Anda di halaman 1dari 11

Developmental Psychology 1991, Vol. 27, No.

3, 505-515

Copyright 1991 by the American Psychological Association, Inc. 0012-1649/91/$3.00

Assessing Sex Typing and Androgyny in Children: The Children's Sex Role Inventory
Janet P. Boldizar
University of Alabama at Birmingham
A measure of sex typing and androgyny in children was developed based on the Bern Sex Role Inventory (BSRI; Bern, 1974). The Children's Sex Role Inventory (CSRI) was correlated with the BSRI in an adult sample to establish the equivalence of the two inventories for use in developmental research spanning middle childhood and adulthood. Ss were 145 third, fourth, sixth, and seventh graders, 130 of whom were retested 1 year later. Reliability was established through high internal consistency of femininity and masculinity scales (average a = .81) and stable test-retest reliabilities averaging .60. Validity of the scales was evident in significant ( p< .01) gender differences on both scales and in confirmation of hypothesized relations between gender-role categories and measures of (a) sex-typed toy and activity preferences; (b) self-perceptions of global self-worth, scholastic competence, social acceptance, athletic competence, physical attractiveness, and behavioral conduct; and (c) cognitive performance. A short form of the CSRI was also developed.

Given the widespread interest in the development of sex typing in children, it is surprising that research in the field has relied on a relatively limited range of assessment measures. The formerly popular projective measures (e.g., the IT Scale for Children; Brown, 1956) have largely been displaced by the use of toy and activity preference measures as the primary indicators of children's sex typing (Beere, 1979; Downs & Langlois, 1988; Huston, 1983), perhaps because these measures are both objective and appropriate for use with preschoolers as well as with older children. Toy and activity preference measures have thus afforded opportunities to study the earliest manifestations of sex typing in children and to chart the development of children's gender-related behaviors and associated cognitions. By the time children reach the elementary school years, however, new options for studying their identification with gender roles become available. As children's self-concepts become less based on physical or behavioral attributes and more based on psychological attributes (Harter, 1983), it becomes possible to expand the study of sex typing in children to incorporate the self-perception of gender-related personality dispositions that are not relevant to the study of sex typing in younger children. Although the personality assessment approach has largely dominated the adult sex-typing and androgyny literature (e.g., Bern's, 1974, Sex Role Inventory [BSRI] and Spence, Helmreich, & Stapp's, 1974, Personal Attributes Questionnaire [PAQ]), it has received relatively little attention from developmental psychologists.
I would like to thank the principal, teachers, and students of the A. D. Henderson University School for their generous cooperation in the collection of these data. I am also indebted to Michelle Bengston, Rosemary Perez-Rossman, Rosemary Shaw, and Patti Smith for their assistance with various phases of this research, as well as to Ken Wilson for his helpful comments and support throughout the project. Correspondence concerning this article should be addressed to Janet P. Boldizar, Department of Psychology, 201 Campbell Hall, University of Alabama, Birmingham, Alabama 35294.
505

The goal of this study was to create a children's questionnaire that is conceptually equivalent to the BSRI (Bern, 1974), a widely used adult sex role measure that assesses femininity and masculinity as independent dimensions.1 Two comprehensive reviews of the sex role literature (Signorella & Jamison, 1986; Whitley, 1983) report that a majority of the studies included in these reviews used the BSRI to assess masculinity, femininity, and androgyny. Given the prevalent use of the BSRI in adult populations, a children's version of this inventory would be highly useful to developmental researchers interested in using the BSRI to examine changing influences of sex typing and androgyny from middle childhood through adulthood. Moreover, the development of the Children's Sex Role Inventory (CSRI) will add a new alternative for those researchers interested in examining sex typing in children from a multidimensional perspective, an approach that has been strongly advocated by many leading researchers in thefield(Downs & Langlois, 1988; Huston, 1983,1985). The theoretical assumptions guiding the development of the CSRI were consistent with those of the BSRI. It was assumed that (a) there exist heterogeneous clusters of personality attributes that have been culturally defined as more appropriate and desirable for one sex than for the other and (b) there are individual differences in the propensity to use these categories as ideal
1 An adaptation of the BSRI for adolescents, the Adolescent Sex Role Inventory (ASRI), was published by Thomas and Robinson (1981). However, this scale was not developed for use with elementary schoolaged children, and the reliability and validity of the ASRI were largely established through administration of the ASRI to undergraduate college students. The only reported administration of the ASRI to seventh and eighth graders established an average item correlation with the BSRI (r = .44), but no data were reported regarding the internal consistency of the masculinity and femininity subscales, test-retest reliability, or construct validity of the scale with the adolescent subjects. Thomas (1983) reported further psychometric properties of the scale, but these data have not, to my knowledge, been published.

506

JANET P. BOLDIZAR

standards of masculinity and femininity in the self-evaluation of personality and behavior. These assumptions are derived from Bern's (1981a, 1981b, 1985) gender schema theory, which proposes that sex-typed individuals (i.e., those who score above the median on gender-congruent attributes but below the median on gender-incongruent attributes) will differ from nonsex-typed individuals in their propensity to organize information along gender-based lines and in their subsequent motivation to enact culturally denned stereotypes of masculine and feminine behavior. The development of the CSRI was also based on the assumptions that, by the early elementary school years, children can discriminate between these culturally defined sex-role attributes and behaviors (Huston, 1983) and that this knowledge can serve to organize children's understandings about their own gender-related personality attributes and behaviors. Only one major attempt has been made by developmental psychologists to assess masculinity and femininity in children. Hall and Halberstadt's (1980) Children's Personal Attributes Questionnaire (CPAQ) was directly derived from Spence et al.'s (1974) PAQ, a measure also developed to assess masculinity and femininity as independent trait dimensions in adults. Unfortunately, in the nearly 10 years since the CPAQ was published, it has rarely been used in published empirical studies. Examination of the Social Sciences Citation Indexes (Institute for Scientific Information, Inc., 1981-1989) from 1981 to mid-1989 and review of recent major journals revealed onlyfivestudies that reported using the CPAQ as a measure of sex typing or androgyny in children (Alpert-Gillis & Connell, 1989; Hall & Halberstadt, 1981; Mitchell, Baker, & Jacklin, 1989; Silvern & Katz, 1986; Wilson & Cairns, 1988, Study 3). This is especially unusual in light of the fact that the CPAQ was published in a major developmental journal and has been cited in several reviews of the sex-typing literature (Huston, 1983, 1985; Signorella & Jamison, 1986; Taylor & Hall, 1982). Data from thefivepublished empirical studies that have used the CPAQ in the past decade and from the originally published scale point to possible explanations for the infrequency of its use. One potential problem is that the majority of these studies report that boys and girls do not differ as expected on the masculinity (M) and femininity (F) scales (i.e., boys do not score significantly higher than girls on masculinity, and girls do not score significantly higher than boys on femininity). The CPAQ failed to discriminate between boys and girls on the M scale in two studies (Hall & Halberstadt, 1980, Study 1; Wilson & Cairns, 1988, Study 3preadolescent group), and on the F scale in two studies (Mitchell et al., 1989; Silvern & Katz, 1986). (No data on sex differences in masculinity and femininity were reported by Alpert-Gillis & Connell, 1989.) Another possible problem involves relations between the M or F scales and sex-typed play preferences or self-perceptions. Silvern and Katz (1986) found that, for boys, there was a marginally significant negative correlation between the CPAQ M scale and masculine play preferences and that there was a marginally significant positive correlation between boys' self-reported same-sex similarity and the CPAQ F scale (ps < .06). Although advocates of multidimensional assessments of gender roles argue that these measures assess potentially independent domains (cf. Downs & Langlois, 1988; Spence, 1985), Taylor

