Reconceptualizing Femininity and Masculinity: From

Reconceptualizing Femininity and Masculinity: From Gender Roles to Gender Self-Confidence Rose Marie Hoffman California State University, Long Beach

L. DiAnne Borders The University of North Carolina at Greensboro

John A. Hattie University of Auckland A paradigm shift is proposed whereby researchers move from a stereotypical sex-role conception of the psychological constructs of femininity and masculinity to one that respects diversity in perspectives of what femininity and masculinity may mean to various individuals. A review of the literature revealed gender self-contldence as a viable yet overlooked construct for masculinity and femininity research. Two dimensions of gender self-contldence. identitled as gender self-detlnition and gender self-acceptance, are proposed as constructs for further exploration based upon analyses of a new measure of gender self-confidence, the Hoffman Gender Scale (HGS). Discriminant validity of the HGS is supported by demonstrating that the two constructs it measures differ from that measured by the Bem Sex-Role Inventory (BSRI). Implications of these findings and recommendations for further investigation of gender self-contldence, gender self-detlnition and gender self-acceptance, as possible pathways toward a broader understanding of the multifaceted nature of gender, are discussed. Historically, the labeling of traits as "feminine" or "masculine" has provided the framework for the measurement of femininity and masculinity. There is substantial support, however, for the contention that the measurement of femininity and masculinity, both as it existed in the past and as it exists today, is inadequate (Ashmore, 1990; Constantinople, 1973; Deaux, 1987; Hoffman, in press; Hoffman & Borders, 2001; Lewin, 1984a, 1984b; Lewin & Wild, 1991; Marsh & Myers, 1986; McCreary, 1990; Morawski, 1987; Spence, 1984, 1985. 1991, 1993, Author's Info: Rose Mane Hoffman. PhD. Department of Educational Psychology. Administration, and Counseling. California State University. Long Beach. ED2 - 187: 1230 Belinower Blvd.; Long Beach. CA 90840-2201; (562) 985-5626; fax (562) 9854534; email rhoffman^ csulb.edu. Journal of Social Beliavior and Personality. 2000. Vol. 15. No. 4, 475-503. ©2000 Select Press. Novato. CA, 415/209-9838.

476

JOURNAL OF SOCIAL BEHAVIOR AND PERSONALITY

1999). This inadequacy stems largely from conceptual confusion about the definiti()n(s) of these constructs. Individual women and men may derive a sense of their femininity or masculinity from many different sources. The elusi veness of these concepts has been demonstrated in the literature by the difflculty that even highly educated, articulate individuals have in specifying what constitutes their masculinity and femininity (Spence & Sawin, 1985; Spence & Buckner, 1995, 2000). From the dawn of masculinity and femininity measurement until the 1970s, some basic but questionable assumptions implicitly guided such research (Constantinople. 1973; Lewin. 1984a). It had been generally accepted that masculinity and femininity consisted of, and were conceptually defined by, lists of traits and interests that were based on sex difference statistics (cf. Butcher, Dahlstrom, Graham, Tellegen, & Kaemmer, 1989; Gough, 1964; Guilford & Martin, 1943; Hathaway & McKinley, 1943; Strong, 1927;Terman & Miles, 1936). In other words, the basis for deciding if a trait was masculine or feminine had been its prevalence among members of one sex or the other. This assumption, as well as the assumptions of bipolarity and unidimensionality, were flnally challenged when newer femininity and masculinity measures [e.g., Bem Sex-Role Inventory (BSRI; Bem, 1974), Personal Attributes Questionnaire (PAQ; Spence, Helmreich, & Stapp, 1974)] revolutionized femininity and masculinity research by promoting the concept of androgyny. Androgyny was deflned by these researchers as the integration of traditionally feminine and traditionally masculine characteristics within a single individual. Social desirability of traits and/or their perceived typicality among men and women replaced sex difference statistics as the stated criterion for femininity and masculinity test items. Moreover, the assumption that it was healthy for individuals to be sex-typed was replaced by the assertion that a combination of traditionally feminine and traditionally masculine qualities could be healthy regardless of one's biological sex (Bem, 1974). The relationship between psychological health and gender was redefined. [For a review of androgyny research, see Cook (1987).] The concept of androgyny encouraged thinking beyond what was stereotypically feminine or masculine; however, it still encouraged individuals to perceive traits as feminine or masculine by categorizing them as such, and these "feminine" and "masculine" traits continued to provide the framework for assessing femininity, masculinity, and androgyny. Although differences between the sexes were deemphasized durincr the 1970s, the words feminine/femininity and masculine/masculinity, as labels for speciflc characteristics and sets of traits, received unprecedented attention as researchers sought to study psychological androgyny.

