WWI WIN MW 293 1006 “l W M mam 2 8605 Michigan State University 11-4899 This is to certify that the _ thesis entitled THE RELATIONSHIPS BETWEEN COLLEGE WOMEN'S SEX ROLE IDENTITIES AND SELF—ESTEEM, AND THEIR PERCEPTIONS OF THEIR PARENTS' SEX ROLE IDENTITIES, SELF-ESTEEM, AND THE QUALITY OF THE PARENT-DAUGHTER RELATIONSHIP “preSented by ELIZABETH MONROE—COOK has been accepted towards fulfillment of the requirements for Ph. D. Counseling Psychology degree in Department of Counseling, Personnel Services and Educational Psychology Major professor Date May 4, 1979 OVERDUE FINES ARE 25¢ PER DAY PER ITEM Return to book drop to remove this checkout from your record. ‘7’.“‘ ‘ xxx '4 0" THE RELATIONSHIPS BETWEEN COLLEGE WOMEN'S SEX ROLE IDENTITIES AND SELF-ESTEEM AND THEIR PERCEPTIONS OF THEIR PARENTS' SEX ROLE IDENTITIES, SELF-ESTEEM AND THE QUALITY OF THE PARENT-DAUGHTER RELATIONSHIP By Elizabeth Monroe-Cook A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Counseling, Personnel Services and Educational Psychology 1979 ABSTRACT THE RELATIONSHIPS BETWEEN COLLEGE WOMEN'S SEx ROLE IDENTITIES AND SELF-ESTEEM AND THEIR PERCEPTIONS OF THEIR PARENTS' SEX ROLE IDENTITIES, SELF-ESTEEM AND THE QUALITY OF THE PARENT-DAUGHTER RELATIONSHIP By Elizabeth Monroe-Cook The problem addressed in the study concerned the definition of sex role identity for college women and some of the variables associated with their perceptions of sex role identity. A descriptive and analytic study was conducted on the relationships between college women's sex role identity and self-esteem, perceptions of each parent's self-esteem,cognitive involvement, warmth-versus- rejection, and each parent's sex role identity. The relationships between subjects' self-esteem and their perceptions of parents' self-esteem, cognitive involvement, warmth-versus;rejection, and each parent's sex role identity were also investigated. The college women in the study (N - 93) were divided into two samples of 96 and 97 respectively for the purpose of conducting a cross-validation of the instruments' reliability and the replication of the analysis of data. The women ranged in age from 17 to 26 years, and were divided evenly across the four academic years. The women who volunteered for the study responded to a questionnaire consisting of the Ben Sex Role Inventory (BSRI, Bem, 1974) for self, mother and father; the Rosenberg Self-Esteem Inventory (RSEI, Rosenberg, 1965) for self, mother and father; the Elizabeth,MOnroe~Cook Parent Behavior Form (PBF, Worell & Worell, 1974 manual) for mother and father; and the Family Information Questionnaire (FIQ) about their relationships with their parents and their feelings about them. Essential to the concept of sex role identity as investigated in the study is the idea that masculinity and femininity are independent domains within the individual, and are thus measured separately by the BSRI. 0n the basis of their scores on the separate masculine and feminine scales on the BSRI, subjects were placed into one Of the following four categories: Androgynous — Above the median on both the mascu— line and feminine scales. Feminine — Above the median on the feminine scale, but below the median on the masculine scale. Masculine — Above the median on the masculine scale, but below the median on the feminine scale. Undifferentiated Below the median on both the mascu- line and feminine scales. A large number of hypotheses were tested through the use of one-way analysis of variance, Pearson product-moment correlation, and Chi-square analysis. Prior to the analysis, extensive work was done on the reliability of the instruments. Because the cross- validation was conducted, the probability of a Type I error is .0025 (alpha for the independent samples was .05). The conclusions which have been emphasized are those based on significant results in both samples. Elizabeth Monroe—Cook Androgynous women in both sample groups did report higher levels of self-esteem than the other three sex role categories, and Undifferentiated women reported lower levels of self-esteem than the other three groups of women. Inconsistent results were obtained about the relationship between the daughters' sex role identities and their perceptions of their parents' self-esteem. However, in both groups there was a positive relationship between the self- esteem the women reported for themselves and the self-esteem they reported for each of their parents. There was also a positive relationship between the daughters' self-esteem and their mothers' cognitive involvement and between self-esteem and mothers' warmth— versus-rejection. Inconsistent results were obtained across both sample groups on fathers' cognitive involvement and warmth-versus- rejection. Androgynous daughters also reported higher levels of maternal warmth—versus-rejection than Feminine, Masculine, or Undif~ ferentiated women. In the last groups of hypotheses it was found that Androgynous women in both samples did report Androgynous mothers in higher proportions than the other groups, and that Undif- ferentiated women reported Undifferentiated mothers more frequently than the other three sex role categories. Inconsistent Chi-square results were obtained for fathers' sex role identities and subjects' sex role identities in the two sample groups for all four of the sex role categories. Feminine women and Masculine women did not report consistently about their mothers in the two sample groups either. The differences between the two sample groups offer a challenge for future research on the correlates of sex role identity in college women. To Josephine and Michael, my parents, and in grateful memory to Bill Kell ii ACKNOWLEDGMENTS To Bill Farquhar, the Chairman of my committee, who has encour- aged and supported my intellectual competence throughout my graduate training, and has profoundly influenced the process of this research, my warm thanks for his teaching and caring. To Sam Plyler, who has Opened so many new ways of thinking about human beings for me, and who has been an important resource in a great many phases of my development-teacher, colleague, and friend-- my enduring appreciation. To Elaine Donelson, who has brought a richness of thinking and being into my life, who has helped me to endorse the androgynous ideal, and whose ideas have shaped this study in an important way, many warm thanks. To Marty Anderson, who has given me considerable insight into the process of completing this project with integrity, and who has given patient, caring attention to this work and person, my caring and gratitude for her friendship. To Al Hammer and Doug Miller, who gave.invaluable help with the statistical analysis of the study, I acknowledge my thanks for understanding and sanity. To the women of the study, my thanks for their endurance of the many questions, and for their interest. To Dale Monroe-Cook, who is my most esteemed colleague in life, my love and gratitude for who he is, and his support. iii TABLE OF CONTENTS LIST OF TABLES . . . . . . . . . LIST OF FIGURES . . . . . . . . . Chapter I. THE PROBLEM . . . . . . . . Introduction . . . . . . . Changing Definitions of Sex Roles . Acquiring, Maintaining, and Changing Sex Roles. . . . . . . Purpose of the Study . . . . . Need for the Study. . . . . . Rationale for Studying College Women Sex Role Identity and Self-Esteem . Perceptions of Parents' Sex Role Identity and SEIf‘ESteem o o o o o Perceptions of Parents' Behavior Toward Daughters. . . . . . . Research Hypotheses . . . . . Sex Role Identity and Self-Esteem . Sex Role Identity and Perceptions of Parents' Self-Esteem . . . . Self-Esteem and Perceptions of Parents' Self-Esteem . . . . Self-Esteem and Perceptions of Parents' Behavior Toward Daughter . . Sex Role Identity and Perceptions of Parents' Behavior Toward Daughter Sex Role Identity and Perceptions of Parents' Sex Role Identities . . Theoretical Constructs and Empirical Bases for Study . . . . . . Roles . . . . . . . . Attitudes . . . . . . . Cognition and Development . . . Identification . . . . . . Dualism. . . . . . . Empirical Evidence . . . . . iv Page xiii 10 10 10 11 11 ll 12 Chapter Definition of Terms . . . . . Limitations of the Study . . . Overview of the Remaining Chapters II. REVIEW OF THE LITERATURE. Sex Role Identity. Masculinity and Femininity as Bipolar Concepts. . . Dualistic Concepts of Masculinity and Femininity . . . . . Masculinity and Femininity as Independent Concepts. . . Measures of Masculinity and Femininity Assuming Independence. Self-Esteem and Sex Role Identity. Parental Variables and Self-Esteem Traditional Analytic Views. More Recent Analytic Views. . . Cognitive-Developmental Theory. . Independent Measures of Masculinity and Femininity and Parental Variables . . . . Discussion of the Literature . Sex Role Identity . . . Self-Esteem, Sex Role Identity, and Parental Variables . . . . Assumptions for Hypotheses. . Summary . . . . . . . III. DESIGN OF THE STUDY . . . . . Sample . . . . . Measures . . . Procedures for Obtaining Subjects' Participation and Consent . . . Procedures for the Collection of Data. Design Features . . . . . . Hypotheses . . . . . . . Page 12 13 14 17 18 18 20 23 25 30 32 33 34 36 38 42 42 43 44 46 47 47 49 50 51 51 52 Chapter IV. Page Sex Role and Self— Esteem . . . . 54 Sex Role and Mother' 3 Self-Esteem . . . 54 Sex Role and Father' 8 Self-Esteem . . . 55 Subjects' Self-Esteem and Perceived Maternal Self— Esteem. . . . . 55 Subjects' Self-Esteem and Perceived Paternal Self-Esteem. . . . . 56 Self-Esteem and Mother's Cognitive Involvement . . . . . 56 Self-Esteem and Father' 8 Cognitive Involvement . . . . . 57 Self-Esteem and Mother's Warmth- versus-Rejection. . . . . . 57 Self-Esteem and Father's Warmth- versus-Rejection. . . . . . 57 Sex Role and Mother' 5 Cognitive Involvement . . . . . . 58 Sex Role and Father' 3 Cognitive Involvement . . . . . . . . 59 Sex Role and Mother's Warmth- versus-Rejection. . . . . . . 59 Sex Role and Father’s Warmth— versus-Rejection. . . . . . . 60 Sex Role for Self and Perceived Sex Role for Mother . . . . . . 61 Sex Role for Self and Perceived Sex Role for Father . . . . . . 62 Analysis of the Data . . . . . . . 63 Tests of Relationship . . . . . . 63 Analysis of Variance . . . . . . 64 Summary . . . . . . . . . . 65 INSTRUMENTATION . . . . . . . . . 67 Cross Validation . . . . . . . 67 Criteria for Eliminating Scale Items . . . 70 Presentation of Instruments to Subjects . . . . . 72 Bem Sex Role Inventory (BSRI). . . . . 73 Bem Sex Role Inventory Scale Reliabilities . . . . . 73 Medians of the Masculine and Feminine Scales . . . . . . . 76 Results of Eliminating Items . . . . 77 vi Chapter Parent Behavior Form (PBF) . Parent Behavior Form Scale Reliabilities . . . . . Results of Eliminating Items . . Rosenberg Self-Esteem Inventory (RSEI) Summary . . . . . . . . V. RESULTS OF THE ANALYSIS . . . . . Sex Role and Self-Esteem . . . . Sex Role and Parents' Self-Esteem. . Sex Role and Mothers' Self-Esteem . Sex Role and Fathers' Self-Esteem . Subjects' Self-Esteem and Other Parental Variables . . . . Subjects' Self-Esteem and Perceived Parental Self-Esteem. . . . Self-Esteem and Parents' Cognitive Involvement . . . . . . Self-Esteem and Parents' Warmth- versus-Rejection. . . . Sex Role and Parents' Cognitive Involvement. . . . . . . Sex Role and Motherts Cognitive Involvement . . . . Sex Role and Father 8 Cognitive Involvement . . . . . Sex Role and Parents' Warmth-versus- Rejection . . . . . . . Sex Role and Mother' 3 Warmth- versus-Rejection. . . . Sex Role and Father's Warmth- versus-Rej fiction. 0 o o 0 Subjects' Sex Roles and Perceived Parental Sex Roles . . . . . Subjects' Sex Roles and Perceptions of Mothers' Sex Roles . . . Subjects' Sex Roles and Perceptions of Fathers' Sex Roles . . . vii Page 80 82 84 85 86 90 9O 96 96 98 101 101 103 104 105 105 109 112 113 115 117 117 122 Chapter Page Family Information Questionnaire . . . 127 (FIQ) Closeness and Perceived Similarity to Parents. . . . . . . 127 Preferences for Help from Parents . . . . . . . . 133 Emotional Support from Parents . . . 134 Quantity of Time with Parents . . . 134 Stability of Relationship with Parents . . . . 135 Quality of Relationship with Parents . . . . . . . 135 Warmth toward Parents . . . . . 135 Respect for Parents . . . . . . 136 Trust toward Parents. . . . . 136 Intensity of the Relationship with Each Parent . . . . . . 136 VI. SUMMARY AND CONCLUSIONS . . . . . . 138 Summary of the Study . . . . . . 138 New Measures of Sex Role Identity . . 138 Correlates of Sex Role Identity . . . 139 Hypotheses . . . . . . . 140 Cross-Validation and Instrumentation . . . . . . 141 Analysis of the Data. . . . . . 144 Conclusions. . . . . . . . 147 Discussion of Results . . . . . . 147 Sex Role Identity and Self-Esteem . . 147 Relationship between Sex Role and Self— Esteem as Affected by Other Variables . . . . . . . 149 Subjects' Self— Esteem and Perceptions of Parents' Self-Esteem . . . . 150 Other Parental Variables. . . . . 151 Androgynous-Undifferentiated Differences . . . . . 152 Measuring Sex Role Identity . . . . 154 Sex Role and Perceptions of Parents' Sex Roles. . . . . . 155 Literature Introduced Since the Development of the Current Study . . . 155 Suggestions for Future Research. . . . 159 viii APPENDICES Page APPENDIX A. ACADEMIC MAJORS 0F SUBJECTS . . . . . . . . 163 B. NOTICE REQUESTING WOMEN'S PARTICIPATION IN THE STUDY. . . . . . . . . . . . . . . 165 C. CONSENT FORM . . . . . . . . . . . . . 167 D. BSRI AND PBF SCALE ITEMS AND ITEM-TOTAL INFORMATION. . . . . . . . . . . . . 168 E. BEM SEX ROLE INVENTORY AS PRESENTED T0 SUBJECTS FOR SELF, MOTHER, AND FATHER . . . . . . . 176 F. PARENT BEHAVIOR FORM AS PRESENTED TO SUBJECTS . . 179 G. ROSENBERG SELF—ESTEEM INVENTORY AS PRESENTED T0 SUBJECTS FOR SELF, MOTHER, AND FATHER . . . . 191 H. ANALYSIS OF VARIANCE SUMMARY AND GROUP MEANS . . 194 I. FAMILY INFORMATION QUESTIONNAIRE AS PRESENTED T0 SUBJECTS. . . . . . . . . . . . . 198 REFERENCE NOTE . . . . . . . . . . . . . . 203 SELECTED BIBLIOGRAPHY . . . . . . . . . . . . 204 Table 1.1 2.1 2.2 3.1 4.1 4.2 4.3 4.4 4.5 4.6 4.7 LIST OF TABLES Summary of Hypotheses about Subjects' Sex Role Identities, and Perceptions of Parents' Sex Role Identities . . . . . . . . . . Scoring Categories for Measures of Femininity and Masculinity as Independent Domains. . . . . . Results of Kelly's and Worell's 1976 Least Significant Difference Analysis, Sex Role Category by Parents, Women Subjects. . . . . . . . . . . Comparison of Some of the Characteristics of the Two Samples . . . . . . . . . . . Bem Sex Role Inventory Masculine Items, Sample Reliability Summary . . . . . . . . . Bem Sex Role Inventory Feminine Items, Sample Reliability Summary . . . . . . . . Bem Sex Role Inventory Scale Medians Based on College Women's Responses . . . . . . . Means, Standard Deviations, and Reliabilities for the Masculine and Feminine Scales of the Bem Sex Role Inventory. . . . . . . . . Summary of Parent Behavior Form Scales Which Form the Factors of Warmth-versus-Rejection and Parental Cognitive Involvement Used in the Analysis of Data. . Means, Standard Deviations, and Reliabilities for the Parent Behavior Form's Scales, mother and Father 0 O O O O O O O O O 0 Means, Standard Deviations, and Reliabilities of the Rosenberg Self-Esteem Inventory for Self, Mbther, and Father, Validation and Cross-Validation Group 8 O O O O O O I O O O O O O Page 27 40 49 74 75 78 79 82 83 86 Table Page 4.8 Summary of Instrumentation Findings, Bem Sex Role Inventory (BSRI), Parent Behavior Form (PBF), and Rosenberg Self-Esteem Inventory (RSEI) . . . 89 5.1 Summary of Results of Study of College Women's Sex Role Identity, Self-Esteem, and Perceptions of Parents . . . . . . . . 91 5.2 Analysis of Variance of the Rosenberg Self-Esteem Inventory Scores for Self for the Four Bem Sex Role Inventory Categories of College WOmen Subjects, Validation and Cross-Validation Groups . . . . 94 5.3 Analysis of Variance of the Rosenberg Self-Esteem Inventory Scores for Mother for the Four Bem Sex Role Inventory Categories of College WOmen Subjects, Validation and Cross-Validation Groups . . . . 97 5.4 Analysis of Variance of the Rosenberg Self-Esteem Inventory Scores for Father for the Four Bem Sex Role Inventory Categories of College Women, Validation and Cross-Validation Groups . . . . 99 5.5 Summary of Correlations Between Subjects' Self-Esteem and Parental Variables Other Than Sex Role, Validation and Cross-Validation Groups . . . . 102 5.6 Analysis of Variance of the Parent Behavior Form's Cognitive Involvement Factor Scores for Mother for the Four Bem Sex Role Inventory Categories, Validation and Cross-Validation Groups . . . . 106 5.7 Analysis of Variance of the Parent Behavior Form's Cognitive Involvement Factor Scores for Father for the Four Bem Sex Role Inventory Categories, Validation and Cross-Validation Groups . . . . 110 5.8 Analysis of Variance of the Parent Behavior Form's Warmtheversus-Rejection Factor Scores for Mother for the Four Bem Sex Role Inventory Categories, Validation and Cross-Validation Groups . . . . 113 5.9 Analysis of Variance of the Parent Behavior Form's Warmthrversus-Rejection Factor Scores for Father for the Four Bem Sex Role Inventory Categories, Validation and Cross-Validation Groups . . . . 116 xi Table Page 5.10 The 4 x 4 Contingency Tables (Frequencies) for Mothers' Sex Role Categories by College Women's Sex Role Categories, Validation and Cross- Validation Groups. . . . . . . . . . 118 5.11 Summary of 2 x 2 Contingency Tables (Frequencies) for Mothers' Sex Role Categories by College Women's Sex Role Categories on the Bem Sex Role Inventory, Validation and Cross-Validation Groups . . . . 120 5.12 The 4 x 4 Contingency Tables (Frequencies) for Fathers' Sex Role Categories by College Women's Sex Role Categories, Validation and Cross- Validation Groups. . . . . . . . . . 123 5.13 Summary of 2 x 2 Contingency Tables (Frequencies) for Fathers' Sex Role Categories by College WOmen's Sex Role Categories on the Bem Sex Role Inventory, Validation and Cross-Validation Groups . . . . 124 5.14 College Women's Reaponses to Family Information Questionnaire (N 8 193) . . . . . . . . 128 6.1 Summary Of Hypotheses and Results of Analysis of College Women's Sex Roles, Self-Esteem, and Perceptions of Parents . . . . . . . . 141 D1. BSRI Feminine Scale, Item-Total Information, Validation Group . . . . . . . . . . 168 D2. BSRI Masculine Scale, Item-Total Information, Validation Group . . . . . . . . . . 169 D3. PBF Scales, Item-Total Information, Validation Group I O O O O O O O O O O O O 170 H1. Analysis of Variance Summary . . . . . . . 194 H2. Group Mbans for Androgynous, Feminine, Masculine, and Undifferentiated College WOmen: RSEI and PBF. . 