and Hall (1982) suggested that these types of anomalous findings may call into question the construct validity of the CPAQ M and F scales. That is, in order to establish the convergent and divergent validity of the scales, masculine traits and behaviors should relate positively to the M scale but not to the F scale, and vice versa.2 Thus, a number of researchers may view these types of results as undermining the usefulness of the CPAQ as a measure of children's sex typing or gender role identification. Because it was apparent that great care went into the construction of the CPAQ, an alternative explanation was sought for the fact that the CPAQ had not produced the types of results that many developmental researchers had been led to expect from the adult and children's sex-typing literature. One possibility was found in differences in the criteria that were used to select the items for the PAQ (Spence et al., 1974) and the BSRI (Bern, 1974). Whereas the BSRI items were selected on the basis of their sex-typed social desirability (i.e., masculine items were judged more socially desirable for men than for women, and feminine items were judged more socially desirable for women than for men), both feminine and masculine items on the original PAQ were similar in their social desirability for both sexes, although feminine items were judged more typical of women than of men and masculine items were judged more typical of men than of women (Bern, 1974; Spence et al., 1974). These selection differences may contribute to the fact that the BSRI and PAQ have been found to differ in their factor structures, with the PAQ M and F scales forming two orthogonal factors representing instrumentality and expressiveness and the BSRI scales forming a multifactorial solution (Pedhazur & Tetenbaum, 1979). Moreover, it has been pointed out that some of the BSRI items do not share the uniformly high levels of social desirability that the PAQ items do (Pedhazur & Tetenbaum, 1979), quite possibly because items that are relatively more socially desirable for one sex than for the other are not necessarily also high on some absolute measure of social desirability. As Spence and Helmreich (1979) noted, "Although there is a fair amount ofoverlap between the PAQ and the BSRI, and a moderate correlation between parallel scales, there are enough differences between the two instruments to suggest that the type of results found with the BSRI would not necessarily be found with the PAQ" (p. 1036). This subtle construction distinction may have important implications for assessing gender role identification in children. It has been suggested that young children's emerging gender role identity is especially likely to reflect cultural prescriptions of gender stereotypes (Huston, 1983; Spence, 1985). Thus, if the BSRI items reflect culturally based definitions of masculinity and femininity (Bern, 1985) that do not exclusively represent the instrumental and expressive domains (as do the M and F scales
2 It must be pointed out that Spence (1985) and Spence and Helmreich (1981) would not interpret thesefindingsas a failure of the CPAQ but rather as disconfirmation of Bern's (1985) assertion that sex role identification should be related to a propensity to enact genderstereotypical behavior. This is because Spence's theory asserts that the M and F scales assess nothing more than instrumental and expressive personality traits and should therefore be independent of gender rolerelated behavior to the extent that such behavior is not simply a reflection of the expression of these personality traits.

CHILDREN'S SEX ROLE INVENTORY of the PAQ; Spence et al., 1974), they may be more likely to conform to the gender stereotypes that guide the sex role development of children than are the PAQ/CPAQ (Hall & Halberstadt, 1980) items. Researchers have also noted that children's gender stereotypes do not include some of the socially desirable attributes that adults would regard as masculine or feminine (e.g., intellectual as masculine) but that they do include some attributes that adults would consider undesirable (Albert & Porter, 1983; Davis, Williams, & Best, 1982; Silvern, 1977). Thus, the inclusion of these attributes in the BSRI (Bern, 1974) may actually increase the correspondence of the BSRI scales to children's gender stereotypes. 3 In summary, the rationale for the development of the CSRI included several considerations. First, although there are two widely used adult measures of masculinity and femininity (i.e., the PAQ [Spence et al., 1974] and the BSRI [Bern, 1974]), only the PAQ has been adapted for use with children (i.e., the CPAQ [Hall & Halberstadt, 1980]). A children's version of the BSRI would permit developmental analyses of changing influences of sex typing and androgyny from middle childhood through adulthood with comparable instruments. Second, the small number of studies reporting use of the CPAQ over the past decade, despite continued high interest in gender role research with children, led to questions regarding the usefulness of CPAQ for assessing sex typing or androgyny in children. Close inspection of those studies that used the CPAQ revealed findings that many researchers would consider anomalous, thus confirming these suspicions. Third, subtle construction differences in the BSRI and the PAQ were noted that suggest that a children's version of the BSRI would provide researchers with a unique alternative measure of sex typing and androgyny in children. Specific hypotheses were advanced to establish the validity of the CSRI. First, criterion group validity was expected to be established through significant sex differences on the scales: Boys were expected to score higher in masculinity than girls, and girls were expected to score higher in femininity than boys. Second, the masculinity and femininity scores from a toy- and activity-preference measure of sex typing were expected to be significantly correlated with the congruent measures of sex role identification on the CSRI. Although these correlations were expected to be modest in strength because each measure assessed a unique dimension of sex typing, it was also assumed that children's self-perceived sex-stereotypic personality attributes would be related to their preferences for gender-related toys and activities (Bern, 1985). Support for this prediction in the adult literature has been mixed, but significant moderate effects have been reported (e.g., Bern & Lenney, 1976; O'Heron & Orlofsky, 1990; Orlofsky & Windle, 1978). Notably, O'Heron and Orlofsky (1990) also reported stronger correlations between the BSRI (Bern, 1974) and their Sex Role Behavior Scale (SRBS) than between the PAQ (Spence et al., 1974) and the SRBS. O'Heron and Orolofsky concluded from these findings that "the long-form BSRI may be a broader measure of sex roles than the short-form or the PAQ, both of which appear to measure self-assertive and expressive traits more narrowly" (p. 136). Silvern and Katz (1986) are the only researchers who have examined the relation between children's gender role identification (assessed using the CPAQ; Hall & Halberstadt, 1980) and