Hoffman etal.

FEMININITY AND MASCULINITY

477

This phenomenon was evidenced by the numbers of studies using the BSRI [i.e., Beere (1990) identifled 795 articles and 167 ERIC documents using the BSRI during the 1980s, and none of those references was a duplicate of those listed in her first book (Beere, 1979)]. The widespread and sometimes indiscriminate use of measures such as the BSRI brought renewed emphasis to what was traditionally considered masculine and feminine, thus reinforcing our stereotypical perceptions of these words. Although masculinity and femininity were no longer perceived as opposite ends of one dimension, the tendency to view them as opposites was inadvertently strengthened. During the 1980s some attempts were made to encourage personal definitions of femininity and masculinity using instruments based on role construct theory (Kelly, 1955). Two role construct repertory tests were developed: (a) the Sex-Rep (Baldwin, Critelli, Stevens, & Russell, 1986) and (b) the Ravinder Sex Role Salience Reptest (Ravinder, 1987). Neither of these instruments was used in more than a few studies. Both are cumbersome to score. Furthermore, despite the intention to circumvent problems related to stereotyping, an invitation to stereotype remained implicit. For example, respondents are asked to describe ways in which other people are feminine or masculine, which may be very different from the respondent's deflnition of femininity or masculinity in relation to self. More recently, a number of important concepts related to masculinity and/or femininity have been explored [e.g., male gender role strain (see Pleck, 1995), gender role conflict (see O'Neil, Good, & Holmes, 1995), womanist identity (see Ossana, Helms, & Leonard, 1992), feminist identity development (Bargad & Hyde, 1991; Fischer et al., 2000, Rickard 1989; all based on model by Downing & Roush, 1985), feminist attitudes (Henley, Meng, O'Brien, McCarthy, & Sockloskie, 1998), feminist ideology (Tolman & Porche, 1999), male reference group identity dependence (Wade & Gelso, 1998)]. Although this growing body of literature offers signiflcant contributions to our understanding of gender, these newer concepts are explored primarily in relation to either men or women. Moreover, those that speciflcally consider femininity and/or masculinity tend to rely on socially rather than personally constructed images of these constructs. The BSRI remains widely used today, despite numerous criticisms of the instrument based on both theoretical and methodological concerns (see Hoffman & Borders, 2001). Hence, the challenge to develop viable alternatives to current masculinity and femininity measures largely remains. Lewin (1984b) argued that it is meaningless to classify traits as either masculine or feminine while simultaneously acknowledging the

