Homophobic Attitudes and Behaviours: Telling Which Teaching Strategies Make a Difference Paul Van de Ven, Laurel Bornholt and Michael Bailey School of Educational Psychology, Measurement and Technology University of Sydney Abstract Comprehensive, reliable and valid measurement of prejudicial attitudes and behaviours toward lesbians and gay males is essential to the effective evaluation of outcomes of recently implemented initiatives for reducing homophobia. A sample of undergraduate (N = 97) and high school (N = 40) students completed three instruments for measuring cognitive (Modified Attitudes Toward Homosexuality Scale: Price, 1987), affective (Affective Reactions to Homosexuality Scale: after Ernulf & Innala, 1987) and behavioural (Homophobic Behaviour of Students Scale: newly developed) reactions to homosexuals. All scales were found to be reliable (internally consistent). The undergraduate students’ responses confirmed the three-factor structure (Homophobic Guilt, Homophobic Anger, Delight) of affects in this domain, and supported the concurrent validity of the behavioural measure. Predictive validity of the behavioural scale was demonstrated by findings of a high school study which showed that students subsequently acted in accordance with their responses to this paper-and-pencil test. The fit of the overall measurement model was slightly less than optimal. However, the strategy of using the three scales to tell which teaching procedures make a difference in attempts to improve understanding and acceptance of homosexuality was judged to be tenable. Homophobic Attitudes and Behaviours: Telling Which Teaching Strategies Make a Difference Paul Van de Ven, Laurel Bornholt and Michael Bailey School of Educational Psychology, Measurement and Technology University of Sydney What we studied and why Homophobia, or anti-homosexual prejudice, has serious ramifications in educational settings and the broader community. The repercussions range from increasingly bellicose victimisation and vilification of gay and lesbian communities to insidious effects against which few are immune. Lesbian and gay youth in schools experience direct maltreatment in the form of verbal or physical abuse (Gonsiorek, 1988; Grayson, 1987; Martin and Hetrick, 1988; Rofes, 1989). As well, they are likely to encounter indirect discrimination from schools which ignore gay and lesbian issues and fail to provide positive images and role models of lesbians and gay men (Clift, 1988; Squirrell, 1989). Entrenched societal and institutional homophobia disproportionately results in lesbian and gay students’ poor academic achievement, dropping out of school, and engaging in self-destructive behaviours of promiscuity, teenage pregnancy, substance abuse and suicide (D’Augelli, 1992; Gonsiorek, 1988; Grayson, 1987; Martin and Hetrick, 1988; Rofes, 1989; Sears, 1991). For all young people, overcoming homophobia is particularly urgent in the context of AIDS education campaigns, for unless they and their educators feel comfortable in dealing with all of the issues, including sexual variations, promotion of safe sex messages will be unsuccessful (Croteau and Morgan, 1989; Rofes, 1989; Squirrell, 1989). For society as a whole, homophobia contributes to the perpetuation of traditional masculine and feminine stereotypes and gender roles (Bhugra, 1987; Forstein, 1988; Gould, 1985), thereby circumscribing male and female areas of interest (Bhugra, 1987; Grayson, 1987), emotional expressions (Wells, 1991) and the attainment of full, intimate relationships (Bhugra, 1987; Gould, 1985). The profound and diverse consequences of homophobia have been recognised by various education authorities that have developed curriculum materials which aim to improve understanding of homosexuality and reduce homophobic attitudes and behaviours. In New South Wales, for example, the Board of Studies has included units of work on homosexuality in the Year 7 to 12 Personal Development, Health and Physical Education syllabuses (e.g., Board of Studies NSW, 1991). Similarly, the Department of School Education has developed for students in Year 9 to 12 classes a six-lesson module of work for improving attitudes and behaviours toward lesbians and gay males (included in Resources for Teaching Against Violence, NSW Department of School Education, 1991). Now that various teaching materials for attenuating homophobia are available to schools, it is imperative to determine the outcomes of such efforts. The studies presented here were designed to determine an appropriate measurement model for homophobic attitudes and behaviours so that researchers and practitioners alike would be in a position to tell which strategies make a difference. Past assessment of homophobia has focussed on cognitive responses of knowledge, beliefs and thoughts about homosexuals. It has been established for some time that cognitive