196 xii Figure 2.1 2.2 4.1 4.2 6.2 LIST OF FIGURES Masculinity and Femininity as Bipolar Concepts . Masculinity and Femininity as Independent concepts 0 O O O O O O O O 0 Decision Model for the Elimination of Individual Scale Items. . . . . . . . . . Decision Model for Choice of Scale to Use in Cross- Validation Group Analysis . . . . . . A Circumplex Model for Sex Role Identity . . . A Sex Role Identity Circumplex, Adjusted for Population Proportions . . . . . . xiii Page 19 24 69 71 161 162 CHAPTER I THE PROBLEM The contemporary women's movement has had profound effects on the theoretical and applied aspects of psychology. Among these effects has been the large amount of thought and research which has gone into the restrictions and strengths of the socially-defined feminine role. The benefits and limitations of the masculine role have also received considerable scrutiny. Furthermore, far fewer psychologists now view stereotypically-feminine women and stereotypically-masculine men as the best or only candidates for good mental health in American culture. Changing Definitions of Sex Roles One of the first major tasks confronted by psychologists attempt- ing to add understanding and flexibility to their attitudes about women's and men's behavior was the identification of the components of sex roles. The complexity of the task has been enormous for several reasons. First of all, psychologists and researchers are subjective participants in the processes which they are trying to objectify. Therefore, in the attempt to move away from the restrictions of the traditional views of men and women, and of masculinity and femininity, it is difficult to know what new restrictions have been included in psychologists' thinking. Second, there has been a great deal of confusion about the meaning of the phrase "sex role" and about which 2 behaviors or attributes have been described by this phrase. Also, the roles being studied have been changing as they are studied. That which was stereotypic 10 years ago may no longer operate in current society. Despite these difficulties, there has been increasing agreement that both constructive and destructive elements of sex roles exist, and agreement that some of the assumptions about sex roles merit change. Some of the changes will be reviewed extensively in Chapter II, Review of the Literature. Acquiring, Maintaining, and Changing Sex Roles In addition to the questions asked by psychologists about the components of male and female sex roles and the values placed on sex roles, there have been many questions asked about how these sex roles are acquired, maintained, and changed. A large amount of research has been done on the possibility of biological causes for men's and women's behavior and attitudes, but little agreement has been reached. There has been agreement that culture has a profound impact on individual role behaviOr. However, little consistent evidence has been produced about the specific social or psychological components of sex role acquisition, maintenance, or change. Although the fact that peo- ple's attitudes may not always coincide with their behavior has been noted frequently, it cannot be asSumed that it is not valid to study attitudes about sex roles because it is also widely assumed that atti- tudes and belief systems predispose people to certain behaviors. In addition, cognition is considered to be central to the process of devel- opment by some theorists (e.g., Kohlberg, 1966). Purpose of the Study The purpose of the present study was to examine the relation- ships which may exist between college women's sex role identities and their self-esteem, their perceptions of their parents' sex role iden- tities, and self-esteem, and their perceptions of their parents' behavior toward them. Need for the Study Information about the attitudes people have about their own and others' behaviors and characteristics is essential for the understanding of sex roles in American culture. Rationale for Studying College Women College women were chosen as the subjects for this study in part because they represent a group of women making the transition from lives under the direct influence of their parents to lives completely indepen- dent of parents. For college students, the transition has been seen as more gradual than it is for their non-college peers, which may make the process of the change easier to observe. The transitional phase has been of great interest because of the information it may afford about those elements of parents' behavior which take on particular importance for the daughter as she achieves her separateness from her parents. Late adolescence has been a period worth studying for many other reasons. Supposedly American culture extends adolescence, and there is not as much information as is needed about the reality of life in the extension, or its implications. Nor is there much information about extended adolescence and sex roles as they will be defined in the cur- rent study. Perhaps most importantly, women were chosen as the subjects of the present study because of the need for information about the psy- chology of women. While direct implications for change may not have been included in this study, it was assumed that information and insight about the relationship between certain facets of a woman's experiences and her image of herself may lead to change. A belief in the impact of insight on personal change is central to much of psychology, and is reflected in the study presented here as well. The hope behind the study was that information about women's reactions to socialization, by parents in particular, would allow women to examine the effects of socialization, extend those which are constructive, and modify those which are not for themselves. The information contained in this study may also be of interest to those who are involved in the socialization of young women. Some of these people, concerned about their impact on girls and women, may find ideas which would be useful in their attempts to equip them with social and personal skills needed for a world which presents many complexities, not the least of which is the role conflicts it holds for women. Lastly, it was hoped that psychologists would see this infor- mation as useful in their research on women, and in their attempts to help women in the clinical setting. Sex Role Identity and Self-Esteem There has been increasing interest in the relationship between sex role identity and self-esteem (e.g., Spence, Helmreich, & Stapp, 1975; Bem, 1977). The current study also included the variable of self- esteem, with the idea of investigating the relationship between self- esteem and parental variables, as well as the relationship between self-esteem and sex role identity. Perceptions of Parents' Sex Role Identities and Self-Esteem At the time of the development of the current study, no study using the newer measures of masculinity and femininity (to be described later) had also asked subjects to report on their parents' sex role identities and self-esteem. Nonetheless, these reports about parents were thought to be possible indicators of important factors in the daughters' sex role identities and self-esteem. Given the assumed process of learning from and imitating one's parents at different stages of life, it seemed important to know what the daughter believes she has learned and has seen in her parents' behavior. The daughters' perceptions of parents' self-esteem was considered as an element of the cognitive and emotional processes which may affect their own self-esteem. Perceptions of Parents' Behavior Toward Daughters It is quite possible that the daughter's perceptions of how she has been treated by her parents can be as influential as how she was actually treated on the daughter's behavior. In fact, perceptual or 6 phenomenological psychologists such as Snygg and Combs argue that the individual's behavior depends on his or her perceptions of situations and of themselves, not on the so-called "objective realities." Psycho- analytic theory argues the importance of introjects on individual behavior, and cognitive theorists emphasize perceptual organization. Combs and Snygg (1959) have said, People do not behave according to the facts as others see them. They behave according to the facts as they see them. What governs behavior from the point of view of the individual him- self are his unique perceptions of himself and the world in which he lives, the meanings things have for him. (p. 17) Therefore, because of the desire to know something about the relation- ships between the parents' behavior and the daughters' behavior, the daughters were asked to describe their parents' behavior. The women in the present study have also described two particular aspects of their self images: their self-esteem or self-worth, and their sex role identity. Research Hypotheses General research hypotheses were formulated for the current study, which resulted in the production of several specific hypotheses. The general hypotheses will be presented in the following section, and the specific, testable hypotheses will be found in Chapter III, Design of the Study. Sex Role Identity and Self-Esteem Young women who report relatively high degrees of both masculinity and femininity were hypothesized to report the highest levels of self-7 esteem. Those women reporting relatively low degrees of both 7 masculinity and femininity were hypothesized to report the lowest levels of self—esteem. These hypotheses are in keeping with the findings of earlier studies (e.g., Spence, Helmreich & Stapp, 1975; Bem, 1977). Sex Role Identity and Perceptions of Parents' Self-Esteem A projection was made that women rating themselves as high in masculinity and high in femininity will also report high levels of self-esteem for each of their parents. Included in this hypothesis is the assumption that if subjects see their parents having positive self images, it is related in some manner to the daughter's having an expan- ded self image. Women reporting the lowest relative levels of mascu- linity and femininity were predicted to report the lowest levels of parental self-esteem. Self-Esteem and Percgptions of Parents' Self-Esteem Positive relationships between the daughter's self-esteem and the levels of self-esteem she ascribes to each of her parents were expected. If it is assumed that the daughters' reports reflect the reality of the parents' self-esteem, many different interpretations could be used with these hypotheses. For example, parents with high self-esteem may react more constructively to their children than do parents with low self-esteem, resulting in a daughter's having a high level of self-esteem. Another possibility is that daughters imitate the self-esteem statements of their parents. However, the position taken in the present study was the former, not the latter. Self-Esteem and Perceptions of Parents' Behavior Toward Daughter In these hypotheses two aspects of parental behavior were empha- sized: warmth and cognitive involvement. The prediction made in this study was that there would be direct, positive relationships between self—esteem and the degree of warmth from each parent. Again, these hypotheses relied on the rather basic notion that if the daughter reported positive, constructive behavior from her parents, she would also report high self-esteem. Sex Role Identity and Perceptions of Parents' Behavior Toward Daughter As in the previous section, these hypotheses emphasized cognitive involvement and warmth from each parent, as perceived by the daughter. It was projected that both warmth and cognitive involvement would be reported to be high from the parents of those daughters who reported high levels of both masculinity and femininity. Similarly, it was hypothesized that daughters reporting low levels of both masculinity and femininity would also report low levels of warmth and cognitive involvement from each of their parents. In past research (Kelly & Worell, 1977) there has been evidence to indicate that cognitive involve- ment from mother and father are influential in the daughters' reports of sex role identity. Sex Role Identity and Perceptions of Parents' Sex Role Identities In this section of hypotheses, a differentiation between mother and father was proposed. First a general hypothesis was formulated that 9 there would be similarity across all the groups of daughters and mothers. For example, if a daughter reported high levels of both masculinity and femininity, it was hypothesized that she would report her mother to have similarly high levels of both masculinity and femininity. If a daughter reported high levels of femininity but not masculinity it was suggested that she would report the same for her mother. With reports of their fathers' identities, however, it was suggested that a counterbalancing would occur, e.g., if a daughter reported high levels of femininity but not masculinity, that she would report her father as having a high level of masculinity, but not femininity. However, high levels of both masculinity and femininity in the daughter or low levels of both were proposed to be related to her perceiving the same in her father. These hypotheses have been summarized in Table 1.1. TABLE 1.1 SUMMARY OF HYPOTHESES ABOUT SUBJECTS' SEX ROLE IDENTITIES, AND PERCEPTIONS 0F PARENTS' SEX ROLE IDENTITIES If Then Daughter's Reports Daughter's Reports Daughter's Reports for Self for Mother for Father High Femininity High Femininity High Femininity High.Masculinity High Masculinity High Masculinity High Femininity High Femininity Low Femininity Low Masculinity Low Masculinity High Masculinity Low Femininity Low Femininity High Femininity High Masculinity High Masculinity Low'Masculinity Low Femininity Low Femininity Low Femininity Low Masculinity Low'Masculinity Low Masculinity 10 Reliance on some traditional views about the development of masculinity and femininity has influenced some of the hypotheses, e.g., the expec- tation that the daughter would report similarity to the same-sex parent. Theoretical Constructs and Empirical Bases for Study NO one theory was drawn upon for the design of the study pre- sented here, nor was it the purpose of this study to investigate a facet of a particular theory. However, there are several constructs from a few theoretical areas which are relevant to the diScussion of the acquisition of sex roles, and the maintenance of sex role identity. In addition, many pieces of empirical evidence have been used in the design of the study. The theoretical and empirical elements will be discussed briefly in this section and explored more fully in Chapter II, Review of the Literature. Out of the field of social psychology, role theory has had a significant influence on the area of sex role studies, and on the ideas presented in the current study. For example. many of the sources which are cited here have used the ideas of Parsons and Bales (1955), who have defined some major aspects of masculinity and femininity. Attitudes Attitudes are considered to be predispositions to behavior by so- cial psychologists. Researchers rely on subjects' reports to understand attitudes, even though recognition is given to the fact that attitudes may not always be related to behavior. The current study alSo relied on subjects' reports about themselves and others as valid measures of attitudes. 11 Cognition and Development Basic to the cognitive-developmental view is the idea that indi- vidual thought is an active determinant of role perceptions. Cognitive- developmental theorists (e.g., Kohlberg, 1966) also posit a develop- mental process, saying that individual role concepts do change with age, depending on both cognitive and social development. Kohlberg (1966), drawing on the work of Piaget, states: "Sex role concepts and attitudes change with age in universal ways because of universal age changes in basic modes of cognitive organization" (p. 83). The present study in- cludes acceptance of the importance of cognition in sex role identity, and the importance of the "experience-linked changes" cited by Kohlberg I (1966), especially in women's experiences with their parents. Identification While the psychobiological emphasis of much of analytic theory did not particularly influence the present study, the ideas presented by analysts about identification will be discussed. Identification is a concept frequently cited in discussions of sex role identity. Object relations theory, for example, includes many ideas about intrapsychic events and intrapsychic development in the context of the environment, particularly early relationships. The importance of the internal experience of external events in psychodynamic theory was the main reason for its inclusion in the thinking behind the present study. Dualism One of the major elements of Jung's theory is his dualistic con- cept of many facets of human nature. Jung's idea that masculinity and 12 femininity both occur in the personality regardless of gender has been drawn on for the current study, and by others who will be cited in this study. Empirical Evidence The majority of the studies cited are recent; they were completed within the last five or 10 years. The reason for restricting the search for evidence to the last 10 years is the fact that the measures of femininity and masculinity which assume them to be independent prin- ciples are themselves so recent. In the current study a construct of sex role identity is being investigated which includes elements of femininity, masculinity, and androgyny. Among the studies cited (and described in Chapter II) are those which operationalize the masculinity and femininity to be viewed as separate principles. The studies include those investigating the development of the new measures of masculinity and femininity, changes in sex role identities, and variables which seem to be related to sex role identity. Definition of Terms For the present study egg rglee_were defined as the expectations American society holds for women's and men's behavior (Bem, 1974). The discussion of sex roles in this study did not include the idea that they are those behaviors which differentiated women from men. Rather than differentiating women from men, sex roles defined which characteristics the culture expects from women, and from men. A specific meaning for sex role identity was defined for use in the current study: sex role identity is the degree to which individuals 13 regard themselves as masculine and feminine, according to the societal norms for these roles (Kagan, 1964). Sex role identity is distinct from sexual identity or sexual preference, terms which refer to the individual's view of self as heterosexual, homosexual, or bisexual. Sex role identity is also different from sex role preference (Lynn, 1959; Biller, 1971), which is how people think they would like to be with regard to cultural ideals for masculinity and femininity. Sex role adoption is distinguished from identity because adoption repre- sents the degree to which others would define an individual as mascu- line and feminine according to societal values (Donelson, 1977). In the research hypotheses about parents' behavior toward their daughters the two elements which were emphasized were wermgh and cognitive involvement (Worell & Worell, Note 1; Kelly & Worell, 1976). Warmth refers to interpersonal caring, interest, and non-rejecting behavior. Cognitive involvement refers to the parents' interest in and encouragement of the child's intellectual competence and curiosity. In the current study self-esteem was defined as general feelings of worth and value, a favorable opinion of self (Rosenberg, 1965). This definition was used for both the daughters' perceptions of themselves and of their parents. Limitations of the Study The study presented here did not include biological aspects of sex roles or sex role identity. As was noted, there has not been much agreement about the effects of biological factors, and the current study will add nothing to the discussion of these factors. Only those 14 elements which are obviously a part of the socialization process, or of the individual's reactions to the socialization process, were included whether or not biology could also be assumed to be a factor. Because the subjects of this study were all women, male-female differences will not be addressed, nor will they be assumed to exist along all the dimensions of the study without further information. On the other side, no assumption was made that the results of this study could automatically be applied to male college students. The generalizations made in the study have been confined to college women. No direct observations of behavior were made in the study; all of the data analyzed were perceptions reported by the subjects. This limitation is particularly important in the discussion of parents' self-esteem and behavior toward their daughters