507

their sex-typed play preferences, and as previously noted, their findings were inconclusive. Although positive correlations were obtained between the CPAQ F scale and feminine play preference in both boys and girls, a marginally significant negative correlation was obtained between the CPAQ M scale and masculine play preference in boys; the correlation was nonsignificant in girls. Confirmation of the predicted relation between sex role identification and sex-typed toy and activity preferences would lend further credibility to the CSRI as an alternative sex-typing measure for children. In addition, construct validity was expected to be demonstrated through assessments of children's self-concepts and cognitive performance. The Self-Perception Profile for Children (Harter, 1985) was used to assess children's self-concepts. The scale includes a subscale that measures global self-worth as well as five domain-specific subscales. The following hypotheses were advanced for each subscale: 1. Global self-worth. Because a number of studies have documented a relation between masculinity and self-esteem in both adults (e.g., Antill & Cunningham, 1979; Deutsch & Gilbert, 1976; Orlofsky, 1977; Spence, Helmreich, & Stapp, 1975) and children (e.g., Hall & Halberstadt, 1980; Lamke, 1982; Massad, 1981), masculinity was expected to be a significant predictor of global self-worth in both boys and girls. Other studies have found, however, that androgynous persons (high in both masculinity and femininity) score even higher than masculine sextyped persons in self-esteem (Alpert-Gillis & Connell, 1989; Bern, 1977; Hall & Halberstadt, 1980; Spence & Helmreich, 1978; Spence et al., 1975). However, when compared directly with masculinity, femininity clearly accounts for far less of the variance in self-esteem (Whitley, 1983). These findings led to the prediction that masculinity would be a more significant predictor of global self-worth than would femininity and that masculine sex-typed and androgynous children would score higher in this measure than would feminine sex-typed or undifferentiated children. 4 2. Scholastic competence. Two studies have directly examined the relation between perceived scholastic competence and masculinity or femininity, and both found that masculinity was
3 The fact that the BSRI F scale is lower in social desirability than the M scale may simply reflect what some consider the social fact that the traits associated with men are more valued than those associated with women, as Taylor and Hall (1982) have suggested. Taylor and Hall also noted that to attempt to camouflage this fact by artificially adjusting these items to reflect uniformly high levels of social desirability may contribute to further confusion concerning the meaning of the scales. 4 Studies using the BSRI and the PAQ or CPAQ have typically reported results in two forms. Some studies have analyzed M and F scores as independent, continuous measures of identification with gender-related personality traits. Assessment of androgyny from this approach is achieved by examining the effects of the Masculinity X Femininity interaction (Hall & Taylor, 1985). Another large body of sex-typing and androgyny literature has used the median-split method for classifying individuals into groups (cf. Bern, 1977; Spence et al., 1975). Individuals who score above the median on both M and F are classified as "androgynous," those above the median on M and below the median on F as "masculine sex typed," those below the median on M and above the median on F as "feminine sex typed," and those below the median on both dimensions as "undifferentiated."

508

JANET P. BOLDIZAR

significantly correlated with self-perceived intellectual or school competence and that this relation was significant only for girls (Hall & Halberstadt, 1980; Wilson & Cairns, 1988). Thus, only masculinity was expected to be significantly related to perceived scholastic competence, and this relation was expected to be especially strong for girls. 3. Social acceptance. Although femininity has been shown to be positively correlated with expressive qualities such as nurturance (Bern, Martyna, & Watson, 1976), empathy (Orlofsky & Windle, 1978; Spence & Helmreich, 1978), and sympathetic listening (Bern, 1975) that might be expected to enhance objective social acceptance, the limited existing data do not provide support for the assumption that these qualities contribute to more positive self-perceptions of social acceptance. Rather, Wilson and Cairns (1988) found that masculinity was the only significant predictor of perceived social competence for both sexes, whereas Hall and Halberstadt (1980) found no significant relations between self-perceived popularity and masculinity or femininity. Thus, only masculinity was expected to predict selfperceived social acceptance. 4. Athletic competence. Wilson and Cairns (1988) found significant correlations between masculinity and perceived physical competence for both boys and girls, but the correlations between femininity and perceived physical competence were nonsignificant. These findings suggested that perceived athletic competence would be strongly associated with masculinity for both sexes. 5. Physical attractiveness. Only Hall and Halberstadt (1980) have examined the relation between sex typing and self-conceptions of physical appearance. They reported significant correlations between masculinity and perceived physical attractiveness, as well as significantly higher scores for androgynous (versus others) children. These findings led to the prediction that both masculinity and androgyny would be related to children's perceived physical attractiveness. 6. Behavioral conduct. This is the only domain in which femininity was expected to play a significant role. Hall and Halberstadt (1980) found that feminine sex-typed children had higher self-concepts with regard to good behavior than did masculine sex-typed children. Feminine sex-typed persons have also been found to be more conforming, submissive, and nonassertive than persons of other gender role categories (Baucom, 1980; Bern, 1975; Berzings, Welling, & Wetter, 1976), which suggests that they may perceive themselves as especially well behaved in the structured classroom setting. Cognitive performance was assessed through IQ and standardized school mathematics and verbal achievement tests. General intelligence was expected to be largely uncorrelated with masculinity and femininity, and if confirmed, this finding would help to establish the discriminant validity of the CSRI. However, Hall and Halberstadt (1980) found that for girls, IQ and masculinity were positively correlated and that higher mathematics achievement scores were negatively correlated with femininity. Thesefindingsare consistent with the results of Signorella and Jamison's (1986) meta-analysis that found that higher masculine and lower feminine scores were associated with better mathematical task performance and that this relation was more consistently observed for girls. With regard to verbal achievement scores, little evidence exists in support of the ste-

reotypical expectation that verbal skills are associated with higher femininity (Huston, 1983). Rather, Signorella and Jamison found that masculinity, but not femininity or androgyny, was associated with better performance on verbal tasks and that this association was especially strong for adolescent girls. Thus, masculinity was expected to be positively related to both mathematics and verbal achievement, and these relations were expected to be stronger for girls than for boys.

Method CSRI Development and Preliminary Validation


The items for the CSRI were adapted directly from the BSRI (Bern, 1974). The CSRI thus included 20 masculine items, 20 feminine items, and 20 neutral items that served asfillers.The CSRI items and their BSRI equivalents are presented in Table 1. The items were presented in a repeating pattern of one neutral item, one feminine item, and one masculine item. Each item was worded as a statement about the self, and subjects were asked to rate themselves according to "how true of you" each one was on the following 4-point scale: 4 = very true of me, 3 = mostly true of me, 2 = a little true of me, and 1 = not at all true of me. Masculinity and femininity scores were computed by averaging the responses to the 20 items on each scale and could therefore range from 4 (highest) to 1 (lowest). To establish initial construct validity of the CSRI and to demonstrate that the CSRI and the BSRI (Bern, 1974) were roughly equivalent measures of masculinity and femininity, the two scales were administered (in counterbalanced order) to a group of 47 adult men and women. The correlation between the BSRI and CSRI M scales was .86, and the correlation between the two F scales was .89 (ps < .001).5

Subjects
The 145 children in this study (71 girls and 74 boys) attended the third (n = 25, mean age = 9.0 years), fourth (n = 43, mean age = 9.8 years), sixth (n = 42, mean age = 11.6 years), and seventh (n = 35, mean age = 12.7 years) grades of a university school serving a middle-class community.6