478


existence of these traits in both sexes. Even more compelling is that Spence herself (author of one of the most widely used "masculinity" and "femininity" measures, the PAQ) has argued repeatedly that the BSRI and the PAQ are basically measures of instrumentality and expressiveness rather than masculinity and femininity (Spence, 1993, 1999; Spence & Buckner, 1995, 2000; Spence & Helmreich, 1981). She has contended that masculinity and femininity should be conceived of differently. The next obvious question is, if masculinity and femininity should not be defined by stereotypical traits, then how should they be deflned? Might there not be a way to honor diversity in individual perspectives of what femininity and masculinity mean? How can we tap into the meaning of femininity or masculinity for a given individual? An examination of the gender identity construct may provide some answers to these important questions. Gender Identity: A Reconceptualization of Femininity and Masculinity The work of Lewin (1984b) and Spence (1985; Spence & Buckner, 1995, 2000) points to the individual's sense of himself or herself as the critical dimension in a discussion of masculinity and femininity measurement. Both Lewin and Spence have suggested that masculinity can be conceived of as maleness and femininiry asfemaleness. Maleness and femaleness are quite different from the more limited notion of stereotypical male and female roles (Spence, 1985), which is the way masculinity and femininity traditionally have been viewed. Maleness and femaleness suggest something about the core of an individual, and gender identity is used to refer to one's subjective feelings of that maleness or femaleness (Basow, 1992; Golombok & Fivush, 1994). Gender identity has been described as "a basic, existential conviction that one is male or female" (Spence & Sawin, 1985, p. 59); a secure sense or conviction of one's own maleness or femaleness (cf. Green, 1974; Money, 1994); and the "individual's awareness of and satisfaction with being a male or female" (Pleck, 1984, p. 220). As such, gender identity refers to one's sense of confidence in and comfort with being either male or female (Lewin, 1984b). In essence, gender identity reflects an individual's concept of himself or herself as male or female (Golombok & Fivush, 1994). Rather than equating masculinity and femininity with stereotypical gender traits and roles, masculinity and femininity can be reconceptualized in terms of the gender identity construct, and, thus, as part of one's selfconcept. Spence (1985) proposed that ''masculinity and femininity, as they refer to an individual's self-concept, be retained and conceptual-

Hoffman et al.


479

ized as gender identity [italics added]: a basic phenomenological sense of one's maleness and femaleness that parallels awareness and acceptance of one's biological sex and is established early in life" (p. 91). Her recent work (Spence & Buckner, 1995, 2000) has extended this line of thinking. Furthermore, Lewin (1984b) suggested that masculinity and femininity be conceptualized as "the gender-relevant aspects of a person' s self-concept [italics added]," thus allowing for "individual variation in the specific content of the self-image as related to gender" (p. 200). Similarly, Ashmore (1990) argued that "gender identity is an individual's structured set of gendered personal identities" (p. 514) and that this is how femininity and masculinity should be conceived, and BlanchardFields, Suhrer-Roussel, and Hertzog (1994) maintained that gender identity and masculinity/femininity "must be deflned in terms of the individual rather than [be bound to] an experimenter-imposed construction" (p. 452). Part of the challenge to researchers seeking to measure masculinity and femininity via gender identity is that the terms gender identity and gender role identity are frequently, and erroneously, used interchangeably. Gender role identity refers to the degree to which a person identifies with or displays societal, but not necessarily personal, deflnitions of masculinity and femininity (cf. Basow, 1992; Mintz & O'Neil. 1990). Gender role identity is the construct at the core of instruments such as the Bem Sex-Role Inventory and the Personal Attributes Questionnaire. The items that comprise the Masculine and Feminine scales of these instruments were selected because they reflected socially prescribed roles and "typical" traits for men and women. Indeed, Bem's (1981) gender schema theory, which she proposed as the foundation for the BSRI, describes how sex-typing is a process whereby individuals learn what attributes, behaviors, and roles are linked with being female or male. According to Bem, sex-typed (vs. androgynous or undifferentiated) individuals conform to what society deems as "sex-appropriate" roles. Gender identity, on the other hand, is a different and broader concept, according to its deflnitions presented above. For example, nonadherence of males and females to societally prescribed gender roles does not necessarily imply uncertainty regarding one's gender identity (Golombok & Fivush, 1994). Nor, if we accept Spence's contention that masculinity and femininity should be conceptualized as gender identity, does nonadherence to societally prescribed gender roles suggest uncertainty about one's masculinity or femininity. As Spence (1985) proposed, individuals may create their own standards or "calculus" for selfassessing maleness or femaleness. For example, Tangri (1972) found that women whose career aspirations were traditionally "masculine" did