because it must be noted that the daughters' descriptions may or may not reflect actual behaviors accurately. Women raised primarily in single-parent families have not been included in the current study. Differences between single-parent and two-parent families are assumed to exist, but are not addressed. Therefore, the generalizations of the present study must be confined to women raised in two-parent families. Overview of Remaining Chapters In Chapter II, relevant literature will be offered, including information from some of the theoretical and hypothetical areas deemed to be most salient for the current study. In the main, however, empirical information from studies directly related to the one 15 presented here will be presented. The information will be organized chiefly along the thematic lines of the general hypotheses. Chapter III will be dedicated to the details of the design and implementation of this study. The sample used for the study will be described, a brief introduction to the study's instruments will be given, and the testable hypotheses will be presented. The plan for analysis of data will be shown, including some of the features of the statistics used. The rationale for conducting a cross—validation study will be introduced. In Chapter IV a detailed explanation of the instruments used for this study will be found. The reliabilities of the instruments for these samples will be included, with information about whether or not removing some items from the instruments improved their reliabilities. A separate chapter on instrumentation has been included because of the large amount of information to be presented, and the fact that instru- mentation is not simply a matter of design nor analysis. The results of the analysis of data will be found in Chapter V, including the results of a survey given to subjects which was not part of the formal hypotheses, but which offers some interesting information about the subjects' feelings about their relationships with their parents. Because this was a cross-validation study, the results of the statistical analysis will be presented in the following manner: Hypothesis tested, results from the first sample (validation group), results from the second sample (cross—validation group). This par- ticular format was intended to make the comparisons between the l6 validation and cross-validation groups easier for the reader to follow. The results of the survey will be reported in both narrative and tabular form. The conclusions drawn on the basis of the instrumentation, analysis, and survey results, and a discussion of alternative expla- nations will be found in Chapter VI. Research completed since the design of the current study will be drawn into the discussion, and suggestions for future research will be explored. CHAPTER II REVIEW OF THE LITERATURE Several distinct areas of study are relevant to the development and implementation of the current study, and will be summarized in the chapter which follows. The first of these areas is the definition of masculinity and femininity. There are a multitude of definitions of masculinity and femininity, and little agreement about which elements of the human personality are described by these terms. The concepts may be viewed philosophically, sociologically, anthropologically, and psychologically. In the present study both psychological and socio— logical viewpoints were used. In the following chapter, two major approaches to the concepts of masculinity and femininity--the bipolar approach and the independent approach--will be discussed. These con- cepts are central to the investigation of sex role identity, and, therefore, to the current study. Another area of study which will be reviewed is self-esteem, par- ticularly as it pertains to sex role identity. Self-esteem is, of course, the subject of much psychological research, and to review even a portion of the general research would be an enormous task. Therefore, the current review will be restricted to the studies which investigated self-esteem as a personality correlate of sex role identity based on the construct of masculinity and femininity as independent principles. 17 18 Some of the theoretical and empirical vantages on the acqui- sition of sex role identity will also be reviewed. The particular focus will be parental variables in the socialization process. Analytic and cognitive-developmental theory (i.e., Kohlberg) will be discussed, and a study (Kelly & Worell, 1976) which was particularly germane to the development of the current study will be presented. Finally, a discussion of the implications of some of the literature for the hypotheses of this study will be found. Sex Role Identity As noted in Chapter I, sex role identity has been defined as the degree to which the individual sees himself or herself as masculine and feminine (Kagan, 1964). While many writers use this particular definition, there are a number of different views about the source(s) of sex role identity. There are also different views about what "masculine" and "feminine" are. Masculinity and Femininity as Bipolar Concepts Traditionally, masculinity and femininity have been viewed as bipolar concepts when studied empirically. Thus, masculinity repre- sented one end of the continuum and femininity the other. Individuals were labeled as masculine or feminine, but not both. This approach is also called the unidimensional view of masculinity and femininity. Test scales such as the Strong Vocational Interest Blank's MF Scale (1936), the Terman and Miles Attitude-Interest Analysis MEF Test (1936), and the Minnesota Multiphasic Personality Inventory's Mf scale (1943) 19 were designed to place an individual on the continuum between masculinity and femininity. Those individuals who scored in the middle, or in cross-sex fashion, were often considered deviant. Sex role identity has sometimes been confounded with sexual preference or orientation as well. As Spence and Helmreich (1978) say, "The governing criterion for inclusion of items has been their capacity to distinguish between men and women and also, in some instances, between homosexuals and hetero- sexuals" (p. 18). MOst of the tests used in the past assumed that a positive score on a masculine trait automatically earned a negative score on a feminine trait. This ipsitive view dominated psychological studies for years, and resulted in a restricted view of the nature of humans. Constantinople (1973) pointed out that there were many problems with these measures, not the least of which is their reverse logic, e.g., if A is masculinity, not-A is femininity, and vice versa. Another problem is the assumption that a unitary trait with two contrasting ends is being measured (see Figure 2.1). 0 1 L X f I I Masculinity Femininity Fig. 2.1. Masculinity and femininity as bipolar constructs Parsons and Bales (1955) proposed a theoretical model in which masculinity was described as "instrumental" behavior, that is, behavior designed to accomplish goals through initiative and independence. The feminine orientation was described as "expressive," designed to 20 please Others and facilitate harmony and understanding. Johnson (1963) used Parsons's and Bales's definitions and argued that the instrumental position precludes the expressive because to be instrumental the indi- vidual ". . . must resist pressures to become affectively involved in the immediate interactional situation" (p. 321). Johnson also argued that the instrumental orientation can be seen as the "disciplined pursuit of goals that transcend the situation" (p. 321). Thus, in Johnson's formulation, the assumption was found that a person must be primarily instrumental or primarily expressive. Not included was the idea that while people may have to choose one mode over the other for a given situation, their choice of modes may change with changing situations. Over time, then, people would have ample opportunity to respond out of both modes, depending on the assessments they make of individual situations. Dualistic Concepts of Masculinity and Femininity The idea that over time the two dimensions called masculinity and femininity can be expressed by the same individual was presented by Bakan (1966), who used definitions similar to those of Parsons and Bales: I have adopted the terms 'agency' and 'communion' to charac- terize two fundamental modalities in the existence of living forms, agency for the existence of an organism as an indivi- dual, and communion for the participation of the individual in some larger organism of which the individual is a part. (p. 15) In a chapter on human sexuality, Bakan assigned these terms to masculinity and femininity, saying, "I propose . . . that what we have been referring to as agency is more characteristically masculine and what we have been 21 referring to as communion is more characteristically feminine" (p. 110). Within the individual, agency and communion can be co-existing domains; thus, masculinity and femininity can co-exist. However, one may be more dominant than the other in any given individual. Bakan continues, For in the male and in the female we have instances of differentiation of function, especially with respect to their roles in reproduction. If we think of agency and communion as two major functions associated with all living substance, then, although agency is greater in the male and communion greater in the female, agency and communion nonetheless charac— terize both. (p. 152) Bakan's position echoes that of Jung, whose description of 22123 as femininity and animus as masculinity is one of the better- known dualistic concepts. Jung also uses Eros to label the feminine principle and Logos for the masculine principle. Jung says that each individual has both dimensions, and that for women the anima is more dominant (conscious), with the animus as unconscious, submerged closer to the core of the psyche. For men Jung postulates the opposite: the animus is expressed outwardly, and the anima is unconscious, or latent. Jung emphasizes the importance for the individual of balancing the two dimensions. Jung has been criticized by some for the fact that he does not seem to place equal emphasis on development for men and for women through the use of the latent dimensions. For example, Singer (1976), quoting the following from Jung, "'Just as a man brings forth his work as a complete creation out of his feminine nature, so the inner masculine side of a woman brings forth creative seeds which have the power to fer- tilize the feminine side of the man'" (p. 47), goes on to say herself, Although Jung asserts that both men and women draw their creative energies from the unconscious, there is a subtle--or perhaps not so subtle--difference. The man's anima helps him to produce his 22 creative work. The woman's animus is supposed to inseminate the man's anima, which thereupon inspires him to produce his creative work. Fortunate, but rare indeed, is the woman whose active animus is furthered by the tender nurturing of a man's anima. (p. 47) Caught up in the patriarchal emphasis of psychoanalysis or not, Jung (1951) does say, The recognition of the anima gives rise, in a man, to a triad, one third of which is transcendent: the masculine subject, the opposing feminine subject, and the transcendent anima. With a woman the situation is reversed. (p. 161) In Jung's psychology, transcendence is central to growth and fulfillment. One of his biographers (Campbell, 1971) points out that Jung thought that the psychological goal of life should be to know all parts of one- self, not to suppress or repress any side. Speaking of Jung, Campbell (1971) says, "And he terms that faculty of the psyche . . . gaining release from the claims of but one or the other of any pair-of—opposites, the Transcendent Function" (p. xxviii). The emphasis placed by Jung on the existence of both masculinity and femininity within each individual and the need for integration of the two is the feature of his work which was used in this study, and not the emphasis on the belief in the inherent, instinctive nature of the two dimensions. As Donelson (1977) states, Jung does seem to assume the association of femininity with femaleness and masculinity with maleness (biopsychological equivalence). However, the kinds of personality attributes he assumes to be feminine or masculine do match well with contemporary role stereotypes. Perhaps more important, his views about the necessity of development of both kinds of characteristics are supported by research and modern conceptions of androgyny.l . . . (p. 27) The "modern conceptions of androgyny" will be discussed shortly. 1From the Greek, andro = male, gyn = female. 23 The dualistic notion of masculinity and femininity has had credence, and has existed for a much longer time than what the empirical use of bipolar measures of masculinity and femininity might suggest. Examples from mythology, religion, classical literature, and philosophy, of the co-existence of masculinity and femininity within the individual, the integrative principle of androgyny, are easy to observe (Heilbrun, 1964; Singer, 1976), and richly rewarding to investigate. For both women and men much of the potential wasted by our culture's devaluation of that which is feminine can be fulfilled through rediscovery of the positive aspects of femininity. Because the over-valuation of that which is masculine has had destructive results, the rediscovery of the positive aspects of masculinity-~in a more balanced perspective--will be necessary as well. Masculinity and Femininity as Independent Concepts Since Constantinople's 1973 critique, research has been conducted on the variety of characteristics or behaviors which can be labeled feminine or masculine without the assumption of bipolarity (Bem, 1974; Spence, Helmreich & Stapp, 1975; Heilbrun, 1976; and Berzins, Welling & Wetter, 1978). The basic themes of instrumental, agentic characteris- tics and behavior for masculinity and expressive, communal characteris- tics and behavior for femininity are found in all of these researchers' work. These studies support Constantinople's idea that masculinity- femininity is not a unitary trait. Rather, they are modes of being and behaving Which are separate from one another, yet related. If allowed to score themselves separately for each domain, subjects will report 24 varying degrees of stereotypic feminine characteristics and varying degrees of stereotypic masculine characteristics. The developers of these new tests argue for the measurement of femininity and masculinity as independent domains (see Figure 2.2). V Femininity \/ Masculinity Fig. 2.2. Masculinity and femininity as independent constructs Studying the concepts of femininity and masculinity as indepen- dent dimensions has encouraged a broader view of emotional health and offered the challenge to re-evaluate earlier data. For example, those subjects scoring comparatively high on both masculine and feminine scales have been labeled "Androgynous" and have been reported to exhi- bit a higher level of situationally-appropriate behaviors than role- restricted behaviors has come from Bem (1974; 1975; with Lenney, 1976). Apparently it is not necessary nor sound to assume that psychological health is limited to adherence to cultural norms, even though there is social pressure to adhere to certain gender-related roles. In fact, it is possible that a high degree of conformity to societal sex roles may hinder women's psychological well-being. For example, Heilbrun (1965) reported that he found a high level of identification with (perceived similarity to) high-feminine mothers among college women he had categorized as "maladjusted" because of 25 their seeking psychological help. The traditional feminine role imposes a lot of restrictions, and is often defined by characteristics which are not as valued in American culture as those ascribed to the masculine role (McKee & Sherriffs, 1957; Rosenkrantz, et a1., 1968; Broverman, et a1., 1972; Spence & Helmreich, 1972; Deaux, 1976). Her review of the literature has led Donelson (1977) to conclude: Boys in our culture initially have more pressure for role behavior than do girls, but more freedom than girls within the role toward which they are pressured. Girls are less rigidly sex-typed than boys, but females suffer more for adhering to sex roles than do males. (p. 121) The penalties associated with roles may be different, but are apparent for both sexes. Measures of Masculinity and Femininity Assuming Independence Four instruments have been designed recently to measure femi- ninity and masculinity as orthogonal dimensions. Familiarity with these measures is important for understanding the hypotheses and procedures of this study. The first of these new measures is the Bem Sex Role Inventory (BSRI) developed by Sandra L. Bem (1974). The BSRI is a 60-item instrument; 20 adjectives were selected as positively-valued masculine characteristics, 20 adjectives were selected as positively- valued feminine characteristics, and 20 adjectives are not gender- related items but were used as a social desirability scale in the development of the BSRI. Bem began with a pool of 400 adjectives-— 200 were either masculine or feminine in tone, and 200 were neutral-- and had two samples of judges (100 in all) "rate the desirability in 26 American society of each of the approximately 400 items" (p. 157). Half the judges were male and half were female. An individual judge rated all 400 items either on their desirability for men or for women. Bem explained the criteria for inclusion of an item in the following manner: A personality characteristic qualified as masculine if it was independently judged by both males and females in both samples to be significantly more desirable for a man than for a woman (p <.05). Similarly, a personality characteristic qualified as feminine if it was independently judged by both males and females in both samples to be significantly more desirable for a woman than a man. . . .' (p (.05) A personality characteristic qualified as neutral with respect to sex . . . a) if it was independently judged by both males and females to be no more desirable for one sex than for the other (t‘.2) and b) if male and female judges did not differ significantly in their overall desirability judgements of that trait (t.2). (p. 157) When Bem first used the inventory, she employed three categories in her scoring procedures: "Masculine," subjects scoring significantly higher on the masculine scale than on the feminine scale; "Feminine," subjects scoring significantly higher on the feminine scale than on the masculine scale; and "Androgynous," subjects scoring about equally on both the masculine and feminine scales. Criticism of the method by which an individual was labeled androgynous mounted (Spence, Helmreich A Stapp, 1975; Strahan, 1975; Baucom, 1976; and Berzins, Welling & Wetter, 1978), and Bem revised the scoring process to include a fourth category, "Undifferentiated," which represents those people who score about equally on the masculine and feminine scales, but also score low on both. In Bem's current work, "Androgynous" represents subjects whose scores on each of the scales are about equally high (see Table 2.1). 