evaluations of homosexuals yield one large Condemnation-Tolerance factor that accounts for most of the variance (Herek, 1984; Kurdek, 1988). However, measurement that taps the complete range of homophobic reactions is essential (Britton, 1990; Herek & Berrill, 1990; Plasek & Allard, 1984). Existing attitude scales such as Price’s (1982) Modified Attitudes Toward Homosexuality Scale (hereafter MATHS) reliably and validly map the cognitive domain of homophobia. However, cognitive measures fail to account for affective or behavioural variables which are an essential part of a tripartite approach to evaluative responses (Eagly & Chaiken, 1993). A more thorough examination of the affective domain is critical to measurement and understanding of homophobia (Bhugra, 1987; Croteau and Kusek, 1992; Ernulf & Innala, 1987; Holtzen & Agresti, 1990; Plasek & Allard, 1984). Ernulf and Innala (1987) have contributed an ‘affective adjective checklist’ that remains the only pencil-and-paper test specifically designed to measure homophobic affects. The checklist contained anecdotes with homosexual content to which individuals responded to 15 value- laden adjectives on a four-point scale of ‘not at all’ to ‘very much’. Validation of this instrument was partly achieved in cross-national replications in Sweden and the United States (Innala & Ernulf, 1992). In these replications, variants of the original checklist were employed to corroborate earlier findings (Ernulf & Innala, 1987) of two affective components of homophobic reactions to a homosexual stimulus: Homophobic Guilt (comprised of Embarrassed, Guilty, Ashamed, Frightened, Awkward) and Homophobic Anger (Angry, Disdainful, Disgusted). These two factors were complemented by a third factor, Delight (Contented, Satisfied, Encouraged, Enlightened, Happy, Proud), which drew together pro-gay and -lesbian affects. The fifteenth adjective, Compassionate, was problematical in that it loaded moderately on both Homophobic Anger and Delight. The present studies contributed to the development of the ‘affective adjective checklist’ by examining the effects of two alternative adjectives and by submitting new data to confirmatory factor analysis. Accepting (hypothesised to load on the Delight factor only) was substituted for the ambiguous Compassionate which may have evoked mixed emotions of pity and acceptance. Also, Despising was used in place of the uncommon Disdainful. The revised checklist was styled Affective Reactions to Homosexuality Scale (hereafter ARHS). Behavioural responses to homosexuality are also an important and neglected area of research (Buhrke, Ben-Ezra, Hurley & Ruprecht, 1992; Croteau & Kusek, 1992; Plasek & Allard, 1984). Direct behavioural measures of, for example, chair placement or interpersonal distance have occasionally been used (Morin & Garfinkle, 1978), but what is surely needed is a broader, practical strategy for assessing behavioural reactions to homosexuality which can be used in pretest-posttest studies with students. To this end, the Homophobic Behaviour of Students Scale (hereafter HBSS) was developed and evaluated. The HBSS is a 10-item questionnaire, with five-point Likert gradations, designed to gauge students’ behavioural intentions toward gay males and lesbians, in the context of teaching about homosexuality. The scale has the advantages of being practical, plausible and inoffensive, of being a measure of commitment to action rather than a cognitive or affective response to ideas, and of being able to be presented on multiple occasions. Six items of the HBSS were designed to measure, across classroom and social situations, the extent to which participants pursued or avoided contact with gay males and lesbians (e.g., Item 2: I would speak individually, in class, with a gay person or lesbian about homosexual issues). Remaining items assessed willingness to act in support of gay and lesbian rights (e.g., Item 8: I would not sign my name to a petition asking the government to make sure gays and lesbians have equal rights with everybody else). Despite increased awareness of the serious negative consequences of homophobia, little educational research attention has been devoted to the problem. In order to address this neglect, as well as to attempt to overcome the measurement deficiencies outlined above, two studies were undertaken to determine an appropriate measurement model of homophobia. The research sought to establish the reliability of the MATHS, ARHS and HBSS; to confirm the hypothesised structure of the ARHS; to validate the HBSS; and to assess the adequacy of the overall measurement strategy. How we went about the research The three scales (MATHS, ARHS, HBSS) were administered to two groups of subjects whose ages spanned the critical range for homophobic attitudes and behaviours: undergraduate arts and education students (N = 97; 26 male, 71 female; age