480


not consider themselves to be masculine; in contrast, they deflned their femininity in a variety of other ways. To describe the nature of an indwiduaVs self-concept as he or she relates it to masculinity or femininity would indeed be a more fruitful approach to understanding human behavior than counting the number of ways in which an individual resembles the "typical" member of his or her own sex (Spence & Helmreich, 1978). It is generally agreed that the BSRI focuses on gender role orientation (Ashmore, 1990; Blanchard-Fields et al., 1994), but does gender role orientation necessarily equal masculinity and femininity? We argue that it does not. Gender role identity, then, is more restrictive than gender identity. Gender identity, as described above, suggests that individuals have considerable latitude in defining what femininity or masculinity might entail for them, rather than mere adherence to prescribed gender roles. Thus, gender identity is more relevant to an exploration of personal or individual meanings of femininity and masculinity. Spence (1985) made a cntical point in suggesting that "[w]hat constitutes an adequate amount of gender-relevant qualities for a given individual is determined by a complex calculus operating below the level of conscious awareness" (p. 83). Individuals' definitions of femininity and masculinity not only vary from person to person, but also differ when individuals are assessing themselves as opposed to others (Spence, 1985). Spence contended that, on some level, people strive to keep their sense of femininity or masculinity intact, using characteristics they attach to gender and that they possess to conflrm their gender identity. Spence (1985) argued that most men and women appear to be more secure than not in their gender identity most of the time. However, developmental tasks (e.g., adolescence) or life events (e.g., divorce) may result in stresses that cause people to periodically doubt their femininity or masculinity and struggle to reafflrm it (Spence, 1985). In addition to developmental or life events, Spence (1985) argued that issues more central to one's identity (e.g., identity as a lesbian or gay man) can be related in various ways to one's sense of femininity or masculinity. For example, the belief held by some heterosexual individuals that lesbians and gay men cannot be "real" women and men reflects the importance that some heterosexual men and women place on sexual orientation in evaluating their own and others' masculinity or femininity. In contrast, however, many lesbians and gay men clearly deflne their femininity and masculinity independently of sexual orientation and feel confident as females and males for many other reasons (Spence, 1985).

Hoffman et al.


481

From Spence's discussion we can extrapolate that a sense of confidence in relation to gender, or a sense of adequacy as a man or woman, is tied to one's gender identity, and, hence, to one's personal sense of femininity or masculinity. Likewise, Lewin (1984b) contended that one's sense of one's own femininity or masculinity was the key to its measurement, arguing that one's "perceived self-image [in relation to gender] will be more compelling than [one's] perceived traits or interests [that have typically been associated with being female or male]" (p. 200). Gender self-concept, then, encompasses gender identity. Rather than looking at the extent to which the characteristics of an individual correspond to traditional sex roles, more worthwhile questions might be: What does it mean to a particular woman to be a woman? How does she define her femininity (femaleness)? Does she conceptualize her femininity in stereotypical or unique ways? Is it important to her? Does she regard herself positively as a woman? What does it mean to a particular man to be a man? How does this man define his masculinity (maleness)? Does he conceptualize his masculinity in stereotypical or unique ways? Is it important to him? Does he regard himself positively as a man? The answers to these questions would seem to reflect one's gender self-concept. Like self-concept in general, gender self-concept is multifaceted. Confldence in self has been identifled as one important aspect of selfconcept (Basow, 1992; Hattie, 1992), and, consequently, of gender selfconcept. According to Lewin (1984b), "MF [masculinity and femininity] tests should assess gender self-confidence " (p. 200). Like Spence, Lewin argued that the focus should be on measuring "individuals' beliefs that they are, or are not, living up to various aspects of ihcir personal genderrelevant self-concepts [italics added]. Do they feel competent as members of their own sex? Are they meeting their own [italics added] standards for masculinity and femininity?" (p. 200). It would seem most appropriate, then, to investigate gender self-confidence as one aspect of gender identity, and, hence, of gender self-concept, as a means toward better understanding femininity and masculinity (see Figure 1 for proposed model of Gender Self-confidence). Gender self-confidence, then, is deflned as the intensity of one's belief that she/he meets her/his personal standards for femininity/masculinity. Gender self-confldence is one aspect of one's gender identity. As already described, gender identity is defined as "a basic, existential conviction that one is male or female" (Spence & Sawin. 1985, p. 59). It is a secure sense or conviction of one's own maleness or femaleness (cf. Green, 1974; Money, 1994). Gender identity is the "individual's awareness of and satisfaction with being a male or female" (Pleck, 1984, p. 220). Gender identity is one aspect of gender self-concept.