27 Bem, conceding to the criticism offered, added the idea that the degree to which both feminine and masculine characteristics are endorsed is important. TABLE 2.1 SCORING CATEGORIES FOR MEASURES OF FEMININITY AND MASCULINITY AS INDEPENDENT DOMAINS Femininity Masculinity High Low (Above Median) (Below Median) High (Above Median) Androgynous Masculine Low (Below Median Feminine Undifferentiated (Indeterminate) The 60 items of the BSRI are given to subjects with instruc- tions to rate themselves on each of the adjectives using a seven—point scale, ranging from 1, "Never or almost never true of me," to 7, "Always or almost always true of me." The scores for the Feminine scale items, and for the Masculine scale items are added to derive the Feminine score and the Masculine score. Because half of the Social Desirability scale items are negative characteristics, their scores are reversed before they are added to the other item scores for this scale. Spence, Helmreich, and Stapp (1974, 1975) were the first to use four scoring categories in their 55-item Personal Attributes Questionnaire (PAQ): 28 l. High-feminine, low-masculine 2. Low-feminine, high-masculine 3. High-feminine, high-masculine 4. Low-feminine, low-masculine Low and high ratings are determined by whether or not the subject scores above the median or below the median on each scale. (In recent studies, Bem has also used the median split technique.) The FAQ includes 18 items considered ideal for all people, but more typical of women; 23 items considered ideal for all people, but more typical of men; and 15 items considered "gender-specific"2 only. Each of the items is bipolar, for example: Not at all Very independent A . . B . . C . . D . . E independent Subjects choose the letter which indicates where they think they fall on the scale. The items were chosen from the Sex Role Stereotype Questionnaire developed by Rosenkrantz, Vogel, Bee, Broverman, and Broverman (1968). Using items from the Personality Research Form (PRF), Berzins, Welling, and Wetter (1978) developed the PRF ANDRO scale of 56 items. The original pool of 64 items they used was chosen on a "rational- intuitive" basis, consistent with themes found in Bem's Masculine and Feminine scales. Later, the items were narrowed down, and validated by psychometric research. For example, 177 judges were given the PRF items and asked to rate them on a seven-point scale ("not at all desirable" to "extremely desirable") in response to the question, 2In 1977 Spence and Helmreich reported that they now call these gender-specific items "Masculinity-Femininity" because this scale is bipolar in nature, in contrast to the separate Masculinity and Femininity scales. 29 "In American society how desirable is it for a MAN (alternately WOMAN) to mark this item TRUE?" (p. 128). Half the judges rated the items for men, and half for women; analysis of variance showed all 56 of the final items to be significant in the predicted direction. The PRF items are statements which are marked true or false by subjects. The PRF ANDRO is scored in the same way the the PAQ and BSRI are scored, and was designed to parallel the function of the BSRI, using standard personality test items. Heilbrun (1976) has developed a scale using items from Gough's and Heilbrun's Adjective Check List (ACL; 1965). Heilbrun intended to present a scale which would be similar to the BSRI, PAQ, and PRF ANDRO . in measuring masculinity and femininity as independent. However, the basis for selection of the 54 items of the Masculinity-Femininity scale hindered this intention. As Heilbrun reports it, The Masculinity-Femininity Scale was derived by identifying those adjectives that discriminated between college males identified with masculine fathers and college females identified with feminine mothers (Cosentino & Heilbrun, 1964). This approach to scale devel- opment sought to compile items that distinguished between two extreme criterion groups differing not only in terms of biological maleness/ femaleness but also in terms of psychological masculinity/fem- ininity. (p. 184) Although he has said that masculinity and femininity have been shown to vary independently in his instrument because the correlations be- tween the Masculinity and Femininity scales are low, Heilbrun has not developed his instrument using a dualistic concept of masculinity and femininity. Echoing Constantinople's (1973) criticism of earlier studies and measures, Kelly and Worell (1977) say that in Heilbrun's instrument a problem occurs because, . . . gender differences in frequency of adjective endorsement was a criterion for item inclusion. This would seem to reintro- duce the question of whether this scale is assessing psychological A? __—_... 30 sex roles in the same manner as conceived by Bem and Spence et a1. or whether it confounds gender differences in item endorsement in the bipolar manner. . . . (p. 1106) In addition, Heilbrun's use of the concept of identification (e.g., "college females identified with feminine motherS") is subject to controversy because of the disagreement about its definition among researchers. Kelly and Worell (1977) have summarized and reviewed these measures and the efforts to validate them. They point out that the validation studies that have been conducted using the new measures of masculinity and femininity can be placed into three categories: "(a) personality correlates to sex role style, (b) parental and auto- biographical variables related to sex role adoption, and (c) behavioral differences among sex-typed and androgynous individuals" (p. 1107). Using the categories suggested by Kelly and Worell, the hypotheses presented in this study can be divided into those addressing "personality correlates" (i.e., self-esteem in the current study), and those addres- sing "parental and autobiographical variables" (parents' self-esteem, sex role identities, and behavior toward daughters in this study). Self-Esteem and Sex Role Identity Spence, Helmreich, and Stapp (1975) used the Texas Social Behavior Inventory (TSBI) to measure subjects' self-esteem as related to their scores on the PAQ. The TSBI is described by Spence, Helmreich, and Stapp as an instrument "designed to determine individuals' self- confidence and competence in interpersonal situations and is generally described as a measure of social self-esteem" (1975, p. 31). These 31 researchers found very high positive relationships between masculinity and self-esteem and between femininity and self-esteem in both sexes. They also found that androgynous subjects reported the highest self- esteem; with subjects reporting high-masculine, low-feminine next; then subjects reporting high-feminine, low-masculine; and last, subjects with low-masculine, low~feminine. The descending order of categories and self—esteem presented above held for both sexes. Spence, Helmreich, and Stapp (1975), suggesting that perhaps endorsing both masculine and feminine characteristics results in higher self-esteem, state the following: The data suggest that androgyny, conceived of as the possession of a high degree of both masculinity and femininity, may lead to the most socially desirable consequences, the absolute strengths of both components influencing attitudinal and behavioral outcomes for the individual. (p. 35) Of course, it is possible to argue the relationship between androgyny and self-esteem in the opposite direction as well, i.e., to say that high levels of self-esteem would encourage some individuals to perform in situationally-appropriate ways, even if they had to cross sex-role boundaries to do so. Perhaps high levels of self-confidence would insulate the individual against anxiety about endorsing cross-sex characteristics. Without further information, it is appropriate to conclude that there is a relationship between androgyny and self-esteem, but not to conclude that one precedes the other, or causes the other. In a 1977 study, Bem used the Texas Social Behavior Inventory with the BSRI to evaluate the relationship of self-esteem with her newly modified sex role categories. She found the predicted relation- ship between androgyny and self-esteem, and, in fact, had findings 32 similar to Spence, Helmreich, and Stapp (1975), "with feminine and undif- ferentiated subjects low in self-esteem and masculine and androgynous subjects high in self-esteem" (Bem, 1977, p. 200). However, unlike Spence et a1., Bem did not find rank order patterns of self-esteem for the four sex role categories for both men and women. For women, but not for men, self-esteem was positively related to both masculinity and femininity, using multiple regression analysis. In Bem's sample, self- esteem was positively related to masculinity but not to femininity for men. So the rank order of the four sex role categories along the continuum of self-esteem did not hold for men in Bem's study, although it did hold for the women. Kelly and Worell (1977) suggest in their critique that "high self-esteem is related mainly to the presence of masculine-typed behavior capabilities and minimally to the presence of feminine-typed characteristics" (p. 1108). While this suggestion is congruent with evidence that American society evaluates masculine characteristics more positively than it does feminine characteristics. it is not congruent with the findings of Spence et a1., nor with Bem's findings about women, sex role identity and self-esteem. In the current study, the self-esteem and sex role identity question has been pursued again, but a shorter self-esteem inventory than the one used by Spence et a1., and by Bem has been used: The Rosenberg Self-Esteem Inventory (RSEI, Rosenberg, 1965). Parental Variables and Sex Role Identipy Prior to the 1970's extensive work had been done, especially with young children, on the effects of parents' behavior upon the development of measures of masculinity and femininity as independent 33 domains, however, not much research has been done on parental variables. Before describing the one known study of orthogonal measures and parental variables, a short review of some relevant ideas about socialization will be presented. Traditional Analytic Views The socialization process is known to include, at least, the effects of parents, other adults, schools, peers, media, and experience. The traditional view, derived from Freud, of the effects of the parents on sex role identity has been that people learn the "appropriate," i.e., same-sex, sex role identities through the assumed process of identification with the same-sex parent. This identification is thought to take place at an early age for the individual, as part of the resolution of the Oedipal conflict. The young girl, who was originally attached primarily to her mother is thought to focus her attention on her father as a sexual object. Finding out that she cannot replace her mother in her father's affections (if all goes well for the successful resolution of the Oedipal conflict) the young girl supposedly returns her attention to the relationship with her mother by identifying with her, i.e., taking on characteristics similar to her mother's. The part of this process which is assumed to be motivated by the daughter's feared loss of mother is called anaclitic identifi- cation. The part which is motivated by the daughter's fear of punishment by mother for her rivalry for father is called defensive identification or identification with the aggressor. Some writers emphasize anacli- tic identification for girls and defensive identification for boys. 34 Whether anaclitic or defensive, or both, this identification is said to begin taking place early in life (3 - 5 years). Erikson (1963), in his presentation on stages of development describes "Identity vs. Role Confusion,‘ a stage said to coincide with puberty and adolescence of the individual. Some authors (e.g., Newman and Newman, 1975) have said that this period involves a recapitulation of sex role identity issues. The difference between the sex role identity task at adolescence and at an early age has been presented by Newman and Newman (1975) in the following way: Several critical experiences occur between early school age and later adolescence which result in a reconceptualization and consolidation of the sex-role identity. First, the child engages in close same-sex peer relationships. These friendships teach the child about the possibility of intimacy among equals. They also expose him to peer norms for appropriate sex-role behavior. During the stage of early adolescence, the impact of the peer group ex- pands to communicate expectations about heterosexual relationships as well as same-sex friendships. Early adolescence also brings the onset of physical changes which the child must incorporate into his or her sex-role identity. (p. 223) The adolescent is thought to be engaging in a great deal of experi- mentation along the path of identity formation--"psychosocia1 mora— torium" in Erikson's terms (1959). The product of this experimentation, or the resolution of the crisis of this life stage is the development of a secure identity which is said to include the choice of the appro- priate sexual role (vs. "Bisexual Confusion"). More Recent Analytic Views Also within psychoanalytic tradition are the ego psychologists and object relations theorists, but their views of development and personality sometimes lead to slightly different conclusions about 35 sex role identity. First of all, some of these writers do not place as heavy an emphasis upon biopsychological determinism as do the classic Freudian theorists. Ego psychologists do not negate the biological component of personality development, but they include additional elements. In her overview of psychoanalytic theory, Chodorow (1978) describes some of these differences: Ego psychology begins with an acceptance of Freud's drive theory. . . It adds a concern with another inborn faculty--the 'system ego'--a combination of functions or 'apparatuses' (per- ception, memory, cognition) in the first instance independent of drives and of psychological conflict. (p.45) However, object relations theorists de-emphasize the biological some— what, and emphasize the social influences upon development, expanding. the ideas of the "cultural school" psychoanalysts such as Horney, Fromm and Thompson. Chodorow, acknowledging the acceptance of the importance of sexuality especially in early development by object relations theorists, also says, However, object-relations theory is distinguished from the instinctual determinists by its different conception of the role of drives with respect to the formation and expression of sexu- ality. Object-relations theorists argue that the child's social relational experience from earliest infancy is determining for psychological growth. (p. 47) Thus the quality and variety of relationships with others, e.g., with the parents, can be seen to influence the individual's development, and sex role identity can be seen as the result of psychodynamic factors within relationships. Sex role identity according to the tenets of object relations theory would not be the simple result of the resolution of the Oedipal conflict as found in traditional 36 Freudian theory, but the result of psychophysiology, the "ego apparatuses" of the ego psychologists and the "person, self, sub— ject in relationship, with conscious and unconscious motives and intentions" (Chodorow, 1978, p. 49). Thus, an active cognitive role of the individual in his or her own development is one of the features of the newer psychoanalytic traditions. However, cognitive processes are central to the develop- mental view of sex role identity as set forth by Kohlberg (1966). Cpgnitive-Developmental Theory The role of individual cognition in development is at the core A of Kohlberg's (1966) ideas about sex role attitudes. Kohlberg separates his ideas from those of the analytic tradition because he says that he refutes the idea of sex role identity as the result of biological instinct. Kohlberg also differs from the analysts in his perception of the role of identification. In simple terms, the analytic tradition seems to say that identification leads to sex role identity, and the cognitive-developmental idea is that sex role identity leads to identif- ication. Kohlberg might say that first the child establishes the fact that she is female, than she looks to the environment to find out how to be and act as a female. She may "identify" with mother because mother provides information or behavior which when c0pied confirms the child's desire to be effective and good, or because she sees mother as similar to how she sees herself. Freudians would cite instinct, sexual drives, and dependency, and behaviorists would cite reinforcement and 37 modeling, but Kohlberg says that the individual's struggle for mastery and a positive self—image is the motivation behind identification with one's parent. Kohlberg does include observational learning or modeling in his theory, but does not agree with many other facets of the behavioral view because they imply that the individual is the passive subject, shaped only by external forces. While including observational learn- ing Kohlberg (1966) points out that, this learning is cognitive in the sense that it is selective and internally organized by relational schemata rather than directly reflecting associations of events in the outer world. In regard to sex-role these schemata that bind events together include con— cepts of the body, the physical and social world, and general categories of relationship. . . . (p. 83) Kohlberg's ideas are derived from the work of Piaget, and many elements run parallel to Piaget's theory. For example, Piaget's theory addresses the changes in cognitive processes about the physical world which are related to age. Kohlberg's ideas address the changes in the individual's cognitive organization of the social world which accompany age changes. Kohlberg cites research on children which provides evidence that they develop an idea of "having an unchangeable sexual identity conceptions of the invariable identity of physical objects" (p. 83), an element of Piaget's theory of cognitive development. The role of parents is their part in the child's social world, as objects of observational learning, as objects of identification, and as sources of information for the child's cognitive organization of sex role identity. Kohlberg says that, "many research findings seem to indicate that parent attitudes differentially stimulate or retard the 38 development of many basic sex-role attitudes, rather than teaching them directly through reinforcement or identification" (p. 84). The belief that the parents' attitudes do have an effect on their child's sex role identity was reflected in the planning of the current study, although it has college-age women as its subjects. While Kohlberg's work is directed primarily at early development, a logical extension of many of his principles can be made to late adolescence, during which cognitive development and experience of the environment are still taking place. One of the extensions which was made in the present study was the importance of knowing something about the parents' attitudes because of their potential effect on the daughter's sex role identity. Independent Measures of Masculinipy and Femininity and Parental Variables Two researchers (Kelly & Worell, 1976) recently conducted a study which included the principle of measuring masculinity and femininity as separate dimensions in an investigation of parental behavior as related to sex role identity. They used social learning as their theoretical base and assumed that Androgynous subjects would have modeled after both of their parents. Kelly and Worell employed the PRF ANDRO (Berzins, Welling, & Wetter, 1978) as their measure of sex role identity and the Parent Behavior Form (PBF, WOrell & Worell, Note 1) as their measure of the parents' behavior toward their children. The PBF (which will also be discussed in Chapter IV) is an instrument composed of 15 scales of nine itens each, of which 13 are used in the analysis of the behavior of each parent. The items are short descriptive statements, which subjects are asked to mark as 39 ' or "not like" their parents at the time the "like," "somewhat like,‘ subjects were 16 years old. The 13 scales used for the analysis of Kelly's and Worell's study were: Warmth, Active Involvement, Egali- tarianism, Cognitive Independence, Cognitive Curiosity, Cognitive Competence, Lax Control, Conformity, Achievement Control, Strict Control, Punitive Control, Hostile Control, and Rejection. The 13 scales have been grouped, on the basis of factor analysis, into three factors. The first is Warmth-versus-Rejection, the second factor is Control, and the third is Cognitive Involvement. The factors are formed through the process of adding (and sometimes subtracting) the scores of the scales found to relate to the factor. Their study included both male and female college undergraduates, but only the results for the women subjects will be discussed in the current review. On the basis of the PRF ANDRO scores, subjects were divided into the four sex role categories noted earlier: Androgynous, Masculine-typed, Feminine-typed, and Indeterminate. Analysis of variance was used to investigate the differences among the four categories for the PBF's scales on each parent and followed by least significant difference analysis (see Table 2.2) "to determine phipp_individual sex role categories differed from one another where significant parent scale differences across all categories occurred" (p. 846). Both Androgynous and Masculine-typed women reported receiving high levels of cognitive and achievement encouragement from their parents, and Indeterminate women reported the least encouragement. In addition, Androgynous women reported the highest levels of maternal encouragement for Cognitive Curiosity and were higher than Masculine-typed women on maternal mo.V a .3. 