Condition) analysis of variance for the MMCS data with failure ability, success context, failure context, success luck and failure luck as dependent variables yielded no significant multi- variate results. Table 37 in Appendix K presents the multi- variate F ratios. Significant univariate comparisons for area of achievement across sex and condition were obtained for success context (5 (1,142) = 5.46; p < .025) and success luck (F (1,142) = 10.10; p