Stimulus Materials
In addition to the CSRI, the children also completed two other questionnaires. The Toy and Activities Preferences Questionnaire (TAPQ; Perry, Perry, & Hynes, 1990) consists of 45 toys and activities typical of children in these age groups and asks children to indicate how much they like or dislike each one by crossing out the appropriate response. Responses are ranked indicating choice on the following 5-point scale: 1 = dislike it a lot, 2 = dislike it a little, 3 = neither dislike nor like it, 4 = like it a little, and 5 = like it a lot. Fifteen of the items are masculine (i.e., significantly more likely to be endorsed by boys than by girls), 15 are feminine (i.e., significantly more likely to be endorsed by girls than by boys), and 15 are neutralfillers.The three types of items are randomly distributed across the questionnaire. Toy and activity preference scores were computed by averaging the 15 responses in each of the F and M scales and could range from 5 (highest) to 1 (lowest). The second questionnaire was the Harter Self-Perception Profile for Children (SPPC; Harter, 1985). This scale consists of six separate subscales that assess children's self-perceptions of global self-worth as well All significance tests are two-tailed. Fifth graders were not included in this study because they were involved in another study at the time of the data collection.
6 5

CHILDREN'S SEX ROLE INVENTORY

509

Table 1 Items From the Children's Sex Role Inventory (CSRI) and Their Equivalents From the Bern Sex Role Inventory (BSRI; Bern, 1974) CSRI Masculine items It's easy for me to make up my mind about things. I can take care of myself. I can control a lot of the kids in my class.* I like to do things that boys and men do. When a decision has to be made, it's easy for me to take a stand." I get pretty angry if someone gets in my way. I am a leader among my friends.* I'd rather do things my own way than take directions from others. When I play games, I really like to win.* I'm willing to work hard to get what I want. I am sure of my abilities." I stand up for what I believe in.* [ would rather do things on my own than ask others for help. 1 am good at sports.* It's easy for me to tell people what I think, even when I know they will probably disagree with me." [ make a strong impression on most people I meet.* [ can get people to do what I want them to do most of the time. [ like to think about and solve problems. I am good at taking charge of things.* I am willing to take risks. Feminine items [ care about what happens to others." When someone's feelings have been hurt, I try to make them feel better." [ usually speak softly. [ am a warm person." [ am a kind and caring person." It's easy for people to get me to believe what they tell me. Sometimes I like to do things that younger kids do. [ don't like to say "bad" words or swear. [ like babies and small children a lot* * Items on the short CSRI. as scholastic competence, social acceptance, athletic competence, physical appearance, and behavioral conduct. Each subscale consists of six items, each of which is followed by a description of two types of "kids": one who is "like" the item and another who is not. The child is first asked to decide which kind of kid is most like herself or himself and then to decide whether this is "really true" or "sort of true." Half of the items present the high-competent type of kidfirst,and half present the low-competent type of kid first. Because this type of structured alternative response format legitimizes either choice (by implying that either response is shared by other children who feel the same way), it controls for social desirability concerns that have been problematic in other children's self-esteem scales (e.g., Piers-Harris Self-Concept Scale; Piers, 1969). Scores were calculated byfirstassigning the following values: 4 (really true of me) or 3 (sort of true of me) for high-competent kids and 2 (sort of true of me) and 1 (really true of me) for low-competent kids. Average scores were then computed from the sum of the six responses within each of the six subscales. Scores could range from 4 (most competent) to 1 (least competent). Measures of cognitive ability were available from school records. The IQ test scores were based on the Cognitive Abilities Test (Thorndike & Hagen, 1986), and the standardized mathematics and verbal achievement scores were taken from the Iowa Tests of Basic Skills (Hieronymous et al., 1986). All tests had been administered about 1 month prior to the data collection. The mean IQ score was 106.64, and the mean standardized mathematics and verbal achievement scores were 108.47 and 108.26, respectively. Compassionate Eager to soothe hurt feelings Soft-spoken Warm Tender Gullible Childlike Does not use harsh language Loves children Makes decisions easily Self-sufficient Dominant Masculine Willing to take a stand Aggressive Acts like a leader Individualistic Competitive Ambitious Self-reliant Defends own beliefs Independent Athletic Assertive Strong personality Forceful Analytical Has leadership abilities Willing to take risks BSRI CSRI Feminine items (continued) I am a gentle person." When there's a disagreement, I usually give in and let others have their way. I am a cheerful person." I feel shy around new people. When I like someone, I do nice things for them to show them how I feel." I feel good when people say nice things about me. I am faithful to my friends. I like to do things that girls and women do." It makes me feel bad when someone else is feeling bad.* I can usually tell when someone needs help. I'm good at understanding other people's problems. Neutral items I am an honest person. I think I'm better than most of the other people I know. People like me." I am a serious person. I have many friends.* I usually get things done on time. It's easy for me to fit into new places." I'm always losing things." I am careful not to say things that will hurt someone's feelin&s I like to do things that other people do." I like to help others. I am a moody person.* I'm the kind of person others can depend on. I like acting in front of other people.* I am a happy person. I never know what I'm going to do from one minute to the next.* I always do what I say I will do." I feel bad when other people have something that I don't have." I try to tell the truth. I like to keep secrets. Sincere Conceited Likable Solemn Friendly (In)efficient Adaptable Unsystematic Tactful Conventional Helpful Moody Responsible Theatrical Happy Unpredictable Reliable Jealous Truthful Secretive Gentle Yielding Cheerful Shy Affectionate Flatterable Loyal Feminine Sympathetic Sensitive to the needs of others Understanding BSRI

510
Procedure

JANET P. BOLDIZAR

The questionnaires were administered to the children in classroom groups at two separate testing sessions. In the first testing session, the children completed the CSRI. The scale was introduced as a "survey about how children your age think about themselves." Children's desks were physically separated and they were given cover sheets to assure the privacy of their answers. The items were read aloud to the third and fourth graders to control for individual differences in reading ability. In the second testing session, which took place approximately 1 week after the first, the SPPC (Harter, 1985) and the TAPQ (Perry et al., 1990) were administered. Approximately 1 year after the first data collection, 130 of the original 145 children (66 girls and 64 boys) from the fourth, fifth, seventh, and eighth grades of the same university school participated in a follow-up testing session to establish the test-retest reliability of the CSRI.7 Children were again tested in classroom groups. They completed the CSRI and a separate short form of the CSRI. The short form was composed of 30 items: 10 masculine, 10 feminine, and 10 neutral. The masculine and feminine items were selected on the basis of their item-total correlations with their respective scale, and the neutral items were those on which girls and boys differed the least.