40 8. V a « «SV «SV «av «EV «EEm8 + «SV «SV «EV «LEV «HA «HA adv «EHwoo 1. «EV RSV *HA «m V «m V «H A «H A «a A «ao>zu m>Hequoo «2A «EV «HA .«HA «HA «xv «EEwoo + « mazm<3 hmfiumh Hug—Ho: Hwfiumh umzuoz Hwiumh Hozuoz Mosumh Hwfiuoz Okum Ase measzmmemazH Ame mszazme Axe azaasom mmomo w.noH ocfi o~ e.nm oca cw nonuwh o.co oca ON o.~o~ oqa ON wonuoz H.0m oaH ON m.~oH oca 0N uHom Anaoua Hamv «wmbomu onH mmomo m.cH~ cad om o.nw ocH ow nonumm m.~m oca ow m.moH oca ON Macao: «.ma cca ow m.moH oqa ON «Ham «ADOKU onH compo: ouoom ouoom couvo: ououm ouoom AHazH Ebfiqxux aaawcfiz Enauxm: aaaacwz mqmz¢ovzw>z~ udou Kmm Sum m.¢ mqm<fi 79 .A~.e unawah ooov wunv madame «an: vacuum mo manuamcw :« pom: mauomca .asnfla Eoua vouavumvcmum « Hm. mo.n~ ee.ma ma ««~a. so.- No.me ma possum «uam. an.m~ mn.om ma ««~a. ma.~H ~s.mm AH guano: «48a. «m.na o~.a~ ea ««a~. e~.a me.ma «a «How Anson“ uuuqaaav macau onaummomu ««Hm. om.aa o~.moa om cw. HH.¢H ms.ew om possum mm. Hm.nH ea.~a om hm. ~H.na as.ocfi oN guano: aw. a~.n~ mm.na o~ an. Ho.o e~.~oH om uaom Amauua Hana macaw zo~aummomo No. om.o~ as. «suaaanu mm : aquH asuaaapa em = msmuu q<=nH>anH uaauz mo nuaasz uaaox uuaasz ugmzmoezm>z~ mqom xmm 2mm mmh mo mm4ma Qm .80) for both parents. Three scales, Cognitive Competence, Achievement Control, and Rejection had reliabilities (standardized item alpha) which were below .80. The lowest was a standardized item alpha of .58 in the validation sample's Achievement Control scale for Mother (see Table 4.6). After computing the reliabilities of the PBF scales for the validation group only four of the nine scales were predicted to increase in reliability through removal of items. Two of the scales were predicted to show an increase for both Mother and Father, and two were predicted to increase for Father only. Appendix D contains the list of the scales and their 83 .a:n~< Baum vanavumvcmuma ---- Na. ca.” os.- Na. ma.N o~.- ---- AN. mo.~ HG.~H as. on.~ -.- zeapomqmz an. on.” -.o~ a mm. cm.n ~a.- ax. as.n cm.z_ u--- an. ex.m Ae.~a am. o~.s -.~a noxezou uaasmcz ---- as. m~.s mo.m~ he. on.m aa.s~ n--- ma. ~o.e mm.a~ mm. mm.~ s~.s~ aoaazou azm=m>mazu< ---- ma. az.s Ne.wa on. ~o.o oo.ma .u.. «a. ha.m Ac.o~ as. No." oo.m~ mozmammzoo M53288 as. ea.” nn.m~ m on. n~.s as.c~ an. na.s o~.o~ u--- mm. as.s om.aa mm. mm.s om.m~ spamoamau m>~e~zuou ---- mm. Nw.n m~.- cw. ~s.s om.- n--- om. a~.n oe.- om. so.s ma.- muzmazmamoza m>~aazuou u--- mm. on.” am.z~ aw. ms.s ms._~ ---- aw. an.n -.m~ an. mm.n Ho.m~ amaz<~x4o>zw usaau< ow. o_.s s~.mfi w am. xe.s ma.- Ea. Hw.s -.H~ om. mm.n hm.o~ w mm. mm.n a~.m~ as. an.a so.m~ mazaummcxu zoaa-mmo¢o zo_p zo~eummomo onpnmmomu zo~a ¢m=ama QM .80), and the remaining three scales can be characterized as moderately reliable in both samples groups. The RosenberggSelf-Esteem Inventory (RSEI) The RSEI consists of 10 sentences to which the women responded using a four point scale. The subjects completed the RSEI once for themselves, and once for each of their parents. The pronouns in the instrument's statements were changed accordingly, because Rosenberg's original version was all first-person statements. Appendix G contains the three versions as the women received them. Five of the RSEI items are scored as high self-esteem if marked 3 or 4, whereas the other five statements, if marked 3 or 4 it would indicate low self-esteem. Initially, Rosenberg (1965) used a Guttman technique to score the inventory. However, it was decided that for the current study the items would be scored separately and a check dOne on the instrument's reliability with that method of scoring. Because half of the items are negative, and half are positive, the women's actual scores had to be recoded during the analysis so that they would be unidirectional, and so that high scores would indicate 86 high self-esteem. After completing the recoding process, the reliabil- ities were computed and found to be acceptable for all three versions of the RSEI administered to the subjects of this study. None of the standardized item alphas fell below .85. Table 4.7 presents the means, standard deviation, and reliabilities for both half samples, and the Self, Mother, and Father versions of the RSEI. TABLE 4.7 MEANS, STANDARD DEVIATIONS, AND RELIABILITIES OF THE ROSENBERG SELF-ESTEEM INVENTORY FOR SELF, MOTHER, AND FATHER, VALIDATION AND CROSS-VALIDATION GROUPS SAMPLE INDIVIDUAL M SD Reliability* VALIDATION GROUP Self 32.06 4.57 .88 Mother 32.72 5.56 .85 Father 33.07 5.73 .91 CROSS-VALIDATION GROUP Self 31.83 4.59 .85 Mother 31.74 6.08 .89 Father 33.41 5.96 .89 * Standardized item alpha. Summary Extensive work was done to establish sample reliabilities on the three major instruments used in the current study. The three instruments, the Bem Sex Role Inventory (BSRI), the Parent Behavior 87 Form (PBF), and the Rosenberg Self-Esteem Inventory (RSEI) were examined separately for the validation and cross—validation samples. The results from the first group were used to predict improvement in the second group's reliabilities through elimination of items in the BSRI and PBF which were not correlating well enough with the other scale items. The length of the scales was also considered in the elimination of items. No items were eleminated from the RSEI. The two scales used from the BSRI were Masculinity and Femininity. Data was obtained for Self, Mother, and Father from each woman in the study. All of the scales except Masculinity for Father improved in the cross—validation group reliabilities through the elimination of a few items. The improved scales were used in the analysis of the cross-validation group's data. The reliability figures of all of the scales were high (none below .80) prior to the elimination of any items, indicating the strength of the BSRI as a whole. Nine of the Parent Behavior Form's 15 scales were used in the present study. The scales used were those which had been described in earlier research as the components of a warmth-versus-rejection factor and a parental cognitive involvement factor. Only four of the nine scales were predicted to make any improvement in this study through the elimination of items, and only one of these four did, in fact, improve for the cross-validation group. Six of the nine PBF scales were highly reliable for the current sample and three were moderately reliable. If a scale's standardized item alpha was not at least .80 in the cross-validation group, it was not used in the analysis of the cross-validation group's data. 88 The Rosenberg Self-Esteem Inventory, which contains 10 items, was found to be highly reliable for both the validation and cross- validation groups. Each item was scored separately, rather than in clusters, and the items were recoded in order to have high scores represent high self-esteem. The summary of information about the findings on the three instruments used in the current study are presented in Table 4.8. 89 ram 5.0 W OF mum FINDINGS. an! an IO“ M (I‘ll). um “VIM m (PIP). A” aoamnc SELF-[STEIN MN! (RSEI) INSTRUMENT Baa the Here it- Did it- flee the Hhich aeale validation ehoeen for elimination tron-validation vaa uaed in SCALE group a elimination increaae acale'a group a the anew“. .80 or beeauae of a for the .00 or of crou- SUBJEC‘I 0F higher? validation emu-validation higher? validation, SCALE group'a reaulta? group? group data. BSRI MNINITY Self yea yea yea no (.79) modified Mother yea yea yea yea modified Father yea yea yea yea modified MSCULINITY Self yea yea yea yea mdified Mother yea yea yea yea modified Father yea 1" 0° V“ {“11 PBF (Factor) WARM"! (WVR) Mother y“ yea no yea full Father y“ yea aome yea modified ACTIVE INVOLVEMENT (WVR) Mother ,0 yea yea yea modified Father y“ yea no yea full EGALI‘IARIM'ISH (WVR) Mother yea no -- yea full Father yea no -- yea full COGNITIVE INDEPENDENCE (COG, WVR) Mother yea no -- yea full Father yea no -- yea full COGNITIVE CURIOSII'Y (COG) Mother yea no -- yea full Father yea yea no yea full COGNITIVE COHPETENCE (COG) Mother no (.67) no -- no (.74) not need Father no (.76) no - no (.75) not need ACHIEVEMENT CONTROL (COG) Mother no (.58) no -- no (.78) not need Father no (.67) no -- no (.79) not used HOSTILE CONTROL (WVR) lbther yea no -- yea full Father yea yea no yea full IEJEC‘I'ION (WVR) Mother no (.74) no -- no (.79) not uaed Father no (.77) no -- yea full RSEI Self yea no -- yea full Bother yea no -- yea full Father yea no -- yea full CHAPTER V RESULTS OF THE ANALYSIS In the chapter which follows, the hypotheses which were for- mulated for the study will be restated, in full and symbolic form. Following each hypothesis the results of the statistical tests which were conducted will be presented, with validation and cross-validation groups shown separately. The format of presenting both samples' results together was chosen to aid the reader's understanding of the general cross-validation results. The hypotheses and results will be organized according to the variables studied, as presented in Chapter III, Design of the Study. The null hypothesis will be presented first, and in those cases in which analysis of variance and £_tests of contrasts were used, the analysis of variance table will be presented following the null hypothesis. The results of all the hypotheses are summarized in Table 5.1. Finally, the results of the survey (the . Family Information Questionnaire, FIQ) given to the women about their relationships with their parents will be presented. Sex Role and Self-Esteem A set of hypotheses were formulated about the relationship between the sex role category assigned to the women subjects on the basis of the Bem Sex Role Inventory (BSRI) and the degree of self- esteem they reported. The following null hypothesis was tested: 90 91 TABLE 5.1 SUMMARY OF RESULTS OF STUDY OF COLLEGE WOMEN'S SEX ROLE IDENTITY, SELF-ESTEEM, AND PERCEPTIONS 0F PARENTS RESULTS SYMBOLIC HYPOTHESES Cross- Validation Validation Group Group HO: SEA= SEF: SEF!8 SEU F 3 6038* F . 5e35* Hla: SEA >SEF, SEM, SEU t_= 2.298* £_= 3.134* Hlb: SEU SEF _.= 2.541* £_= l.OO7 H0: SEMA= SEMF= SEMM= SEMU F = 3.38* F = 3.53* Hla: SEMA >SEMF, SEMM, SEMU £_= 3.017* “E = 1.253 Hlb: SEMU SEFF, SEFPI’ SEFU E g 3e112* "' Hlb: SEFU (SEFA, SEFF, SEFM £ 3 1.578 ”'- ho: No relationship between SE & SEM H1 ° Positive relationship between r = .23* r = .24* SE and SEM HO: No relationship between SE & SEF H1 . Positive relationship between r = .28* r = .21* SE and SEF H0: No relationship between SE & COGM H1 . Positive relationship between r = .19* r = .35* SE and COGM 92 TABLE 5.l--Continued SYMBOLIC HYPOTHESES HO: No relationship between SE & COGF H1: Positive relationship between SE and COGF H0: No relationship between SE & WVRM H1: Positive relationship between SE and WVRM H0: No relationship between SE & WVRF H1: Positive relationship between SE and WVRF H0: COGMA= COGMF= COGMM= COGMU H13: COGMA> COGMF, COGMU Hlb: COGMU (COGMA, COGMF, COGMU H0: COGMA = COGMM H0: COGFA= COGFF= COGFM= COGFU H13: 0061;, > COGFF, COGFU Hlb: cocru<:cocFA, COGFM, COGFF H0: COGMA =' COGFM H0: WVRMAB WVRMF= WVRMM= WVRMU H18: WVRMA>WVRMF, WVRMm, WVRMU Hlb: WVRMuWVRFF, WVRFM, WVRFU -- f; = 1.714 Hlb: WVRFUI SE F’ M’ U In both the validation and cross—validation samples, the comparison of group means yielded significant differences in the predicted direction. In the validation group, £_= 2.298 (p <.05, one-tailed), and in the cross- validation group, 3 = 3.134 (p <.05, one-tailed). The next contrast was planned on the basis of the following hypothesis: Hlb: Subjects classified as Undifferentiated by their scores on the BSRI will score lower on the RSEI than subjects classified as Androgynous, Feminine, or Masculine. Stated symbolically: H : SE < SE , SE , SE 1b U A F M In both sample groups, the women categorized as Undifferentiated scored significantly lower on self-esteem than the other three BSRI categories. The validation sample produced at_t value of 3.165 (p <.05, one-tailed), and the cross-validation group had 3 = 3.047 (p <.05, one-tailed). The last hypothesis which was tested for subjects' sex role and subjects' self-esteem was: ch: Subjects classified as Masculine by their scores on the BSRI will score higher on the RSEI than subjects classified as Feminine. 96 In symbolic form: 0 > ch. SEM SE F For the validation group, comparison of the means of these two categories of subjects did yield a significant result (5 = 2.541, p <.05, one- tailed). However, the t test of the same contrast in the cross—validation group was not significant at the predicted level (5 = 1.007). Therefore, the significant finding in the validation group may be spurious. Sex Role and Parents Self-Esteem Another set of hypotheses which were tested in this study were formulated around predictions of the women's reports of their parent's self—esteem on the basis of the subjects' own sex role categories. Sex Role and Mothers' Self-Esteem The first of the null hypotheses about parents was: H0: .Ng difference will exist among the four groups of subjects established by scores on the BSRI in the Self-Esteem scores (SEM) they ascribe to their mothers on the RSEI. In both sample groups, the analysis of variance (Table 5.3) of the means of the four groups of subjects produced by BSRI scores generated significant values. In the validation group, the null hypothesis was rejected because of the F value of 3.38 (p‘<.05), and in the cross-validation sample, the sex role groups were assumed to be different because F = 3.53 (p <.05). 97 TABLE 5.3 ANALYSIS OF VARIANCE OF THE ROSENBERG SELF-ESTEEM INVENTORY SCORES FOR MOTHER FOR THE FOUR BEM SEX ROLE INVENTORY CATEGORIES OF COLLEGE WOMEN SUBJECTS, VALIDATION AND CROSS- VALIDATION GROUPS SAMPLE GROUP SOURCE df SS MS F VALIDATION GROUP * Sex Role Category 3 293.73 97.91 3.38 Within-groups 84 2434.26 28.98 Total 87 2727.99 CROSS-VALIDATION GROUP * Sex Role Category 3 375.17 125.06 3.53 Within—groups 84 2978.27 35.46 Total 87 3353.44 p < .05 Planned contrasts were also tested to compare subjects' reports of their mothers' self-esteem on the basis of the subjects' own sex role categories. The first alternate hypothesis was: H18: Subjects classified as Androgynous by their scores on the BSRI will report higher levels of self-esteem on the RSEI for their mothers than subjects classified as Feminine, Masculine, or Undifferentiated. Stated symbolically: H : SEM > SEM , SEM , SEM la A F M U In the validation sample, the comparison of group means indicated that Androgynous subjects reported higher self-esteem for their mothers than the other three groups (£_ 8 3.017, p <.05, one-tailed). However, the cross-validation group did not produce the same results when group 98 means were compared (£_= 1.253), therefore, the Androgynous category cannot be said to report higher maternal self-esteem than the other three sex role categories for the cross-validation sample. Another hypothesis formulated about sex role category and maternal self-esteem was: Hlb: Subjects classified as Undifferentiated by their scores on the BSRI will report lower levels of self-esteem on the RSEI for their mothers than subjects classified as Androgynous, Feminine, or Masculine. The symbolic form is: H : SEM < SEM , SEM , SEM 1b U A F M In the validation sample the test of the contrast did not show that Undifferentiated subjects reported lower self-esteem for their mothers than the other three categories (t_= -.325). However, in the cross-validation sample the £_va1ue obtained was significant (£_= 3.180, p <.05, one-tailed), indicating that the Undifferentiated subjects reported significantly lower levels of maternal self—esteem than the other three sex role categories. Sex Role and Fathers' Self-Esteem Similar to the hypotheses about subjects' self—esteem and their mothers' self-esteem were the following hypotheses about the women's self-esteem and their reports of their fathers' self-esteem. The null hypothesis was: H0: .Ng difference will exist among the four groups of subjects 99 established by scores on the BSRI in the self-esteem (SEF) scores they ascribe to their fathers on the RSEI. Stated in symbolic form: H : SEF = SEF = SEF = SEF O A F M U Through the results of the analysis of variance (Table 5.4) the four sex role groups of the validation group have been shown to be different (F = 3.70, p<:.05). The four groups were not shown to be significantly different in the cross—validation group subjects' reports of paternal self-esteem (F = 2.04). TABLE 5 . 4 ANALYSIS OF VARIANCE OF THE ROSENBERG SELF-ESTEEM INVENTORY SCORES FOR FATHER FOR THE FOUR BEM SEX ROLE INVENTORY CATEGORIES OF COLLEGE WOMEN, VALIDATION AND CROSS-VALIDATION GROUPS SAMPLE GROUP SOURCE df SS MS F VALIDATION GROUP * Sex Role Category 3 336.78 112.29 3.70 Within-groups 85 2581.85 30.38 Total 88 2918.72 CROSS VALIDATION GROUP * Sex Role Category 3 218.87 72.96 2.04 Within-groups 81 2899.91 35.80 Total ' 84 3118.78 * p < .05 Because the analysis of variance of father's self-esteem by sex role category was not significant for the cross-validation sample, the t tests of the predicted contrasts are not relevant. In the validation group, the first alternate hypothesis to be tested for these two variables was: 100 Hla: Subjects classified as Androgynous by their scores on the BSRI will report higher levels of self-esteem on the RSEI for their fathers than subjects classified as Feminine, Masculine, or Undifferentiated. In symbolic form: H : SEF' > SEF , SEF , SEF 1a A F M U The £_test comparing group means did produce a significant value (p = 3.112, p <.05, one-tailed), indicating that in the validation group, the Androgynous women reported higher levels of self-esteem for their fathers than the other three sex role categories. The next alternate hypothesis which was tested for the validation group was: H : Subjects classified as Undifferentiated by their scores on lb the BSRI will report lower levels of self-esteem on the RSEI for their fathers than subjects classified as Androgynous, Feminine, or Masculine. Stating the hypothesis symbolically: H : SEF < SEF , SEF , SEF lb U A F U When the above contrast was tested, the value obtained was not significant (5 - 1.578). Therefore, neither the validation nor the cross-validation group offered evidence that Undifferentiated subjects would be more likely to report low levels of paternal self-esteem than the Androgynous, Feminine, or Masculine college women. 