482


Gender Self-Concept Gender Identity I Gender Self-Confidence I

FIGURE 1 Proposed Model of Gender Self-Confldence

Gender self-concept is defined as one's perception of self as a man or a woman. It is broader than gender identity in that one perceives oneself as male or female whether or not one has a secure sense or conviction of one's maleness or femaleness. Gender self-concept reflects that which is personally relevant to the individual about being male or female. One's gender self-concept may or may not include a strong gender identity. In turn, one's gender identity may or may not include much gender self-confldence. The American Heritage Dictionary (1982) defines feminine as "[o]f or belonging to the female sex" (p. 496) and femininity as "[tjhe quality or condition of being feminine; womanliness" (p. 496). Masculine is defined as [o]f or pertaining to men or boys; male" (p. 769). These deflnitions are consistent with Spence's view that masculinity and femininity can be conceptualized as gender identity, or as she argued, as maleness and femaleness, and with Lewin's (1984b), Ashmore's (1990), and Blanchard-Fields et al.'s (1994) perspectives that each individual must be allowed the latitude to determine what these concepts mean to him or her. It is difflcult for many people, however, to focus on anything but the stereotypical sex-role deflnitions of femininity and masculinity, OT feminine and masculine, when they hear these words. Indeed, some readers of this article have had difficulty conceptualizing masculinity and femininity as possibly different from sex-role stereotypes and have erroneously concluded that we are suggesting that Spence supported an

Hoffman etal.


483

essentialist view of women and men, (i.e., that women, as a group, have personality characteristics that are innate, and that those characteristics are different from those that men innately possess) and further concluded that we are upholding that perspective. On the contrary, we argue for one's freedom to form one's own definitions of these constructs. Yes, gender is socially constructed; however, it does not follow that all definitions of femininity and masculinity are based on sex-role stereotypes. Despite the literature that supports exploring femininity and mascuhnity as aspects of the self-concepts of women and men, respectively, and the identification of gender self-confidence as key to the measurement of femininity and masculinity, no such approach has been taken previously. Lewin (personal communication) noted that this is a difflcult task. The intent of the studies that follow was to begin an exploration of gender self-confldence as a step toward a broader understanding of femininity and masculinity. The investigation of gender self-confidence requires a tool to measure it. Because this construct had not yet been defined empirically, it was necessary to develop an instrument that could assess it. Scale development was conducted in two steps: first, a preliminary assessment tool was developed and pilot-tested with a college undergraduate population (n= 146); then, a refined instrument was administered two months later to a second group of college undergraduates (n = 371). The initial development, pilot-testing, and revision of the scale is reported in Study 1. Study 2 reports results of administration of the revised scale and construct validity as well as discriminant validity data. STUDY 1: DEVELOPMENT OF THE HOFFMAN GENDER SCALE (HGS) Method An initial set of items to assess gender self-confldence was developed by posing open-ended questions to individuals about their notions of gender self-confidence, and then sorting their responses into topical categories (Crocker & Algina, 1986). Graduate students in a test development class asked various individuals to provide phrases or sentences that they felt represented confidence in oneself as a female or male. Generally, responses could be classified as either cognitive or affective, and were related to security, self-regard, and identity as a member of one's biological sex. After sorting, editing, and removing redundant items, an item review was conducted in which qualified colleagues and