CSRI and TAPQ scales were correlated, the TAPQ M scale was not associated with children's self-perceptions of competence, as was the CSRI M scale. This finding suggests that these two types of scales measure different dimensions of sex typing in children and that each can make a unique contribution to our understanding of this phenomenon. It was not clear, however, that this was the case for the CSRI and TAPQ F scales because both were positively correlated with the behavioral conduct subscale of the SPPC. To determine whether the femininity measures were making independent contributions to children's self-perceptions of behavioral conduct, a regression analysis was performed in which sex was entered into the equation first, followed by CSRI and TAPQ F scores. Results confirmed that both CSRI and TAPQ F scores made significant independent contributions to the prediction of self-perceived behavioral conduct, Fs(3,141 ) = 21.54 and 3.95, ps < .001 and .05, respectively. Finally, the lack of significant relations between the CSRI M or F scales and any of the cognitive measures demonstrates the discriminant validity of the scales. Regression Analyses of the SPPC Subscales

Results The reliability of the two scales was assessed through alpha coefficients of internal consistency. The alpha for the CSRI M scale was .75, and the alpha for the CSRI F scale was .84. The correlation between the two scales was nonsignificant, demonstrating the expected independence of the two scales. The TAPQ (Perry et al., 1990) M and F scale alphas were .89 and .90, respectively, and the correlation between these two scales was also nonsignificant. The SPPC (Harter, 1985) alpha coefficients were .80, .81, .79, .78, .79, and .78 for the subscales scholastic competence, social acceptance, athletic competence, physical attractiveness, behavioral conduct, and global self-worth, respectively. Tests of significant differences between girls and boys on the measures of interest, along with the respective means and standard deviations are presented in Table 2. As predicted, girls scored significantly higher on both the CSRI and TAPQ (Perry et al., 1990) F scales and the behavioral conduct subscale of the SPPC (Harter, 1985) than did boys. Boys scored significantly higher on the M scales of the CSRI and the TAPQ, as well as on the athletic competence and physical attractiveness subscales of the SPPC, than did girls. These findings of gender differences on the SPPC parallel those reported by Harter (1985) in her description of the psychometric properties of the SPPC. No significant gender differences were observed for any of the measures of cognitive performance. Correlations between the CSRI M and F scales, the TAPQ (Perry et al., 1990) M and F scales, and the other measures of interest are reported in Table 3. In support of the validity of the CSRI, the CSRI F scale was significantly positively correlated with the TAPQ F scale and with the SPPC (Harter, 1985) behavioral conduct subscale. The CSRI F scale was negatively correlated with the TAPQ M scale and with age. The CSRI M scale was significantly positively correlated with the TAPQ M scale, as well as with the global self-worth, scholastic competence, social acceptance, athletic competence, and physical attractiveness subscales of the SPPC. It is noteworthy that, although the

To determine the extent to which scores on the CSRI M and F scales contributed to the differences in children's self-perceptions of competence over and above the contributions of sex, regression analyses were performed on each of the subscales with sex entered into the equation first, followed by masculinity and femininity and finally by the Masculinity X Femininity interaction term. Only masculinity was a significant predictor of global self-worth, F(3,141) = 11.78, p < .001; scholastic competence, F(3, 141) = 16.93, p < .001; social acceptance, F(3, 141)= 13.07, p < .001; and athletic competence, F(3, 141) = 31.30, p< .001. Femininity significantly predicted behavioral conduct, F(3,141) = 36.09, p < .001. Both sex and masculinity were significant predictors of perceived physical attractiveness, Fs(3,141) = 6.94 and 7.49, respectively, p < .01 for both. The Masculinity X Femininity interaction term did not make any significant contributions to the equations over and above those noted previously for the main effects of masculinity and femininity. These findings are highly consistent with those reported in both the adult and child sex-typing literature and provide strong support for the construct validity of the CSRI. 8
7 Minor revisions were made in the wording of some of the items that attenuated the reliability of the original scales (primarily on the M scale) in order to improve the internal consistency of the CSRI scales. 8 To determine whether results using the CSRI were comparable to those reported in studies that compared sex-typed and androgynous persons using the popular median-split method of classification (e.g., Bern, 1977; Spence et al., 1975), additional analyses directly contrasting the groups of interest were performed. To do this, contrast terms were created (cf. Taylor & Hall, 1982) to compare children classified as androgynous (by median-split method), masculine sex-typed, or feminine sex-typed with others. These contrast terms were then simultaneously entered into the regression equations after entering sex. (It was not possible to examine gender interactions with these contrast variables because of instability created by the extremely small ns in the cross-sex-typed groups.) Androgyny and masculine sex typing made independent positive contributions to self-perceived scholastic competence, .Fs(4,140) = 4.40 and 5.83, respectively, p < .05 for both; social

CHILDREN'S SEX ROLE INVENTORY

511

Table 2 Gender Differences on the CSRI Scales, the TAPQ Sacks, the SPPC Subscales, and Measures ofCognitive Performance
Girls" Measure CSRI Femininity Masculinity TAPQ Feminine Masculine SPPC Global self-worth Scholastic competence Social acceptance Athletic competence Physical attractiveness Behavioral conduct Cognitive performance IQ Verbal achievement Mathematics achievement M 3.10 2.69 2.78 2.35 3.32 2.98 3.07 2.92 2.80 3.19 107.02 109.40 107.21 SD 0.37 0.36 0.83 0.92 0.58 0.68 0.76 0.66 0.72 0.51 13.15 16.14 12.82 M 2.60 2.88 1.63 3.19 3.29 2.92 3.12 3.16 3.23 2.77 106.27 107.18 109.69 Boys" SD 0.40 0.38 0.60 0.87 0.62 0.63 0.66 0.60 0.61 0.66 13.46 12.84 15.54 '(143) 7.68*** -3.19** 9.58*** -5.69*** 0.30 0.61 -0.45 -2.37* -3.88*** 4.37*** 0.34 0.92 -1.04

Note. CSRI = Children's Sex Role Inventory; TAPQ = Toy and Activities Preferences Questionnaire (Perry et al., 1990); SPPC = Harter Self-Perception Profile for Children (Harter, 1985). "n = 71. b n = 74. *p<.05. **/><.01. ***p< .001.

Possible gender differences in the effects of masculinity, femininity, and the Masculinity X Femininity interaction term on the SPPC (Harter, 1985) subscales were tested by adding the interactions of sex with these variables to the equations. Gender interactions did not make significant contributions in any of the SPPC subscale equations. Regression Analyses of the Cognitive Performance Variables Regression analyses were performed following the same procedures described above. Neither the IQ nor the verbal achieveacceptance, Fs(4,140) = 5.97 and 6.52, respectively, p < .05 for both; and athletic competence, Fs(4,140) = 11.91 and 13.98, respectively, p < .001 for both. Androgyny and feminine sex typing made independent positive contributions to perceived behavioral conduct, Fs(4, 140) = 5.68 and 4.13, respectively, p < .05 for both. Only androgyny was positively associated with global self-worth, F(4,140) = 6.33, p < .05, and none of the gender role categories made significant contributions to the prediction of perceived physical attractiveness after controlling for sex. Thesefindingsare quite consistent with those of other studies that have used this method to classify subjects into gender role categories, and they suggest that androgynous children (who have high levels of both masculinity and femininity) may be somewhat better adjusted across the total range of competence and self-worth measures than either masculine or feminine sex-typed children. However, in light of the earlier regression analyses that found that only masculinity made significant contributions to all of the subscales except behavioral conduct (in which femininity was the most important predictor), the inescapable conclusion is that being high in masculinity is the basis of the androgynous children's advantage in global self-worth, scholastic competence, social acceptance, and athletic competence.