101 Subjects' Self-Esteem and Other Parental Variables A series of hypotheses was tested about the women's self- esteem as affected by parental variables other than sex role. The results of the tests of correlation are shown in Table 5.5, and will be discussed individually in the following sections. Subjects' Self—Esteem and Perceived Parental Self-Esteem Two null and alternate hypotheses were tested about the relationship between the women's views of their own self-esteem and their views of each of their parents' self-esteem using the Pearson product-moment correlation statistic. The first of these two sets of hypotheses was: HO: N2 relationship will be found between subjects' self-esteem scores on the RSEI and the self-esteem scores they ascribe to their mothers on the RSEI. H : A positive relationship will be found between subjects' self- esteem scores on the RSEI, and the self—esteem scores they ascribe to their mothers on the RSEI. In both the validation and the cross-validation groups, the value obtained was significant in the predicted direction. In the validation group r = .23 (p< .05, n = 82), and in the cross-validation group, r = .24 (p‘<.05, n = 86), indicating a positive relationship between the women's reports of self-esteem and their reports of mother's self- esteem. 102 TABLE 5.5 SUMMARY OF CORRELATIONS BETWEEN SUBJECTS' SELF-ESTEEM AND PARENTAL VARIABLES OTHER THAN SEX ROLE, VALIDATION AND CROSS— VALIDATION GROUPS DEPENDENT VARIABLE INDEPENDENT VARIABLE VALIDATION GROUP CROSS-VALIDATION GROUP n r n r Subjects' Self-Esteem Mother's Self-Esteem 82 .23* 86 .24* Father's Self-Esteem 85 .28* 83 .21* Mother's Cognitive S4 .19* 88 .35* Involvement Father's Cognitive 86 .09 89 .21* Involvement Mother's Warmth-versus- 84 .31* 89 .33* Rejection Father's Warmth-versus- 86 .14 88 .28* Rejection *p < .05. The second set of hypotheses which were tested were: H : 'Ng relationship will be found between subjects' self-esteem 0 scores on the RSEI and the self-esteem scores they ascribe to their fathers on the RSEI. H : A positive relationship will be found between subjects' self-esteem scores on teh RSEI and the self-esteem scores they ascribe to their fathers on the RSEI. Again, in both the validation and the cross-validation groups, the test of relationship produced significant results. The validation group had r - .28 (p <.05, n - 85), and the cross—validation group 103 produced r = .21 (p <.05, n = 83). Thus, the predicted positive relationship between subjects' self-esteem and their reports of their fathers' self-esteem received support in both sample groups. Self-Esteem and Parents' Cognitive Involvement The following two sets of hypotheses were formulated about the relationship between the subjects' self—esteem and their perceptions of their parents' involvement with, and encouragement of, their cogni— tive development. The statistical tests were performed on the data obtained from the Rosenberg Self-Esteem Inventory (RSEI) for self and the Parent Behavior Form (PBF) for mother and for father. The first set of hypotheses, tested with the Pearson product— moment correlation, were: HO: .Ng relationship will be found between subjects' self-esteem scores on the RSEI and the scores they ascribe to their mothers on the PBF's Cognitive Involvement Factor. H : A positive relationship will be found between subjects' self-esteem scores on the RSEI and the scores they ascribe to their mothers on the PBF's Cognitive Involvement Factor. The correlation between the self-esteem reported by the women and their reports of their mothers' cognitive involvement was positive and significant in both the validation and cross-validation groups. In the validation group, r = .19 (p‘<.05, n = 84), and in the cross— validation group r = .35 (p <.05, n = 88), indicating support for the expected relationship between the subjects' self-esteem and their perceptions of their mothers' cognitive involvement. 104 The following set of hypotheses was formulated about the women's self-esteem and their fathers' cognitive involvement: Ho: .Ng_relationship will be found between subjects' self-esteem scores on the RSEI and the scores they ascribe to their fathers on the PBF's Cognitive Involvement Factor. H : A positive relationship will be found between subjects' self-esteem scores on the RSEI and the scores they ascribe to their fathers on the PBF's Cognitive Involvement Factor. The value obtained for the Pearson product-moment correlation in the validation group was not sufficiently large (r = .09, n = 86) to permit the rejection of the null hypothesis. However, in the cross-validation group, the correlation between the subjects' self— esteem and their perceptions of their fathers cognitive involvement was significant, r = .21 (p <.05, n = 89). Self-Esteem and Parents' Warmth-versus—Rejection The following set of hypotheses concern the relationship between the subject's self—esteem and their perceptions of the warmth, contrasted with rejection, they received from their mothers: HO: ‘Ng relationship will be found between subjects' self-esteem scores on the RSEI and the scores they ascribe to their mothers on the PBF's Warmth-versus-Rejection Factor. H : A positive relationship will be found between subjects' self- esteem scores on the RSEI and the scores they ascribe to their fathers on the PBF's Warmth-versus-Rejection Factor. The Pearson product—moment correlation for both the validation and cross-validation groups indicated a positive, significant relationship 105 between the women's self-esteem and the warmth they perceived from their mothers. In the validation group, r - .31 (p‘3.05, n = 84), and in the cross-validation group, r = .33 (p‘<.05, n - 89). Mixed results were obtained when a similar set of hypotheses were tested about subjects' perceptions of fathers' Warmth-versus- Rejection. The null and alternate hypotheses were: HO: N9 relationship will be found between subjects' self-esteem scores on the RSEI and the scores they ascribe to their fathers on the PBF's Warmth-versus Rejection Factor. H : A_positive relationship will be found between subjects' self—esteem scores on the RSEI and the scores they ascribe to their fathers on the PBF's Warmth-versus-Rejection Factor. For the test of the hypothesis in the validation group the correla- tion coefficient was not sufficiently large (r = .14, n - 86) to per- mit the rejection of the null hypothesis. However, in the cross- validation group, the Pearson product-moment test produced a value of r 8 .28 (p <.05, n = 88), indicating a positive relationship between subjects' self-esteem and father's warmth-versus-rejection for that group. Sex Role and Parents' Cognitive Involvement The following group of hypotheses were designed to test the relationship between the sex role categories into which subjects were placed using the BSRI and their reports concerning each parent's level of cognitive involvement. Sex Role and Mother's Cognitive Involvement The following null hypothesis was tested about the college women and their mothers: 106 H : ‘Ng difference will exist among the four groups of subjects established by scores on the BSRI in the scores they ascribe to their mothers on the PBF's Cognitive Involvement (COGM) Factor. In symbolic form: HO: COGMA = COGMF = COGMM = COGMU Analysis of variance (Table 5.6) of the hypothesis produced signifi- cant results only in the cross-validation sample. The F value in the validation group was 1.31, which is not sufficiently high to allow for the rejection of the null hypothesis, or for the testing of the a_priori contrasts. Nonetheless, in the cross-validation group, the F value of 3.59 was significant (p‘<.05), indicating differences among the four sex role categories in their reports of maternal cognitive involvement. 0 TABLE 5.6 ANALYSIS OF VARIANCE OF THE PARENT BEHAVIOR FORM'S COGNITIVE INVOLVEMENT FACTOR SCORES FOR MOTHER FOR THE FOUR BEM SEX ROLE INVENTORY CATEGORIES, VALIDATION AND CROSS-VALIDATION GROUPS SAMPLE GROUP SOURCE df SS MS F VALIDATION GROUP Sex Role Category 3 463.53 154.51 1.31 Within-groups 85 10388.13 118.05 Total 88 10851.66 CROSS—VALIDATION GROUP Sex Role Category 3 615.07 205.02 3.59* Within-groups 86 4509.34 57.09 Total 89 5524.41 *p <.05 107 The first of the alternate hypotheses about sex role and maternal cognitive involvement which was tested was; H13: Subjects classified as Androgynous by their scores on the BSRI will report higher levels of Cognitive Involvement on the PBF for their mothers than subjects classified as Feminine or Undifferentiated. Stated in symbolic form: H13: COGMA > COGMF, COGMU Because the analysis of variance did not produce a significant value for the test of the general null hypothesis in the validation group, the contrast of the alternate hypothesis was tested by the £_statistic only for the cross-validation group. In the cross-validation group, a £_of 2.531 (p <.05, one-tailed) was found, confirming the predicted difference between the Androgynous and the Feminine and Undifferentiated categories in their reports of maternal cognitive involvement. The second of the alternate hypotheses was: Hlb: Subjects classified as Undifferentiated by their scores on the BSRI will report 1233; levels of Qggnitive Involvement on the PBF for their mothers than subjects classified as Androgynous, Feminine, or Masculine. Stated symbolically: Hlb: COGMU < COGMA, COGMF, COGMM The result of the t test of the contrast was significant for the cross-validation group, with a t_va1ue of 2.531 (p‘<.05, one-tailed). The contrast was not tested for the validation group, as noted earlier, because the analysis of variance did not produce a significant F value. 108 The Androgynous and Masculine categories were not expected to differ in their reports of their parents' cognitive involvement because of the assumption that such involvement could be predicted for either sex role style equally well. Therefore, a second null hypo- thesis was constructed for sex role and parental cognitive involvement concerning only the Androgynous and Masculine subjects. For mother's cognitive involvement, the null hypothesis was: HO: There will be 22 difference between subjects classified as Androgynous and subjects classified as Masculine on the BSRI in the scores they ascribe to their mothers on the PBF's Cognitive Involvement Factor. In symbolic form: H0: COGMA = COGMM Because it was expected a_priori that a t test of the group means would not produce a value high enough to warrant the rejection of the null hypothesis, no directional alternate hypothesis was formulated. In the validation group, the analysis of variance (Table 5.6) revealed that significant differences did not exist among the four sex role category groups. Therefore, the £_value of .789 which resulted from the comparison of the Androgynous and Masculine groups was anticipated. The null hypothesis about these two groups could not be rejected. The general statement which can be made is that in the validation group, the Androgynous and Masculine categories did not report sig- nificantly different levels of cognitive involvement from their mothers, nor were these two categories significantly different than the Feminine or Undifferentiated women in their reports of maternal cognitive involvement. 109 The results were different for the cross-validation group, however. The 5 test of the Androgynous and Masculine group means produced a value of 2.092 (p'<.05, two-tailed), indicating that the Androgynous and Masculine groups had reported different levels of cognitive involvement for their mothers (the Androgynous group mean was higher than the Masculine group mean). The expectation that the null hypothesis would not be rejected was not met. Sex Role and Father's Cognitive Involvement The following hypotheses were tested about the relationship between the women's sex role categories on the BSRI and their reports of their father's cognitive involvement, measured by the PBF. The null hypothesis was: H0: N2 difference will exist among the four groups of subjects established by scores on the BSRI in the scores they ascribe to their fathers on the PBF's Cognitive Involvement Factor. In symbolic form, the hypothesis was: H : COGF = COGF 8 COGFM = COGF 0 A F U Inconsistent results were obtained from the two sample groups when the one—way analysis of variance (Table 5.7) was conducted for the test of the null hypothesis. In the validation group the null hypothesis was rejected because of F = 3.00 (p <.05). However, in the cross- validation group, the F value was low, .58, which meant that the null hypothesis could not be rejected, and that testing the g_priori contrasts in that group would not be valid. 110 TABLE 5.7 ANALYSIS OF VARIANCE OF THE PARENT BEHAVIOR FORM'S COGNITIVE INVOLVEMENT FACTOR SCORES FOR FATHER FOR THE FOUR BEM SEX ROLE INVENTORY CATEGORIES, VALIDATION AND CROSS-VALIDATION GROUPS SAMPLE GROUP SOURCE df SS MS F VALIDATION GROUP Sex Role Category 3 1184.13 394.71 3.00* Within—groups 89 11724.34 131.73 Total 92 12908.47 CROSS-VALIDATION GROUP Sex Role Category 3 96.30 32.10 .58 Within-groups 87 4832.89 55.55 Total 90 4929.19 *p < .05 The a_priori contrasts, based on the alternate hypotheses, were tested for the validation group. The first of these was: H1: Subjects classified as Androgynous by their scores on the BSRI will report higher levels of ngnitive Involvement on the PBF for their fathers than subjects classified as Feminine or Undifferentiated. Stated symbolically: H COGFA > COGFF, COGFU 1: For the validation group, Androgynous subjects were found to report higher levels of paternal cognitive involvement than the Feminine and Undifferentiated categories (3 - 2.485, p‘<.05, one-tailed). 111 The second of the alternate hypotheses about sex role and father's cognitive involvement was: Hlb: Subjects classified as Undifferentiated by their scores on the BSRI will report lggg£_1evels of gggnitive Involvement on the PBF for their fathers than subjects classified as Androgynous, Feminine, or Masculine. In symbolic form: Hlb: COGF < COGFA, COGFF, COGFM U The.g test of the hypothesis produced a value of 2.256, which is sig- nificant (p‘<.05, one-tailed). Therefore, the Undifferentiated cate- gory of the validation group did report lower levels of paternal cogni-- tive involvement than the other three sex role categories. As was done with the subjects' reports of their mothers' cognitive involvement, 8 separate, specific null hypothesis was for- mulated about the Androgynous and Masculine women's reports of their faters' cognitive involvement. The null hypothesis was: HO: There will be 39 difference between subjects classified as Androgynous and subjects classified as Masculine on the BSRI in the scores they ascribe to their fathers on the PBF's Cognitive Involvement Factor. Symbolically stated: H0: COGFA = COGFM Again, because it was expected that the null hypothesis would not be rejected, no alternate hypothesis was written predicting a difference between the two groups. 112 The £_test in the validation group produced the expected result: £_= .150. The low value indicated that the Androgynous and Masculine subjects could not be said to be reporting different levels of cognitive involvement from their fathers. In addition, because the earlier null hypothesis that the four sex role groups were not different in their reports of paternal cognitive involve- ment had been rejected, it can be said that while the Androgynous and Masculine women do not differ from one another, they were dif- ferent from the Feminine and Undifferentiated categories, in the validation group. In contrast, the null hypothesis that the four sex role categories would not differ in their reports of paternal cognitive involvement had not been rejected for the cross-validation group. Therefore, it seemed likely that the Androgynous and Masculine group means would not be significantly different. The £_value of .898 confirmed that the null hypothesis about the Androgynous and Masculine subjects should not be rejected, but unlike the validation group, it is not possible to say that the Androgynous and Masculine groups were different from the Feminine and Undifferentiated groups in the cross-validation sample. Sex Role and Parents' Warmth-Versus-Rejection The hypotheses which follow concern differences among the four sex role categories in the women's reports of their parents' warmth-versus-rejection on the PBF. 113 Sex Role and Mother's Warmthdversus-Rejection The first null hypothesis concerned the women's sex role and the level of warmth-versus-rejection they reported for their mothers: HO: N3 difference will exist among the four groups of subjects established by scores on the BSRI in the scores they ascribe to their mothers on the PBF's Warmth-versus-Rejection Factor. In its symbolic form: H0: WVRMA = WVRMF = WVRMM = WVRM U When the analysis of variance (Table 5.8) was conducted, significant values were obtained for both the validation sample (F - 3.02, p‘<.05) and the cross-validation sample (F - 2.74, p <.05). The results indi- cate that differences among the four sex role categories do exist in their reports of mother's warmth-versus-rejection. TABLE 5.8 ANALYSIS OF VARIANCE OF THE PARENT BEHAVIOR FORM'S WARMTH—VERSUS- REJECTION FACTOR SCORES FOR MOTHER FOR THE FOUR BEM SEX ROLE INVENTORY CATEGORIES, VALIDATION AND CROSS-VALIDATION GROUPS SAMPLE GROUP SOURCE df SS MS F VALIDATION GROUP Sex Role Category 3 3484.90 1161.63 3.02* Within-groups 88 33814.76 384.26 Total 91 37299.65 CROSS-VALIDATION GROUP Sex Role Category 3 2163.97 721.32 2.74* Within-groups 87 22884.60 263.04 Total 90 25048.57 *p <.05 114 The first of the planned contrasts which was tested by the £_statistic concerned the following alternate hypothesis: Hla' Subjects classified as Androgynous by their scores on the BSRI will report higher levels of the Warmth-versus-Rejection Factor on the PBF for their mothers than subjects classified as Feminine, Masculine, or Undifferentiated. Stated symbolically: Hla: WVRMA > WVRMF, WVRMM, WVRMU In the validation group the value of £_which was obtained was 2.655, which is significant (p‘<.05, one-tailed). A significant value was also obtained in the cross-validation group, t_- 2.396 (p‘<.05, one— tailed). Therefore, the Androgynous category in both sample groups did report higher levels of maternal warmth-versus-rejection than the Feminine, Masculine, or Undifferentiated categories. The second of the alternate hypotheses about sex role cate- gory and mother's warmth-versus-rejection was: Hlb: Subjects classified as Undifferentiated by their scores on the BSRI will report lower levels of the Warmth-versus- Rejection Factor on the PBF for their mothers than subjects classified as Andrggynous, Feminine, or Masculine. In symbolic form: Hlb‘ WVRMu (mm W» "VRMM In neither the validation nor the cross-validation groups was the.§ test of this particular contrast significant. The value obtained for the validation sample was £_= 1.461, and the value in the cross- validation sample was t_= 1.631, indicating that the Undifferentiated 115 women did not report significantly lower levels of maternal warmth- versus-rejection than the other three sex role categories. Sex Role and Father's Warmthr~versuséRejection The null hypothesis which was tested by one-way analysis of variance was: HO: N2_difference will exist among the four groups of subjects established by scores on the BSRI in the scores they ascribe to their fathers on the PBF's Warmth-versus-Rejection Factor. Stating the hypothesis symbolically, it is: Ho: WVRFA = WVRF F I WVRFM = WVRF U The F value obtained through the analysis of variance (Table 5.9) for the validation group was not significant (F - 1.46). Therefore, the null hypothesis was not rejected and the planned contrasts derived from the alternate hypotheses which follow were not carried out for the validation group. However, the analysis of variance in the cross- validation group resulted in F = 3.86, which was significant (p< .05), indicating differences among the four sex role categories in the levels of paternal warmth-versus-rejection they reported. The first of the alternate hypotheses, tested by the 5 statistic for the cross-validation group alone, was: Hla: Subjects classified as Andrpgynous by their scores on the BSRI will report higher levels of the Warmth-versus-Rejection Factor on the PBF for their fathers than subjects classified as Feminine, Masculine, or Undifferentiated. Stated symbolically: H1. 