range 18 to 44 years, EMBED Equation = 21.3 years, sd = 3.98), and students in two Year 9 classes at a coeducational high school (N = 40; 24 male, 16 female; age range 14 to 15 years, EMBED Equation = 15.1 years, sd = 0.44). Data from both groups were used to compute Cronbach alpha reliability coefficients for each scale. To determine the predictive validity of the HBSS, the high school students were subsequently provided with opportunities to participate, or abstain from participation, in each of the ten activities included in the scale. It was hypothesised that the scale would have predictive validity if the HBSS item means of Participants were lower (less behaviourally homophobic) than those of the Abstainers. Undergraduate responses were used for three additional purposes: 1. To assess the concurrent validity of the HBSS by examining the intercorrelations among the scales. Concurrent validity would be demonstrated if the HBSS had positive correlations with the MATHS, Homophobic Guilt and Homophobic Anger variables, and a negative correlation with Delight. 2. To test the hypothesis that the affective domain of homophobic reactions is a multidimensional construct consisting of three factors: Homophobic Guilt, Homophobic Anger and Delight. 3. To examine the adequacy of the overall measurement model comprised of five factors: Cognition, Homophobic Guilt, Homophobic Anger, Delight and Behaviour. This model is presented schematically as Figure 1. What we discovered Reliability. Descriptive information (means, standard deviations, internal consistencies) about all scales is presented in Table 1. All three domains of homophobic response could be measured reliably. The MATHS and each of the sub-scales of the ARHS were highly reliable, and likewise, the newly developed HBSS had sound reliability. Correlations among Variables. Homophobic cognition as measured by the MATHS was strongly correlated with one of the homophobic affects, Anger (r = .78), and moderately so with the other, Guilt (r = .51). Homophobic Anger and Homophobic Guilt were themselves strongly correlated (r = .70). The positive affect Delight had a moderate negative correlation with both the MATHS (r = -.61) and Homophobic Anger (r = -.55), but only a weak negative association with Homophobic Guilt (r = -.26). EMBED CDraw \s \* mergeformat Figure 1. Hypothesised structure of five-factor model of homophobia. Concurrent Validity SYMBOL 190 \f "Symbol" HBSS. Correlations between the HBSS and the MATHS and ARHS variables supported the concurrent validity interpretations of the HBSS. Homophobic Behaviour was strongly positively correlated with Homophobic Cognition (r = .78); less so with the Homophobic Affects, Anger (r = .66) and Guilt (r = .38). As anticipated, Homophobic Behaviour was negatively correlated with the affective variable Delight (r = -.56). Table 1 Means, Standard Deviations and Internal Consistencies of the Modified Attitudes Toward Homosexuality Scale (MATHS), Affective Reactions to Homosexuality Scale (ARHS) and Homophobic Behaviour of Students Scale (HBSS) for High School (N = 40) and Undergraduate (N = 97) Students High School Students Undergraduate Students Scalea EMBED Equation (sd) Cronbach Alpha EMBED Equation (sd) Cronbach Alpha MATHS 160.1 (50.6) .94 120.3 (50.3) .96 ARHSb Homophobic Guilt 48.6 (17.7) .90 40.6 (16.0) .93 Homophobic Anger 45.8 (14.4) .93 35.0 (12.8) .93 Delight 99.4 (17.6) .94 92.3 (17.2) .93 HBSS 36.1 1(9.7) .86 25.1 1(8.0) .81 a Ranges: MATHS 30–270, Homophobic Guilt 20–80, Homophobic Anger 16–64, Delight 28–112, HBSS 10–50. b Based on the final version of the ARHS following confirmatory factor analyses. Predictive Validity SYMBOL 190 \f "Symbol" HBSS. The predictive validity of the HBSS was supported by findings that the mean scores on individual items of Participants and Abstainers were in the expected direction. That is, the group that actually participated in each of the activities had lower HBSS means than the group that refused to participate. Results of ten ‘independent’ t-tests of these data are presented in Table 2. Applying the Bonferroni principle to these planned comparisons, at an experimentwise error rate of .05, alpha level of .005 should be set for each comparison. At this level, six of the ten items were successful predictors. The best predictors were the four items which required participants to act in support of gay and lesbian rights by signing petitions. Only Items 1, 4, 5 and 6 failed to attain statistical significance. Confirmatory Factor Analysis SYMBOL 190 \f "Symbol" ARHS. Using exploratory factor analysis results from studies by Innala and Ernulf (1992), a three-factor model (IV below) was hypothesised. To check whether this model provided the best fit of the data, and to ensure a three-factor solution was indeed warranted, the hypothesised model was tested against theoretically plausible