ment analyses revealed significant predictors, so only the results of the mathematics achievement analysis are discussed. Although neither masculinity or femininity, nor the Masculinity X Femininity interaction term contributed significantly to the prediction of mathematics achievement, the interactions of sex with the CSRI variables did make significant contributions to the equations. Separate regression analyses by sex were performed to determine significant proportions of the variance in the mathematics achievement scores of girls and boys accounted for by masculinity, femininity, and Masculinity X Femininity interaction. For girls, only femininity was a significant predictor of mathematics achievement scores, F(i, 67) = 5.38, p < .05, and, consistent with earlier findings, higher CSRI F scores detracted from girls' mathematics performance. In contrast, only masculinity was a significant predictor of boys' mathematics achievement, FQ, 70) = 4.02, p < .05, and higher CSRI M scores enhanced boys' performance. Age-Related Changes in Femininity and Masculinity To explore possible developmental changes in CSRI scores, analyses of variance (ANOVAs) on femininity and masculinity with grade and sex entered as between-subjects factors were performed. Aside from the highly significant sex effects already discussed, the femininity analysis revealed a significant main effect of grade, F(3,137) = 4.30, p < .01, which was qualified by a significant Grade X Sex interaction, F(3,137) = 3.94, p < .01. Consistent with the negative correlation between age and femininity previously reported, there was a general decline in CSRI F scores for older children. However, the Sex X Grade interaction, illustrated in Figure 1, suggested that although there was a decline in girls' CSRI F scores between the third and sixth

512

JANET P. BOLDIZAR

Table 3 Correlations Between Masculinity, Femininity, Children's Self-Perceptions, Cognitive Performance, and Age
CSRI Measure CSRI Femininity Masculinity TAPQ Feminine SPPC Global self-worth Scholastic competence Social acceptance Athletic competence Physical attractiveness Behavioral conduct Cognitive performance
IQ Age

TAPQ Masculinity -.02

Femininity

Feminine .59*** -.09

Masculine -.22** .23** -.11 -.13 -.07


.00

.07 .15 .03

-.07 -.13 .53*** -.09 -.09


.02

.27** .32*** .30*** .46*** .29*** -.13


.01 .11 .07

.08 .09 .09 -.06 -.06 .44*** -.04 -.05 .08 -.23***

-.02
.13

-.12
.02 .06 .03

Mathematics achievement Verbal achievement

-.28***

-.04

-.10

Note. All tests of significance are based on 143 degrees of freedom. CSRI = Children's Sex Role Inventory; TAPQ = Toy and Activity Preferences Questionnaire; SPPC = Harter Self-Perception Profile for Children. *p<.05. **/?<.01. ***/>< .001.

grades, the seventh-grade girls' F scores were actually slightly higher. The pattern was nearly reversed for the boys, whose CSRI F scores increased slightly between the third and fourth grades and then declined in the sixth and seventh grades. Another way to describe the interaction is to note that the greatest sex differences in CSRI F scores existed among the youngest and oldest groups of children. There were no significant main effects or interactions involving grade in the masculinity analysis. One-Year Follow-Up Results Alpha coefficients of internal consistency of the CSRI F and M scales were .85 and .80, respectively. The test-retest reliabilities were .71 for the F scale and.56 for the M scale (ps < .001).

Internal consistency of the short CSRI F and M scales was evident in respective alpha coefficients of .89 and .78. The validity of the short CSRI was evident in high correlations between the original and short F and M scales (rs = .91 and .83, respectively, p < .001 for both), as well as large significant differences between boys and girls on the two scales. On the short CSRI F scale, girls scored significantly higher than boys, Ms = 3.37 and 2.74, respectively, /(128) = 7.18, p < .001, and on the short CSRI M scale, boys scored significantly higher than girls, Ms = 3.05 and 2.80, respectively, /(128) = 3.15, p < .01. Change in CSRI M and F scores over 1 year was assessed by comparing the group averages broken down by grade and sex from \fear 1 to Year 2. No significant changes were found within these groups, although trends paralleled the age changes previously described.

Discussion'
I Girls ^ ^ Boys

Figure 1. Mean femininity scores as a function of grade and sex.

The results of these studies indicate that the CSRI is a reliable and valid instrument for the assessment of sex typing and androgyny in children. Reliability was demonstrated by high levels of internal consistency (average a = .81 for the two scales over the two testing sessions) and stability over a 1-year period (average r =. 64). These psychometric properties compare favorably with those reported by Hall and Halberstadt (1980) for the CPAQ; the average internal consistency coefficient was .60, and average test-retest reliability was .44. Validity of the CSRI was evident in confirmed predictions of relations between CSRI M or F scales and gender-based toy and activity preferences, selfperceptions, and cognitive abilities, as well as of significant differences between boys and girls on the M and F scales ( p< .01 for both scales). High correlations (average r = .88) were also

CHILDREN'S SEX ROLE INVENTORY

513

obtained between the CSRI and the BSRI (Bern, 1974) scales in an adult sample, suggesting that the two inventories are roughly equivalent and could be used to assess the development of sex typing and androgyny from middle childhood through adulthood. In addition, a short form of the CSRI was developed from the items with the highest item-total correlations on the full-length CSRI. The short CSRI M and F scales were significantly correlated with the analogous full-length CSRI M and F scales (average r = .84), and the short CSRI was also successful in differentiating between boys and girls on the M and F scales at probabilities less than .01. Significant correlations between CSRI M or F scales and the congruent M and F scales of a toy- and activity-preference measure of sex typing provided support for the construct validity of the CSRI. These findings are consistent with Bern's (1981a, 1985) assertion that the gender-based personality attributes assessed by the BSRI (Bern, 1974) may serve as global organizers of sex-related cognitions and behaviors. Confirmation of predicted relations between CSRI M or F scales and domain-specific self-concept measures provided further evidence for the validity of the CSRI. Multiple regression analyses revealed that masculinity was a significant predictor of self-perceived scholastic competence, social acceptance, athletic competence, physical attractiveness, and global self-worth and that femininity was a significant predictor of perceived positive behavioral conduct. These findings are consistent with those of studies that have found that masculinity accounts for the majority of the variance in self-esteem measures but that femininity may be more strongly related to some components of self-esteem than is masculinity (Whitley, 1983). The fact that CSRI masculinity, but not TAPQ (Perry et al., 1990) masculinity, predicted children's self-perceived competence in five of the six domains was an interesting finding, especially in light of the moderate correlation between the two scales. This implies that it is not the types of toys and activities that children prefer but their self-perceived personality attributes that contribute to the relation between masculinity and self-esteem. Thisfindingis consistent with those of other studies that have found weaker relations between measures of psychological adjustment and masculine sex-typed behavioral preferences than between adjustment and masculine sex-type selfperceptions (O'Heron & Orlofsky, 1990; Silvern & Katz, 1986). Although a number of explanations have been suggested to explain the relation between self-perceived masculinity and selfesteem (for reviews, see Cook, 1985; Taylor & Hall, 1982; Whitley, 1983), the fact that the two self-perception scales (i.e., the CSRI and the SPPC, Harter, 1985) may assess conceptually overlapping domains (Nichols, Licht, & Pearl, 1982) cannot be overlooked. For example, the CSRI M scale items that refer to being good at sports, being a leader among friends, and thinking about and solving problems could clearly be linked to the athletic competence, social acceptance, and scholastic competence subscales of the SPPC, respectively. However, because the TAPQ M scale also contains a number of sports-related items that might have been expected to relate to the SPPC athletic competence subscale, the differences may not be solely attributable to shared measurement variance. Further research is needed to explain why masculine stereotypical self-percep-