484


members of a graduate class in test development were asked to informally assess the items for wording, accuracy, ambiguity, and other technical flaws. The resulting instrument consisted of 20 statements to which respondents were asked to indicate their level of agreement according to a 6-point format, ranging from " 1 , " corresponding to Strongly Disagree, to "6," corresponding to Strongly Agree. Labels also were assigned to the intermediary categories (2 = Disagree: 3 = Somewhat Agree: 4 = Tend to Agree: 5 = Moderately A^ree). These labels were selected in an attempt to increase reliable variance, to maximize the probability that they would form an equal-interval scale, and to address the likelihood that respondents' self-ratings are more positive than negative. Specifically, four positive and two negative anchors were used to ensure that the variance was most reliable across all points in the scale, particularly where the majority of respondents might rate themselves (see Klockars & Yamagishi, 1988; Lam & Stevens, 1994 for support of this approach). Items 5 and 10 were worded negatively; therefore, they required reverse-scoring. Two forms of the instrument were used in order to improve readability based on sex of respondent: Form A was worded for a female audience and Form B was written for male respondents. Other than substituting male fox female and masculinity for femininity, all items

were identical between the two forms. Following the 20 items was the question: "What do vow mean by femininity (or masculinity)?" Despite the fact that previous research suggests that this is a difficult question to answer, we wanted to provide an opportunity for respondents to think about how they viewed these terms. This question also was included as a means to explore at a later date the idiosyncratic meanings of the constructs that the respondents were self-assessing. [Our intent in asking female respondents to focus on their notions of femininity/femaleness (rather than femininity and masculinity) in relation to self and asking male respondents to focus on their notions of masculinity/maleness (rather than masculinity and femininity) in relation to self, was to be congruent with Spence's perspective (1985, 1999; Spence & Buckner, 1995, 2000) that masculinity and femininity be conceptualized as gender identity, or "sense of one's maleness [or] femaleness that parallel's awareness and acceptance of one's biological sex" (p. 91).] The instrument was described as the Hoffman Gender Scale (HGS) (Table 1) in order to reduce the emphasis on the self-confidence component. Participants The original version of the HGS was administered by the flrst author to students enrolled in seven undergraduate classes in the counseling

Hoffman etal.

TABLE 1


485

Hoffman Gender Scale (Form A) (Pilot Version of Instrument)

Please indicate your level ol agreement with each of the follovMng statements by rating it a " 1 , " "2," "3," "4," "5," or "6" as follows: J Strongly Disgree

2 Disagree

3 Somewhat Agree

4 5 Tend to Moderately Agree Agree

1. When I think of myself as a female. I really feel good. 2. I am confident in my femininity. 3. I feel competent as a member of my sex. 4. My self-image is feminine. 5. I have doubts about my femininity. 6. I meet my personal standards for femininity. 7. My perception of myself is positively associated with my biological sex. 8. I am proud to be a female. 9. I am secure in my femininity. 10. Being a female makes me uneasy. 11. I think of myself as feminine. 12. I define myself largely in terms of my femininity. 13. I accept myself as a member of my biological sex. 14. My identity is strongly tied to my femininity. 15. I have a high regard for myself as a female. 16. I am happy with myself as a female. 17. I am very comfortable being a female. 18. My sense of myself as a female is positive. 19. Being a female contributes a great deal to my sense of confidence. 20. Femininity is an important aspect of my self-concept. What do you mean by femininity?