two- and three-factor rival models: Table 2 Comparisons of Homophobic Behaviour of Students Scale (HBSS) Single-Item Means for High School Students (N = 40) Participating and Abstaining from Participation in Activities Abstainers Participants T-Tests Item n EMBED Equation (sd) n EMBED Equation (sd) t p 1. Small class group 23 4.17 (1.19) 9 2.67 (1.58) 2.93 <.01 2. Individually 23 4.57 (0.99) 9 2.78 (1.48) 3.98 <.005 3. Class address 18 3.83 (1.20) 17 2.06 (1.56) 3.78 <.005 4. Lunchtime meeting 18 4.39 (1.20) 18 3.94 (1.26) 1.09 NS 5. Lunchtime BBQ 18 4.06 (1.43) 18 2.89 (1.68) 2.24 <.05 6. Video tape 9 3.78 (1.48) 24 2.50 (1.41) 2.28 <.05 7. Stop violence petition 23 4.22 (1.00) 17 1.59 (0.80) 8.95 <.005 8. Equal rights petition 25 4.16 (1.11) 15 1.47 (0.74) 8.35 <.005 9. Partnership petition 36 4.36 (0.96) 4 1.25 (0.50) 6.33 <.005 10. Adoption petition 38 4.47 (0.83) 2 1.00 (0.00) 5.85 <.005 I. Two factors with the Homophobic Guilt and Homophobic Anger items comprising a Rejection factor; Delight items (including Accepting) comprising an Acceptance factor. II. Two factors (as for I) with Accepting free to load on both the Rejection and Acceptance factors. III. Three factors (Homophobic Guilt, Homophobic Anger, Delight) with paths to Accepting fixed at zero. Table 3 Summary of Goodness-of-Fit Data for Models Evaluated by Confirmatory Factor Analyses Based on Undergraduate Student (N = 97) Responses Model df SYMBOL 67 \f "Symbol" 2 SYMBOL 67 \f "Symbol" 2/df GFI AGFI RMSR I 89 293.36 3.30 .707 .605 .117 II 88 271.07 3.08 .726 .627 .101 III 88 308.85 3.51 .729 .630 .187 IV 87 248.75 2.86 .748 .652 .107 V 86 223.28 2.60 .779 .691 .090 Abbreviations: GFI = Goodness of fit Index; AGFI = Adjusted Goodness of Fit Index; RMSR = Root Mean Square Residual IV. Three factors (as for III) with Accepting retained and loaded on the Delight factor only. V. Three factors (as for III) with Accepting free to load on both the Homophobic Anger and Delight factors. The goodness-of-fit data from the confirmatory factor analyses are presented in Table 3. The major findings are summarised as: 1. Model III in which paths to Accepting were fixed at zero yielded the poorest fit of all the models. 2. Models in which Accepting was cross-loaded on the Rejection (II) and Homophobic Anger (V) factors provided a better fit than corresponding models (I and IV respectively) with only one path to Accepting. 3. The three-factor models (IV and V) yielded a closer fit than either of the two-factor models (I and II). 4. Of all the models, Model V accounted for the greatest proportion of covariation among the scale items, although the overall fit was less than optimal. When compared with the hypothesised model (IV), the optimal model (V) yielded a SYMBOL 67 \f "Symbol" 2 reduction of 25.47 at the cost of only one degree of freedom; a highly significant improvement in goodness- of-fit. Adequacy of the Overall Measurement Model. The value of the coefficient of determination (1.00) from LISREL output confirmed that the reliability of the measurement model as a whole was exceptionally high. The adequacy of the model, though, in representing the observed data would bear some improvement. This statement is based on the value of the Goodness of Fit Index (0.515) and the Adjusted Goodness of Fit Index (0.474). The Root Mean Square Residual (0.083) also indicated some degree of misfit in the model. If, on the other hand, we assessed the hypothesised model based on subjective indices of fit, the picture was much improved. For example, SYMBOL 67 \f "Symbol" 2/df ratio was 1.95, representing an acceptable fit to the data. Implications of our results Most previous measurements of homophobia were inadequate because they relied on a unidimensional construct of attitudes toward homosexuals, namely cognition of homosexuality. The cognitive domain is an important one in the sphere of anti-homosexual prejudice, but it is only one part of the beliefs-feelings- actions triptych that is homophobia. Thorough assessment of homophobia entails the evaluation of affective and behavioural reactions as well as cognitive responses. To this end, the present studies provide a comprehensive yet practical strategy for determining students’ levels of homophobia. The MATHS, ARHS and HBSS individually and in combination are highly reliable. The validity of the newly developed HBSS was substantiated by the results of these studies. Correlations between the HBSS and the scales of negative and positive cognition and affects supported the concurrent validity of the HBSS. The predictive validity of the HBSS was verified by the results of the individual item analyses based on high school students’ participation or non-participation in each HBSS activity. Only four items failed, in terms of statistical significance, to predict subsequent behaviour. Plausibly, in the case of at least two of the items (lunchtime