tions, but not masculine sex-typed behavioral preferences, are related to children's self-esteem. Developmental changes in femininity across grades indicated that the trend of declining CSRI F scores for girls between the third and sixth grades reversed itself for girls in the seventh grade. At the same time, boys' CSRI F scores increased between the third and fourth grades and then declined in the higher grades. This pattern suggests that, rather than a monotonic decrease in rigidity of sex role stereotypes as children get older (at least as far as the feminine self-concept is concerned), there is an initial decline followed by a resurgence of sex role stereotyping as children approach adolescence. It is possible that changes in social psychological awareness created by newly salient demands to conform to adult sex role stereotypical expectations in interactions with the opposite sex may be responsible for this phenomenon. It should be noted that only femininity was related to age in this sample and that masculinity was unrelated to age for either boys (r = .01) or girls (r = .13). Other studies have reported that, although the generalization of movement toward more rigid sex typing during adolescence is characteristic of adolescent females, the trend is for males to move toward more androgynous identification during later adolescence (Hyde & Phillis, 1979; Ziegler, Dusek, & Carter, 1984). Because the seventh and eighth grades are likely to represent a developmental turning point with respect to sex typing and androgyny, it is important that future studies be designed to follow the development of gender role identification from middle childhood through adolescence. The independence of the CSRI F or M scales and IQ or verbal achievement provided evidence of the discriminant validity of the scales. However, regression analyses revealed that, whereas masculinity positively predicted boys' mathematics achievement scores, femininity was a significant negative predictor of girls' mathematics achievement scores. Further analyses determined that the CSRI M scale was negatively, but not significantly, correlated with girls' mathematics achievement scores (r = -.08) and that the CSRI F scale was positively, but not significantly, correlated with boys' mathematics achievement (r = . 12). One possible explanation for this gender difference in the influences of masculinity and femininity on mathematics achievement is that the CSRI M and F effects on mathematics achievement were mediated by the types of activities in which boys and girls typically engage. That is, if masculine boys typically engage in masculine sex-typed activities that enhance mathematics performance and feminine girls typically engage in feminine sex-typed activities that do not promote mathematics performance, the observed sex differences might be explained. However, the lack of significant correlations between the TAPQ (Perry et al., 1990) scales and math achievement for either boys (r = .02 for TAPQ M; r = -.02 for TAPQ F) or girls (r = .03 for TAPQ M; r = .02 for TAPQ F) ruled out this possibility. Although it is possible to speculate on other explanations for this phenomenon (e.g., interactions between gender-stereotypic role expectancies and the classroom environment or divergent relations between mathematics achievement and values associated with girls' and boys' gender role stereotypes), to do so here would go beyond the scope of this article. More research is needed to pinpoint the exact mechanisms through which femi-

514

JANET P. BOLDIZAR Berzings, J. I., Welling, M. A., & Wetter, R. E. (1976, September). Androgynous vs. traditional sex roles and the interpersonal behavior circle. Paper presented at the annual meeting of the American Psychological Association, Washington, DC. Brown, D. G. (1956). Sex-role preference in young children. Psychological Monographs, 70 (14, Serial No. 421). Cook, E. P. (1985). Psychological androgyny. New York: Pergamon Press. Davis, S. W, Williams, J. E., & Best, D. L. (1982). Sex-trait stereotypes in the self- and peer descriptions of third grade children. Sex Roles, 5,315-331. Deutsch, C. J., & Gilbert, L. A. (1976). Sex role stereotypes: Effects on perceptions of self and others and on personal adjustment. Journal of Counseling Psychology, 23, 373-379. Downs, C. A., & Langlois, J. H. (1988). Sex typing: Construct and measurement issues. Sex Roles, 18, 87-100. Hall, J. A., & Halberstadt, A. G. (1980). Masculinity and femininity in children: Development of the Children's Personal Attributes Questionnaire. Developmental Psychology, 16, 270-280. Hall, J. A., & Halberstadt, A. G. (1981). Sex roles and nonverbal communication. Sex Roles, 7, 273-287. Hall, J. A., & Taylor, M. C. (1985). Psychological androgyny and the Masculinity X Femininity interaction. Journal ofPersonality and Social Psychology, 49, 429-435. Harter, S. (1983). Developmental perspectives on the self-system. In M. Hetherington (Vol. Ed.) & P. H. Mussen (Series Ed.), Handbook of child psychology: Vol. 4. Socialization, personality, and social development (pp. 275-386). New York: Wiley. Harter, S. L. (1985). Manual for the Self-Perception Profile for Children. Denver, CO: University of Denver. Hieronymus, A. N., Hoover, H. D., Lindquist, E. F, Oberley, K. R., Cantor, N. K., Burdick, D. D., Lewis, E. L., Hyde, E. L., QuailsPayne, A. L. (1986). Iowa Tests of Basic Skills. Chicago: Riverside. Huston, A. C. (1983). Sex-typing. In M. Hetherington (Vol. Ed.) & P. H. Mussen (Series Ed.), Handbook ofchild psychology: Vol. 4. Socialization, personality, and social development (pp. 387-468). New \brk: Wiley. Huston, A. C. (1985). The development of sex typing: Themes from recent research. Developmental Review, 5, 1-17. Hyde, J., & Phillis, D. (1979). Androgyny across the life-span. Developmental Psychology, 15, 334-336. Institute for Scientific Information, Inc. (1981-1989). Social Sciences Citation Index. Philadelphia, PA: Author. Lamke, L. K. (1982). The impact of sex-role orientation on self-esteem in early adolescence. Child Development, 53, 1530-1535. Massad, C. M. (1981). Sex role identity and adjustment during adolescence. Child Development, 52, 1290-1298. Mitchell, J. E., Baker, L. A., & Jacklin, C. N. (1989). Masculinity and femininity in twin children: Genetic and environmental factors. Child Development, 60, 1475-1485. Nichols, J. A., Licht, B. G., & Pearl, R. A. (1982). Some dangers of using personality questionnaires to study personality. Psychological Bulletin, 92, 572-580. O'Heron, C. A., & Orlofsky, J. L. (1990). Stereotypic and nonstereotypic sex role trait and behavior orientations, gender identity, and psychologica\adjmtment. Journal ofPersonality and Social Psychology, 58, 134-143. Orlofsky, J. L. (1977). Sex-role orientation, identity formation and selfesteem in college men and women. Sex Roles, 3, 561-575. Orlofsky, J. L., & Windle, M. T. (1978). Sex-role orientation, behavioral adaptability and personal adjustment. Sex Roles, 4, 801-811. Pedhazur, E. J., & Tetenbaum, T. J. (1979). Bern Sex Role Inventory: A theoretical and methodological critique. Journal of Personality and Social Psychology, 37, 996-1016.