3. WVRFA>WVRFF, WVRFM, WVRFU 116 TABLE 5.9 ANALYSIS OF VARIANCE OF THE PARENT BEHAVIOR FORM'S WARMTH- VERSUS-REJECTION FACTOR SCORES FOR FATHER FOR THE FOUR BEM SEX ROLE INVENTORY CATEGORIES, VALIDATION AND CROSS-VALIDATION GROUPS SAMPLE GROUP SOURCE df SS MS F VALIDATION GROUP Sex Role Category 3 2116.97 705.66 1.46 Within-groups 89 43099.16 484.26 Total 92 45216.13 CROSS-VALIDATION GROUP Sex Role Category 3 3710.29 1236.76 3.86* Within-groups 86 27576.21 320.65 Total 89 31286.50 *p < .05 The alternate hypothesis about Androgynous subjects and paternal warmth-versus-rejection was not supported by the results of the“; test in the cross-validation group (2': 1.714). The second alternate hypothesis--again, tested only for the cross-validation group--was: Hlb: Subjects classified as Undifferentiated by their scores on the BSRI will report lower levels of the Warmth-versus- Rejection Factor on the PBF for their fathers than subjects classified as Androgynous, Feminine, or Masculine. In symbolic form: 117 The hypothesis stated above also failed to receive support when the planned contrast was tested (5.: .512) for the cross-validation group. The hypotheses about sex role and paternal warmth-versus- rejection were not supported in either of the sample groups. There were not significant differences among the four sex role categories in the validation group. Differences in the level of paternal warmth-versus-rejection among the four sex role categories were found in the cross-validation group, but not in the predicted direc- tions. A summary of the results of all the analyses of variance can be found in Appendix H, in addition to a table containing the group means for the four sex role categories for the analysis of variance and planned comparions reported in the previous sections. Subjects' Sex Roles and Perceived Parental Sex Roles The relationship between the subjects' reports of their own sex role behaviors and those of each of their parents was also studied. As stated earlier, each woman completed the BSRI for herself, her mother, and her father. To test the hypotheses about these data, which were nominal in nature, the Chi-square statistic was chosen. The results of these analyses will be reported separately for the women's mothers and fathers in the following sections. Subjects' Sex Roles and Perceptions of Mothers' Sex Roles A general null hypothesis was formulated about the subjects' self-reports and their reports about their mothers on the BSRI: 118 HO: No relationship will exist between the subjects' (sex role) categories scored on the BSRI and their mothers' (sex role) categories scored on the BSRI according to the subjects' reports. A 4 x 4 contingency table (Chi-square) was computed to test the null hypothesis (Table 5.10). In the validation group, x2 - 38.42 (p <.05, 9 d.f.), and in the cross-validation group, x2 - 31.33 (p <.05, 9 d.f.), indicating that in both samples the subjects' reports about their mothers' behavior on the BSRI was related to their reports about themselves on the BSRI. TABLE 5.10 THE 4 x 4 CONTINGENCY TABLES (FREQUENCIES) FOR MOTHERS' SEX ROLE CATEGORIES BY COLLEGE WOMEN'S SEX ROLE CATEGORIES, VALIDATION AND CROSS- VALIDATION GROUPS MOTHERS .A F M u VALIDATION GROUP COLLEGE WOMEN A 16 6 5 1 F 2 9 2 6 M 2 2 7 s U 2 a 9 12 x2 - 38.42, p <.05 CROSS-VALIDATION GROUP COLLEGE WOMEN A 12 7 3 2 F 7 8 a 3 M 6 3 5 6 U 1 2 7 16 x2 - 31.33, p'<.05 NOTE: A - Androgynous, F a Feminine, M-II Masculine, and U - Undifferentiated. 119 In order to test the specific alternate hypotheses about the relationship between the subjects' reports for themselves and for their mothers, a series of 2 x 2 contingency tables had to be con- structed by collapsing categories in sequence, and re—computing the Chinsquare value for each table so constructed. Because the expected cell frequencies sometimes fell below five, the Yates correction for continuity was applied. All of the Chi-square values reported for 2 x 2 contingency tables (d.f. = l) are thus corrected Chi-square figures. The results of this process have been summarized in Table 5.11 and will be discussed, one hypothesis at a time, in the following section. The first of the alternate hypotheses which suggested a relationship between subjects' sex roles and mothers’ sex roles was: Hla: Subjects classified as Androgynous by their scores on the BSRI will report mothers classified as Androgynous on the BSRI more frequently than subjects classified as Feminine, Masculine, or Undifferentiated. The hypothesis received support in both the validation and crossé validation groups. For the validation group the corrected X2 - 21.03 (p <.05), and for the cross-validation group, x2 = 6.19 (p‘<.05). Androgynous women did report Androgynous mothers more frequently than the other three sex role groups. The second alternate hypothesis was: Hlb: Subjects classified as Feminine by their scores on the BSRI will report mothers classified as Feminine on the BSRI mgr; frequently than subjects classified as Androgynous, Masculine, or Undifferentiated. 120 .poumwucmumumwvcs u 2 wow .mCNHSUmmz n z .mCHGHEmm u m .msoczwouv:< u < "meoz mm=Hm> mumscmlfisu pmuomuuoos me. vs no. va HoooHflNX WC. "Nx @moN "NX mHo© "Nx mm NH z.s.< mm «H =.m.< mm NN a.z.< cm «H =.z.s zmzos oN 0N a mN m z «N m a NN NH 4 momqaou «gnome onsummomo mods mc.va mc.va co m uNx NN N uNx NN.o uNx mo.NNuNx Nm NN z.s.< mm 0N =.m.< mm NH =.:.< on o =.:.m zmzoz mN NN a m N z oN m A NN 0N < momguoo «macaw one z.m.< : =.m.< : =.z.< s =.::m < mmmmhoz mmsomu ZOHHImeMU 92¢ ZOHH .Vmoezm>zH maom xmm 2mm mzH zo mmHmOOmHa mumsumlwnu Ovuoouuou « .pmumHuCONOWMNOCD u D van .ocHasommz mo.va on.~ umx «4mm.c an mH. umx on. umx S 2 z...H.< Hm NH =.m.< mm «H 2.2;. 3 NH 2.2...H 2883 «H HH : 8 o 2 S N ."H NH 6 < 5858 «macaw onsnmmomo 8J6 moJa om.m an em.H an on. uNx mm.a uNx Nm HH zéé Nm 8 =.m.< mm NH =.:.< am a mans zmzos mH NH H. S N z NH 0 .H 3H NH < momHHoo «mDOMU ZOHH z.m.< a =.z.< N =.m.< z 2.2.m < mmmmeImmomU 92¢ onH .MMOHZM>ZH MAOM xmm 2mm m:& 20 mmHMOUmHSEF, SEM, SEU * Confirmed Confirmed Hlb: SEU SEF Confirmed -- H0: SEMA= SEMF= SEMM= SEMU * Reject H0 Reject H0 H0: SEFA= SEFF= SEFM= SEFU Reject H0 Failed to reject... H13: SEFA >SEFF, SEFM, SEFU confirmEd -- Hlb: SEFU COGFF, COGFU Confirmed -- Hlb: COGFUwWVRMP WVRMM, WVRMU * Confirmed Confirmed Hlb: WVRMUWVRFF, WVRFM, WVRFU -- -- . < -_ _- Hlb. WVRFU WVRFA, WVRFF, WVRFM H : No relationship between SRS * Reject H Reject H 0 0 and SRM Hla: Androgynous Ss report * Confirmed Confirmed Androgynous mothers Hlb: Feminine 83 report Confirmed -- Feminine mothers H1 : Masculine 83 report -- Confirmed c Masculine mothers Hld: Undifferentiated 83 report * Confirmed Confirmed Undifferentiated mothers H : No relationship between SRS Reject H0 Failed to and SRF reject... Hla: Androgynous 83 report Confirmed -- Androgynous fathers Hlb: Feminine 88 report -- -- Masculine fathers ch: Masculine Ss report -- --** Feminine fathers Hld: Undifferentiated Ss report Confirmed -- Undifferentiated fathers *Significant in both sample groups. **Significant in the direction Opposite of that predicted. NOTE: A - Androgynous, F = Feminine, M = Masculine, I_J_ - Undif- ferentiated; §§_- Self-Esteem, Self; SEM,= Self-Esteem, Mother; §§§.' Self-Esteem, Father; COGM = Cognitive Involvement, Mother; COGF - Cognitive Involvement, Father; WVRM - Warmth-vs.-Rejection, Mbther; WVRF a Warmth-vs.-Rejection, Father; §§§ a Sex Role, Self; SRM - Sex Role, Mother; and §_R_1: 8 Sex Role, Father. 144 BSRI and the PBF were in an acceptable range of reliability prior to the removal of any Of their items. Two other benefits were derived from the cross—validation model. The first was the assurance that the likelihood of making a Type I error (i.e., of falsely rejecting the null hypothesis) did not become inordinately large because of conducting a large number Of tests on the same data pool. Second, it became easier to identify just which variables were unstable over the two sample groups, and thus would be more likely to be unstable in the population, if not accounted for by the unreliability of instruments. Analysis of the Data Three different statistical techniques were used to test the hypotheses. They were chosen for their applicability to the type of variable (discrete or continuous) and for their ability to answer the questions contained in the hypotheses. Analysis of variance and t tests of g_priori contrasts were used to test those hypotheses with sex role category as the dependent variable, and variables such as self-esteem as independent variables. Pearson product-moment corre- lation was used to test the hypotheses about relationships between continuous variables. For the hypotheses about relationships between the women's sex role categories and their parents' sex role categories, Chi-square analysis was employed. Conclusions The following conclusions (1 - 9) have been derived from the results which were significant (p < .0025) in both the validation (V) and cross-validation (CV) groups. 145 Reports of self-esteem did differ according to the sex role categories to which the college women were assigned on the basis of their BSRI scores. a. Androgynous college women seemed to have higher levels of self-esteem than those women who were sex-typed, or who reported restricted masculinity and femininity. Another way of stating the conclusion is that college women reporting high levels of self-esteem also reported high levels of masculine characteristics and high levels of feminine characteristics. b. Undifferentiated college women reported 123;; levels of self—esteem than those who were sex-typed or who reported masculinity and femininity in proportionately high degrees. Alternately, those women who reported low levels of self- esteem also reported lower levels of masculine characteris- tics or feminine characteristics. The women in the four sex role categories did differ from one another in the amount of self-esteem they reported for mothers. The direction of the differences was not consistent over the two sample groups, however. There was a positive relationship of moderate strength (rV = .23, rCV = .24) between the women's self-esteem and the self-esteem they reported for their mothers. There was a positive relationship Of moderate strength (rV = .28, rCV = .21) between the women's self-esteem and the self-esteem they reported for their fathers. 146 There was a positive relationship of low to moderate strength (rV = .19, r .35) between the women's self-esteem and the CV= cognitive involvement they perceived for their mothers. There was a positive relationship of moderate strength (rv = .31, rCV = .33) between the women's self—esteem and the warmth-versus- rejection they perceived for their mothers. Reports of mothers' warmth-versus-rejgction did differ according to the sex role category to which the women were assigned. a. Androgynous women seemed to have perceived their mothers as warmer and less rejecting than women who were classified as Feminine, Masculine, or Undifferentiated. Androgynous college women reported Androgynous mothers in signi- ficantly higher proportions than what could be expected by chance. Undifferentiated college women reported Undifferentiated mothers in significantly higher proportions than what could be expected by chance. The following conclusions (10 - 11) have been derived from the results Of the tests of hypotheses which were significant in only one of the sample groups. 10. 11. Of the 10 hypotheses which were rejected (nulls) or confirmed (alternates) in the validation group, but not in the cross- validation group, seven involved some measure Of the women's fathers. Of the 10 hypotheses which were rejected or confirmed in the cross-validation group, but not in the validation group, half involved some measure of their mothers and half involved some measure of their fathers. 1463 The following conclusions (12 - 19) are based on those hypotheses which failed to be rejected or were not confirmed in either the validation or cross—validation groups. 12. Undifferentiated women did not report significantly lower levels of self-esteem for their fathers than the Androgynous, Feminine, or Masculine women. 13. Undifferentiated women did not report significantly lower levels of warmth-versus-rejection from their mothers than the other three sex role categories. 14. Androgypous women did not report significantly higher levels of warmth-versus-rejection from their fathers than the other three sex role categories. 147 15. Undifferentiated women did not report significantly lower levels of warmth-versus rejection from their fathers than Androgynous, Feminine, or Masculine women. 16. Masculine women did not report Masculine mothers in propor- tions greater than that expected by chance. 17. Feminine women did not report Masculine fathers in propor- tions greater than that expected by chance. 18. Masculine women did not report Feminine fathers in propor— tions greater than that expected by chance. In the cross— validation group not only did they fail to report father's sex role in the predicted direction, but they also reported significantly fewer Feminine fathers than that expected by chance. 19. The hypotheses suggesting a counterbalance between father's sex role and daughter's sex role failed to be supported. A better argument could be made for father—daughter similarity than for counterbalance. Discussion of Results In the following sections, some Of the findings will be discussed in light of earlier findings. Relevant theory will be re-introduced, and some alternative ideas will be presented. Sex Role Identity and Self-Esteem As predicted by the results of previous studies by Bem (1977) and Spence, Helmreich, and Stapp (1975), a relationship between the the women's sex role identities and their self-esteem was found in 148 the current study. As Bem, and Spence and associates have found, Androgynous women had the highest level Of self-esteem and Undif- ferentiated women had the lowest level of self-esteem. Spence, Helmreich, and Stapp (1975) reported a clear rank order for their women subjects, which meant that in addition to the distinction between Androgynous and Undifferentiated women, Masculine women reported significantly higher levels of self-esteem than did Feminine women. Bem (1977) did not find a rank ordering of the same magnitude between Masculine and Feminine women, although her Masculine group mean on self-esteem was slightly larger than her Feminine group mean. In addition, people (e.g., Kelly & Worell, 1977) do seem to believe that higher self-esteem is logically related to masculine characteris- tics because American culture values that which is masculine more than that which is feminine. Extending this thought further, it may be said that for women, even the supposed deviation of having cross—sex characteristics would be offset by the cultural valuation of masculine characteristics. However, the results of the present study do not make the issue of Masculine-typed and Feminine-typed women's self— esteem any clearer. In the validation sample the Masculine women did report higher self-esteem than the Feminine women, but in the cross- validation sample this distinction was not found between the two subgroups. Women who are significantly sex-typed may be so for dif— ferent enough reasons that predicting consistent differences between the two sex-typed groups would be extremely difficult. The phenomenon of sex-typing may be capricious as it is currently measured, or other factors may be contributing which have not been adequately incorporated into the current study or previous studies. 149 Relationship between Sex Role and Self-Esteem as Affected by Other Variables Whether or not the distinction between Masculine-typed and Feminine-typed women were found in both sample groups, the question concerning the nature of the relationship between sex role identity and self-esteem still had to be addressed. In Chapter II, it was indicated that the relationship between sex role identity and self-esteem could be reciprocal (i.e., affect one another with alternating impact) or could be symmetrical. Spence, Helmreich, and Stapp (1975) said: "The correlations between positive relationship with self-esteem suggest that the two factors may function in an additive way to determine an individual's self-concept and behaviors" (p. 35). Knowing that there was likely to be a relationship between self-esteem and sex role iden- tity, it was also considered important in the current study to inves- tigate the possibility of other variables which would affect self- esteem and sex role identity, implying that the relationship between the two was, in part, due to being "caused" by the others. A few consistent results were obtained of a variable's being related to both self-esteem and sex role identity in this study. Mother's self-esteem did differentiate among the sex role groups, and was positively related to the daughter's self-esteem in both the validation and cross-validation groups. Mother's warmth-versus- rejection did differentiate between the sex role categories, with Androgynous women reporting higher levels than the other three categories, and maternal warmth-versus rejection was positively related to daughter's 150 self-esteem in both sample groups. Mother is more influential-—or at least her impact is clearer to the daughter--in the relationship between sex role and self-esteem than is father. In the current sample, the greater influence may be due to the fact that the women spend more time with their mothers when growing up, as they reported in the FIQ. Or, mother's greater influence may be due to same—sex similarity, as posited by social learning theory and cognitive-developmental theory. Or, mother may have had a greater influence because of the daughter's preference for mother's emotional support and her having received more emotional support from mother than from father (FIQ results). The effect may also be due to a confluence of these and other variables as well. Subjects' Self-Esteem and Perceptions of Parents' Self—Esteem Father's and mother's self-esteem were both positively related to daughter's self-esteem. In one sense daughters learn their self- esteem from their parents. However, this result may also be due to leniency error, the tendency of the daughter to rate those closely related to her higher on any dimension. Following Kohlberg's (1966) thought, though, that the driving force behind much of individual development is competence and a positive self image, it can also be assumed that the daughter's self-esteem is enhanced as her perceptions of how to derive self-esteem become more sophisticated, and affected by the amount of constructive information she receives. High levels of parental self-esteem could be a rich source of constructive information for the daughter about how to conduct herself to achieve the same. 151 Other Parental Variables The variables of parents' cognitive involvement and warmth- versus-rejection (Kelly & Worell, 1976) were not related to the women's sex role categories consistently across the validation and cross-validation groups. Significant results in the validation group not confirmed by the cross-validation groups are more difficult to interpret than the reverse situation because theoretically the PBF and BSRI have been strengthened for the cross-validation group. If the results in the validation groups were significant, using instru- ments which supposedly introduce 2253 error than in the cross-validation group, it seemed logical that the tests would be significant using instruments which supposedly introduce less error. Other factors must be called upon to account for the inconsistency. For example, the PBF and BSRI may not have been improved sufficiently through the removal of items to reduce the amount Of error introduced in the cross-validation analysis. Another of the possible explanations is that the variables under consideration (e.g., daughter's sex role, parents' cognitive involvement, and parents' warmth-versus-rejection) may not have been measured validly, or may represent unstable features of the population. Were it not for the fact that both the BSRI and PBF'were involved in results which were significant in the validation, group, but not the cross-validation group, it would have been easier to attribute significant results in the cross-validation group, but not the validation group, to the improvement of scale reliabilities through the elimination of a few items. However, it is not possible 152 make the assumption that the instruments were improved; therefore, the questions about validity and variables' stability are introduced again. In short, the results of the current study did not confirm Kelly's and Worell's findings, and several possibilities exist as explanations, for example: 1. The BSRI used in the current study and the PRF ANDRO used by Kelly and Worell may measure different sex role variables. 