6 Strongly Agree

486


TABLE 2

Descriptive Statistics by Sex of Respondent and Factor Loadings (Pilot Study Data) Female.s

Mean 1. FEELGOOD 5.46 2. CONFID 5.51 3. COM PET 4. IMAGE 5. DOUBTS 6. STAND 7. BIOLOGY 8. PROUD 9. SECURE 10. UNEASY

5.54 5.46 5.18 5.45 5.20 5.77 5.62 5.29

11. MEFEMALE 5.39 12. DEFINE 4.58 13. ACCEPT 5.77 14. IDENTITY 4.81 15. REGARD 5.50 16. HAPPY 5.64 17. COMFORT 5.76 18. SENSE 5.66 19. CONTRIB 4.84 20. SELFCONP 5.05

Males

ItemTotal CorreSD lations .70 .93 .87 .81 .90 .44 .94 .78

1.38 .95 1.19 .73 .80 1.28 .91 1.32 .67 1.28 .98 .91 .73 .82 1.16 .98

.46 .77 .48

.68 .79 .79 .74 .75 .55 .67 .73 .68 .73 .69 .55 .67

Mean

5.44 5.56 5.59 5.24 4.94 5.56 5.26 5.59 5.67 5.30 5.40 4.70 5.83 4.70 5.43 5.81 5.80 5.74 4.81 4.89

SD .97 .72 .80 1.12 1.57 .84 1.17

.90 .67 1.18 1.14 1.48 .54 1.43 1.04 .55 .60 .68 1.47 1.37

ItemTotal Correlations

.79 .68 .63 .76 .45 .36 .61 .78 .52 .76 .88 .81 .46 .82 .80 .60 .66 .72 .70 .79

Factor Loadings

.72 .78 .44 .74 .44 .70 .45 .73 .72 .50 .78 .69 .46 .65 .79 .79 .87 .86 .59 .66

= 146 (Females = 92: Males = 54)

department of a moderately-sized university in the southeastern United States. Although participation was voluntary, none of the students declined. Respondents included 92 women and 54 men. Results For ease of statistical analysis, each of the 20 items was described using a word or an abbreviation (see Table 2). Means, standard deviations, item-total correlations, and factor loadings are provided in Table 2,

Hoffman et al.


487

Biology Identity Define Contrib Selfconp

Compet

Accept

Uneasy Doubts Feelgood Image Proud

Mefennale

Secure Confid Regard Stand Happy Comfort Sense

FIGURE 2 Plot of a Factor Analysis of Hoffman Gender Scale Items (Pilot Sample)

Coefficient alpha, computed to assess internal consistency of the instrument, was .94 for the female sample and .94 for the male sample also. There were no overall mean differences between male and female respondents across the 20 items, [Multivariate F{20, 125) = .96, p = .52].

A maximum likelihood factor analysis clearly indicated the presence of one factor, which explained 5 0 ^ of the total variance. An inspection of the loadings on the second unrotated factor revealed no high loadings. Further scrutiny of this single factor revealed that although one factor was evident, a continuum ranging from gender selfacceptance to gender self-deflnition appeared to exist (see Figure 2). As indicated by Figure 2 (also see Tables 1 and 2 for clarification of items).