ninity detracts from girls' mathematics performance and masculinity enhances boys' mathematics performance. Given the robust psychometric properties of the CSRI, it should prove to be a useful new alternative for researchers interested in studying sex typing and androgyny in children. Taylor and Hall (1982) have distinguished between tests of construct validity in the development of new androgyny instruments and tests of substantive androgyny hypotheses; the primary focus of the present study was necessarily limited to the former. It is clear that future studies that use the CSRI must be designed to address issues of substantive hypothesis testing related to the many unanswered questions concerning the theoretical meaning and interpretation of the M and F scales (cf. Bern, 1985; Spence, 1985), as well as those concerning relations among the CSRI, CPAQ (Hall & Halberstadt, 1980), and other children's sex-typing constructs (cf. Huston, 1983). Future studies should also address the developmental and gender-related issues that have been raised here and elsewhere in the sex-typing and androgyny literature. Although the data indicate that many observed sex differences in cognitive and behavioral performance can be accounted for by sex role socialization processes in which girls and boys come to identify with congruent sex roles, the larger question that remains unanswered is how these socialized roles interact with sex to produce different paths of development for boys and girls.

References
Albert, A. A., & Porter, J. R. (1983). Age patterns in the development of children's gender-role stereotypes. Sex Roles, 9, 59-68. Alpert-Gillis, L. J., & Connell, J. P. (1989). Gender and sex-role influences on children's self-esteem. Journal of Personality, 57, 97-114. Antill, J. K., & Cunningham, J. D. (1979). Self-esteem as a function of masculinity in both sexes. Journal ofConsulting and Clinical Psychology, 47, 783-785. Baucom, D. H. (1980). Independent CPI masculinity and femininity scales: Psychological correlates and a sex-role typology. Journal of Personality Assessment, 44, 262-271. Beere, C. A. (1979). Women and women's issues: A handbook oftests and measures. San Francisco: Jossey-Bass. Bern, S. L. (1974). The measurement of psychological androgyny. Journal of Consulting and Clinical Psychology, 42,155-162. Bern, S. L. (1975). Sex role adaptability: One consequence of psychological androgyny. Journal of Personality and Social Psychology, 31, 634-643. Bern, S. L. (1977). On the utility of alternative procedures for assessing psychological androgyny. Journal oj Consulting and Clinical Psychology, 45, 196-205. Bern, S. L. (1981a). Gender schema theory: A cognitive account of sex typing. Psychological Review, 88, 354-364. Bern, S. L. (1981b). The BSRI and gender schema theory: A reply to Spence and Helmreich. Psychological Review, 88, 369-371. Bern, S. L. (1985). Androgyny and gender schema theory: A conceptual and empirical integration. In T. B. Sonderegger (Ed.), Nebraska Symposium on Motivation: Psychology and gender (pp. 179-226). Lincoln: University of Nebraska Press. Bern, S. L., & Lenney, E. (1976). Sex typing and the avoidance of crosssex behavior. Journal of Personality and Social Psychology, 33, 4854. Bern, S. L., Martyna, W, & Watson, C. (1976). Sex typing and androgyny: Further explorations of the expressive domain. Journal of Personality and Social Psychology, 34, 1016-1023.

CHILDREN'S SEX ROLE INVENTORY

515

Perry, L. C, Perry, D. G., & Hynes, J. (1990). The relation of sex-typing Spence, J. X, Helmreich, R. L., & Stapp, J. (1975). Ratings of self and peers on sex role attributes and their relation to self-esteem and to the outcomes children expect for sex-typed behavior. Unpubconceptions of masculinity and femininity. Journal of Personality lished manuscript, Florida Atlantic University, Boca Raton. Piers, E. V. (1969). Manual for the Piers-Harris Children's Self Concept and Social Psychology, 32, 29-39. Scale. Nashville, TN: Counselor Recordings and Tests. Taylor, M. C, & Hall, J. A. (1982). Psychological androgyny: Theories, Signorella, M. L., & Jamison, W (1986). Masculinity, femininity, anmethods, and conclusions. Psychological Bulletin, 92, 347-366. drogyny, and cognitive performance: A meta-analysis. Psychological Thomas, S. J. (1983, April). Providing a concrete context for sex role Bulletin, 100, 207-228. endosement [sic] in adolescents: Some measurement considerations. Silvern, L. E. (1977). Children's sex role preferences: Stronger among Paper presented at the annual meeting of the American Educational girls than among boys. Sex Roles, 3,159-171. Research Association, Montreal, Quebec, Canada. Silvern, L. E, & Katz, P. A. (1986). Gender roles and adjustment in Thomas, S. J., & Robinson, M. (1981). Development of a measure of elementary school children: A multidimensional approach. Sex androgyny for young adolescents. Journal of Early Adolescence, 1, Roles, 14,181-202. ! 95-209. Spence, J. T. (1985). Gender identity and its implications for the conThorndike, R. L., & Hagen, E. (1986). Cognitive Abilities Test. Chicago: cepts of masculinity and femininity. In T. B. Sonderegger (Ed.), NeRiverside. braska symposium on motivation: Psychology and gender (pp. 59- Whitley, B. E. (1983). Sex role orientation and self esteem: A critical 95). Lincoln: University of Nebraska Press. meta-analy tical review. Journal ofPersonality and Social Psychology, Spence, J. X, & Helmreich, R. L. (1978). Masculinity and femininity: 44, 765-785. Their psychological dimensions, correlates, and antecedents. Austin:Wilson, R., & Cairns, E. (1988). Sex-role attributes, perceived compeUniversity of Texas Press. tence, and the development of depression in adolescence. Journal of Spence, J. T., & Helmreich, R. L. (1979). The many faces of androgyny: Child Psychology and Psychiatry, 29, 635-650. A reply to Locksley and Colten. Journal of Personality and Social Ziegler, C, Dusek, J., & Carter, D. (1984). Self-concept and sex-role Psychology, 37,1032-1046. orientation: An investigation of multidimensional aspects of a perSpence, J. T., & Helmreich, R. L. (1981). Androgyny versus gender sonality development in adolescence. Journal of Early Adolescence, schema: A comment on Bern's gender schema theory. Psychological 4, 25-39. Review, 88, 365-368. Spence, J. T., Helmreich, R. L., & Stapp, J. (1974). The Personal AttriReceived November 30,1989 butes Questionnaire: A measure of sex role stereotypes and mascuRevision received December 3,1990 linity-femininity. JSAS Catalog ofSelected Documents in Psychology, Accepted December 6,1990 4,43 (Ms. No. 617).

Anda mungkin juga menyukai