2. Using the PBF factors as opposed to individual PBF scales may have changed the nature of the variables actually measured (although it should have increased reliability by increasing the length of the scale). 3. Unlike Kelly's and Worell's subjects, the women in the current study were not asked to describe their parents as they were when the women were 16 years old. Androgynous-Undifferentiated Differences The Androgynous category Of women and the Undifferentiated category of women differed from one another most consistently, as expected. Self-esteem, mother's self-esteem, and mother's warmth- versus-rejection were the three variables in which the Androgynous group was higher than the Undifferentiated group in both samples (see Appendix I). Father's self—esteem, mother's cognitive involvement and father's cognitive involvement were the three variables for which Androgynous women reported higher levels than the Undifferentiated women in one or the other Of the two sample groups (see Appendix I). 153 No other subset of sex role categories evidenced such con- sistent differences, leading to the speculation that the Androgynous and Undifferentiated categories include a broad and stable cross- section of variables, while the Masculine and Feminine categories are indicative Of a narrower and less stable set of variables. Androgynous and Undifferentiated women may be more easily distinguished by parental variables, because the Androgynous person's "enhanced" functioning would be more likely if she had positive, con- structive experiences with her parents, and more likely to be asso- ciated with parents because the culture does not teach androgyny for women. Undifferentiated women's constricted functioning might be based on deprivation, especially with parents, and because of the deprivation these women may not be capable of incorporating more posi- tive experiences with people other than their parents. There is a great deal of information from the culture about being a feminine-typed woman. It is more difficult to know what is associated with being feminine-typed, however, because that individual could be drawing from her parents or may be drawing from the culture and her peers, and repressing or suppressing her instrumentality to maintain close relationships. Masculine-typed women occasionally reported low parental warmth in the current sample, leading to the idea that they might be suppressing the expressive domain because of psychological pain associated with it, and turning to the masculine domain for their mastery and self~worth. It becomes quite clear that studies need to be done on multiple variables which lead to sex role identity as well as on personality variables correlated with sex role identity. 154 Measuring Sex Role Identity Although the orthogonal measures of masculinity and femininity have added an important dimension to the study of sex role identity, some of the phenomena associated with sex role identity may be obscured for now because the categories being used are still gross measures of extremely complex variables. If there were a way to incorporate Bem's early notion of scoring the balance between masculinity and femininity, and Spence's and associates' idea about the comparative degree of masculinity and femininity, it might be possible to use samller categories, and make finer distinctions among subjects. Sex Role and Perceptions of Parents' Sex Roles Because the subjects did report similarity to their mothers, modeling, identification and cognitive-development processes can all be argued. The impact of the tendency for the women to also report similarity to their fathers, however, is to highlight cognitive- developmental theory because both the social learning construct of modeling and the construct of identification emphasize same-sex factors. With cognitive-developmental theory, competence, mastery and self-esteem are the goals, and the basic Premise is that the daughter identifies with the person or persons who help her to achieve those goals. While Kohlberg (1966) did indicate a belief in the impact Of the same-sex parent, a combination of his theory and the dualistic view of masculinity and femininity leads to the idea that mother or or father or both could be helpful sources of information and encourage- ment of daughter's competence and self-esteem. 155 Literature Introduced Since the Development of the Current Study The research and writings on the subject of sex role identity, especially the dualistic concept of masculinity and femininity, have been continuing to increase. Some of these works, relevant to the current study, will be reviewed briefly because future research should incorporate these works as well. The most important and most meticulous of these new studies was reported in a book by Spence and Helmreich (1977). They have added many dimensions in their latest work, for example, different age and socioeconomic groups, subjects' reports of parents' sex role identities,‘ and parents' reports of their own sex role identities. They established "couple types" for sets of parents of subjects, for all the possible combinations of mother-father sex role identities, e.g., Androgynous— Androgynous, Androgynous—Feminine. They then investigated correlates of these couple types for their college student subjects. They have constructed another instrument with which to measure sex role identity correlates, the Work and Family Orientation Questionnaire (WOFO). They administered yet another instrument, the Parental Attitudes Questionnaire, and in factor analysis found seven scales significant for both males and females: Father positivity, Mother Positivity, Father Democracy, Mother Democracy, Rule Enforcement, Family Protec- tiveness, and Sex-Role Enforcement. Two factors were found to be specific to women, Female Family Harmony and Mother Supportiveness, and Female Standards. 156 The wealth Of their findings cannot be adequately summarized here. Spence and Helmreich continued to find support for dualistic notions of masculinity and femininity, for the relationship between parental behaviors and sex role identity, for the importance of the children's perceptions of their parents, and for the importance of parental identities and other environmental factors. As for its relevance to the current study, their book has only one major limitation: its publication occurred after the implementation of this study. A study presenting results that indicated the "flexibility and adjustment were generally associated with masculinity rather than androgyny for both males and females" (p. 298) was conducted by Jones, Chernovetz, and Hansson (1978). These researchers inves- tigated sex role identity as measured by the BSRI, and a number of other variables: attitudes toward women's liberation, locus of control, neurosis, self-esteem, problems with alcohol, creativity, political awareness, confidence in one's ability, helplessness, and sexual maturity. However, although they included many possible cor— relates of sex role identity and have presented a fresh challenge, they may have further clouded the issues by returning to Bem's original scoring for the BSRI which produces three categories: Androgynous, Masculine, and Feminine. They used the earlier scoring procedure despite acknowledging in their review Of the literature that Bem had acceded to the criticism of Spence and others, and began using the four—category scoring procedure. It would be interesting and valuable to re-analyze the data Jones, Chernovetz, and Hansson obtained 157 with the addition Of the median-split technique for scoring the BSRI. Their inclusion of a large number of possible correlates is important. Kelly, Furman, and Young (1978) completed the first study of interscale comparability for the BSRI (Bem, 1974), PAQ (Spence, Helmreich & Stapp, 1975), PRF ANDRO (Berzins, Welling & Wetter, 1978), and the ACL Masculinity and Femininity scales (Heilbrun, 1976). Their basic conclusion was that the four inventories may be measuring quite different subgroups and that it is inappropriate to assume cross-sex comparability.3 They also conclude that "when sex role styles are dichotomized into broad typological quadrants, as is the current practice in sex role research, substantial predictive utility may be lost" (p. 1574). A study addressing one aspect of the present study was con- ducted by Doherty and Schmidt (1978), investigating the relation- ship between sex role identity and self-esteem among college women. Doherty and Schmidt used the BSRI, which they scored in the revised manner, and the Tennessee Self-Concept Inventory. All of their significant differences occurred between the Androgynous group and one of the other sex role categories, and, "androgynous women scored higher than undifferentiated women on six self-concept dimensions, higher than masculine women on five dimensions, and higher 3This finding adds credence to the idea that differences between PBF-and-sex-role results in the current study and in Kelly's and Worell's (1976) study are due to the fact that the BSRI was used in the current study, and the PRF ANDRO was used in their study, to measure sex role identity. 158 than feminine women on two dimensions" (p. 496). They also found a trend for feminine women's scoring higher than masculine and undifferentiated women, although the differences were not significant. In a study of subjects older (between 40 and 50 years) than those typically studied, O'Connor, Mann, and Bardwick (1978) found support for Spence's and associates' work, with androgynous self- descriptions predicting the highest levels of self-esteem. However, they are careful to say that theirs are middle-class subjects, as are most college students. Lastly, a comprehensive text on sex roles and women has been written by Frieze, Parsons, Johnson, Ruble, and Zellman (1978) which includes review and synthesis of a large amount of the literature relevant to the psychology of women and sex role identity. One of the areas they have covered is Kohlberg's cognitive-developmental theory. In a chapter for which Parsons was the primary author, it is stated that: The evidence is somewhat equivocal for Kohlberg's prediction that one's sex and sex role valuing would emerge naturally due to the egocentric thinking of preschool children. . . . While most girls express positive feelings toward being female and toward the female role, there is clear evidence of ambivalence. . . . What it may suggest is that cultural processes interact with cognitive- developmental processes to influence the acquisition of any behavior. (p. 132) Parsons also indicates the importance of this interaction between cognition and environment during adolescence, when, according to Piaget's theory, peOple are making the change from concrete operations to formal reasoning. Adolescent girls may be likely to try less rigid role behavior. According to Parsons, having decided egocentri- cally that being a girl is best, but hearing from the culture that 159 being a boy is best, the girl feels ambivalence which leads to her trying as many cross-sex roles as she can while retaining same—sex roles. Another factor, then, is the type of environment in which she exists. The argument can be made for the current study that the college environment may not include as much pressure for confor— mity as some other environments for late adolescents. Suggestions for Future Research Because of the complexity Of the relationships between sex role identity and other variables it would be helpful to be able to say more about the degree of impact each of the variables does have, and how they might be rank ordered. One way of accomplishing the task would be to use a multivariate technique for analysis, especially knowing from the current study which areas are unclear. For example, multiple regression analysis (with dummy variables, because sex role as it is now measured is discrete), or discriminant function analysis could be used. These types Of analysis might allow researchers to say, for example, what amount of the variance in sex role category is accounted for by self-esteem, or perceptions of parental sex role identity. If the data from the present study were to be analyzed again, using one Of the multivariate techniques, the inconsistent results across the two sample groups might be clarified, or new ques- tions could be developed. Sex role identity should be studied with samples having dif- ferent ages, socioeconomic status, and ethnic backgrounds than white, middle-class college students. Spence and Helmreich (1977) have encouraged this step in their book. They have studied high school 160 students with broad backgrounds, and other cultural subgroups, and speak to the particular difficulty in reaching non-white groups. It would also be interesting to compare the findings on college student samples with findings on late adolescents and young adults who are not attending college. Not only could their perceptions be compared, but also their environments could be contrasted to test for the interaction between their perceptions and environments. Careful logic should be applied to the interpretation Of the types of relationships which are being studied, so that variables are not assumed to be asymmetrically (one causes the other) related when they may, in fact, be reciprocal, or symmetrically related. It would be profitable to compare the PBF (Worell & Worell, Note 1) to the Parental Attitudes Questionnaire developed by Spence and Helmreich (1978) to evaluate which is the stronger measure, or which elements from each could best be combined for future studies on parental antecedents of sex role identity. Studies Of large samples, such as that by Bem and Lenney (1976), might provide norm groups for the sex role categories rather than using the median-split technique on a sample-by-sample basis. In addition, finer distinctions should be made than what the four cate- gories currently in use provide. One way of conceptualizing how these finer distinctions might be made is a circumplex, which would include a High-Low dimension, and a Feminine4Masculine dimension. The circumplex could theoretically be divided into smaller and smaller "slices of pie" and concentric circles within the circumplex could also represent degree (see Figure 6.1). 161 H15! H A I k '7 2‘ $2 _ V. 2 '* cc— E E u.) 1 .7 Low Fig. 6.1 A Circumplex Model for Sex Role Identity 162 Combining the findings about norm groups with the principle Of a circumplex might show the need to divide the circumplex at a point other than midpoint (for example, see Figure 6.2) in order to indicate the proportions which are predicted for the population. H‘GH FeHlMlNlry A Sex Role Identity Circumplex, Adjusted Fig. 6.2. for Population Proportions However conceptualized, it is clear that as the understanding of sex role identity becomes more sophisticated, the models will need to be made more complicated, too. From the instability of some results in the present study, it can be concluded that a number of phenomena are lost or masked by the use of the four large categories for sex role identity. APPENDICES APPENDIX A ACADEMIC MAJORS OF SUBJECTS APPENDIX A ACADEMIC MAJORS OF SUBJECTS Agriculture and Natural Resources 1 — Animal Husbandry 4 - Animal Tech 1 - Fisheries & Wildlife 1 - Food Systems Economics & Management 1 - Packaging 1 - Park Design 1 - Recreation Arts and Letters HNwl-‘NHNNH I Classical Studies English French German History Humanities Music Therapy Pre-Law Theatre on C3 :1“ :3 (‘D U) U) NHU‘IQ I N U.) I Accounting Business Economics Hotel, Restaurant and Institutional Management Marketing Travel and Tourism Communication Arts and Sciences pas~uao~haxa I Advertising Advertising/German Audiology & Speech Science Communications Journalism Telecommunication/Social Science Pre-Law Education ~41~1d1a1a1»1d6H Hmamm. aommH. mamam. NmmNN. NmmNN. NNNHN. mamamsmH mass: 66: Doc mmoa mHNam. maHmN.u Namam. aHNmH. mmama. mmaHN. mxHHsHHsO aammm. mNaNH. mNNam. NmmHN. aNaNN. HmNmH. wHmHHHso mmmmm. maNHm. NmNNm. aNmHm. ommma. Nmoam. Ammaoa mammm. mHmNm. mHaNm. Nmmom. mmNmN. Hmmmm. sums Nmmmm. aommN. maoam. aNmaN. momma. HmmNH. cmxoamummom mwcfiammm Nammm. oaomm. mammm. mamNN. mammm. mammm. Dun; asuooa as Human mommm. amoma. mmaam. momma. «mama. ommNm. OumaonmmaaoO mHNmm. amaaa. ammam. mmmmm. momma. aamma. mcHscmuaOosas mumsuo mo ammmm. mmama. maNNm. HNaNm. moama. aONmm. mamas 6:6 6O 6>HmHmamm NHmmm. mNNam. Nommm. NNaNN. momsa. omamm. OHummumaaam Nmamm. HNHmH. mammm. aHsam. mNHNN. aaNaH. oaHaHema aamam. NNmmm. mammm. omama. maamm. omooN. Hmmoa maamm. mmoNN. omamm. mmomm. ammaa. aamao. OHmmOOOOmHm masmm. «Naaa. ammam. ommma. ammma. mamas. mumconommmm omNam. momma. Naaam. momNH. smoom. oNaaH. mam NNon. mNmam. mmomm. amaHm. maamm. momma. Hamummso mammm. NmNam. NHmma. Honm. Nmmma. mmmaN. maHsHaHa nmemgmn .mmoo QMHMAMQ .mmou nmemqmn .mmoo zmaH aH H .onaHDHumaaou mNmoa. mmmHm. HmNNa. mmmam. mmamm. amaam. OHDAHH6=6H>HNGH NNHoa. moama. NaNHa. mmNmN. mNaNm. onmm. OmamOH 6 mm muo< Hamoa. NNHNm. msoNa. mmmam. mHaNm. aaaam. O>Hammnmm< Hamoa. HmNHN. mNoNa. mmNmN. omaam. mammm. samba m mama 6O maHHHHz NmOHa. amNHm. mNaNa. commH. aHoam. ammmH. maHHsuamz Hmmoa. NNHam. ooNNa. HmNHm. oamam. HaNom. OaaaHaoa amaoa. NoNNs. NoaNa. mmamm. NONmm. asoam. OsmHuHmmsmumHmm mmmoa. NmaHm. amHNa. NmHNm. oammm. Nmaom. aHHamO muonHumm amass amHHa. NNmNa. ammNa. mammm. momma. mamam. mamHO mama om maHHHHz mmNoa. NHNmN. amNHa. ommaN. mmHNm. maNNN. mOHuHHHmm aHsmOOmmaH mam mOOHa. mamas. HaNNa. NNOHm. ammmm. omHmN. HROHOmHaq< mNOHa. omoma. amaNa. NmNNm. Nmmam. mmon. Hammouom Nasoa. mammm. NoaHa. NmNmN. NNmam. Hmamm. aOHHmcomuma acouum mmaoa. NNHNN. mNaHa. mNCNN. maHNm. Naoma. 6>Hmummm< HmmHa. HmNmm. maaNa. moaaN. mmmmm. momma. OHOmHsu< mHNca. omHam. mmaHa. mHmaa. oHNNm. ooamm. mamaamammaH ammoa. amamm. mHNNa. ammNm. Naomm. oaoma. mmmHHmm :36 accwmon HmNHa. samNa. ONHNa. HaNmm. maomm. NmNms. OsmHHmAuHHOm amamHma .xaou amamHma .mmoo amamHma .mmoo smaH mH Hmaoa :NHH aH Hmaoa zmaH EH H .ona