488


the top flve items seemed to represent a construct that might be best described as gender self-definition: the bottom seven seemed to represent a construct that could be identified as gender self-acceptance. Thus, although a single factor (i.e., gender self-confidence) was identified based on its explaining so much of the variance, this factor did not appear to be unidimensional (Hattie. 1984. 1985). Discussion This part of the scale de\ elopment process resulted in preliminary support for the use of the HGS with three modifications that would ensure a more sophisticated instrument. First, the existing scale was modified to increase discrimination between certain categories (e.g., replace Moderately Agree as the second anchor with Agree). Second, some test items that converged around the middle of the continuum [i.e.. items 3 (Compet). 13 (Accept). 10(Uneasy), 5 (Doubts). 1 (Feelgood). 4 (Image). 8 (Proud), and 11 (Mefemale)] were eliminated to increase the probability of obtaining two separate factors (see Figure 2). As shown in Figure 2, the plot of the factor indicated a single hyperplane clearly marked by two ends, providing cause to explore a two-factor structure in a revised scale. Third, two additional test items were added to the upper end of the continuum to strengthen the "definition" construct of the gender selfcontldence factor and to provide an equal number of items at both ends of the original continuum. The test items that were added to the revised scale are: (1) "When I am asked to describe mvself. beins female/male is one of the first things I think o f and (9) "Being a female/male is a critical part of how I vievv myself" [See Appendix A for Form A (Female) and Appendix B for Form B (Male) of the revised HGS.] Finally, to provide added insurance that female and male respondents would complete the appropriate form of the instrument (although no such problem had occurred), directions to this effect were clarified at the top of the instrument. (See Appendices A and B.) STUDY 2: INITIAL VALIDATION OF THE REVISED HGS As described in the previous study, the pilot HGS was revised based upon an interest in a possible two-factor structure. The two ends of the continuum that described the one factor that was evident in Study 1 were identitled as gender self-acceptance and gender self-deflnition. Gender self-definition refers to how strong a component of one's identity one considers one's self-defined temininity or masculinity to be. Individuals with ver) strong gender self-detlnition attach a great deal of imponance

Hoffman et al.


489

to their femaleness or maleness. Gender self-acceptance refers to how comfortable an individual is as a member of his or her gender. Individuals with strong gender self-acceptance view themselves positively as females and as males, but do not necessarily view their gender as a critical component of their identity. In addition to exploring the possibility that the revised HGS might have a two-factor structure, we also wanted to assess construct validity by examining any relationship between participants* HGS scores and their scores on the Bem Sex-Role Inventory (BSRI; Bem, 1974). The BSRI has been described as the most widely used masculinity and femininity measure (Beere, 1990). Although gender self-deflnition and gender self-acceptance may be influenced by cultural stereotypes, they are conceptualized as different from adherence/nonadherence to sextyped behavior; thus, no relationship between HGS subscale scores and BSRI classification was predicted. Method Participants The modified HGS was administered to 273 women and 98 men in a variety of undergraduate classes in the departments of human development, public health, and counseling as well as student-athlete study groups at the same university as in Study 1. This sample was designed to obtain a cross-section of undergraduate students representing a diversity of backgrounds and interests. A power analysis determined that a sample size of 198 was sufficient for medium effects and to ensure a power of .80(Cohen, 1977). Age of respondents ranged from 17 to 46 (M = 20.45, SD = 4.12, median =19). Most of the participants were White/Caucasian (n = 244). A total of 91 identified as African American, Hispanic, Native American, Asian, and Other. (Because the numbers in each individual ethnic minority category were inadequate for analysis, data were combined.) Ethnic identity was not reported by 36 participants. There were 132 first-year students, 84 sophomores, 70 juniors, and 49 seniors. Measures Bem Sex-Role Inventory (BSRl). The BSRI (Bem, 1974) consists of 60 personality characteristics on which respondents are asked to rate themselves using a 7-point scale (1 = Never or almost never true: 1 = Always or almost always true). Twenty of the characteristics are stereotypically feminine (e.g., affectionate, gentle, sympathetic), 20 are stereotypically masculine (e.g., independent, forceful, dominant), and 20 are considered filler items by virtue of their gender neutrality (e.g., conscientious, conceited, truthful). Unlike the ''feminine" items and the

490


"masculine" items, which were all identified as socially desirable for their respective sex, 10 of the "gender-neutral" items were identified as desirable for both sexes (e.g., adaptable, sincere) and the other 10 as undesirable for both sexes (e.g., inefficient, jealous). These 20 items were used to comprise a measure of Social Desirability in response. Standard scoring of the BSRI involves the use of a median-split classification system whereby four distinct quadrants are formed. Individuals are classified as androgynous if they rate higher than the median on both dimensions (masculinity and femininity), undifferentiated if they rate lower than the median on both dimensions,/