meeting and BBQ) it may have been because these were the final activities to be conducted. By this time (eight days after questionnaire administration) all students were accustomed to seeing various gay or lesbian people visit their classrooms and may therefore have been more willing to participate than at the outset. Alternatively, discussions among students as the study progressed may have quelled initial hesitations. Were it possible to design a study in which all the activities could be presented concurrently and immediately after the questionnaire, it could confidently be expected that most, if not all, of the HBSS items would successfully predict behavioural intention at that time. On this premise, it is recommended that all 10 items of the HBSS be retained. Whereas cognition (Herek, 1984; Kurdek, 1988) and behaviour (confirmed by independent analysis) are unidimensional constructs in this domain, the goodness-of-fit data from the ARHS analyses suggest that the covariance among this scale’s items may be conceptualised as reflecting the multidimensional nature of affective responses to homosexuals. A three-factor model incorporating a cross-loading of Accepting on Homophobic Anger provides the most satisfactory mapping of homophobic affects. The substitute adjective Accepting, contrary to hypothesis, does not overcome the mixed connotations found to be associated with Compassionate (Ernulf & Innala, 1987). Further development of the ARHS may be warranted, but, for the present, all the positive items should be retained and used to score the Delight sub-scale. Not Accepting (i.e., Rejecting) makes a meaningful and substantial contribution, together with Despising, Disgusted and Angry, to Homophobic Anger. The other homophobic affect, Guilt, constitutes a separate and equally important construct in attempts to understand and account for homophobic reactions. Although it could not be specifically tested in the present study (due to the limited sample), the covariance structure reported here is expected to be invariant across genders. In previous findings (Ernulf & Innala, 1987; Innala & Ernulf, 1992), males scored higher than females, but main and interaction effects of sex of participant and sex of target on these factors were not significant. Correlations among the cognitive, affective and behavioural variables indicate the diverse and interrelated nature of responses to homosexuality. People who hold negative cognitive beliefs about homosexuals tend to feel angry, and behave in a negative way, toward gay males and lesbians. They are also apt to feel guilty and less delighted about lesbian and gay partnerships. Homophobic Anger and Homophobic Guilt are to a moderate degree reciprocal affects, but likely to be manifested in different ways. Anger is more prone than Guilt to be associated with feelings disparaging of gay and lesbian sexual orientations and relationships, and more likely to be expressed in negative behaviours toward homosexuals. Perpetrators of anti-gay and - lesbian violence may therefore tend to be individuals with feelings of anger rather than individuals with repressed negative emotions of shame and guilt. This would give the lie to popular supposition that much homophobic violence and victimisation is simply an expression of fear of latent homosexual tendencies or heightened guilt response. Prejudice against homosexuals has resulted in widespread vilification and victimisation of gay males and lesbians. Urgent interventions are called for, especially with young adults and adolescents, including high school students, who are responsible for much of the harassment and violence (Britton, 1990; Clift, 1988; Herek & Berrill, 1990). A major shortcoming of research into homophobia has been the lack of a set of scales which can reliably, validly and practically measure beyond homophobic cognition to assess affects (Bhugra, 1987; Ernulf & Innala, 1987; Holtzen & Agresti, 1990) and behaviours (Buhrke et al., 1992; Croteau & Kusek, 1992; Plasek & Allard, 1984). The outcomes of recent initiatives aimed at improving understanding and acceptance of lesbians and gay males will only be fully exposed by using multiple dependent measures of cognitive, affective and behavioural responses to homosexuality. In this regard, the present studies make a difference by providing future researchers with an enhanced capability to conduct workable research which is more theoretically comprehensive and more psychometrically rigorous than previous endeavours. References Bhugra, D. (1987). Homophobia: A review of the literature. Sexual and Marital Therapy, 2, 169-177. Board of Studies NSW. (1991). Syllabus — Years 7-10: Personal development, health and physical education. North Sydney: Author. Britton, D. M. (1990). Homophobia and homosociality: An analysis of boundary maintenance. Sociological Quarterly, 31, 423-439. Buhrke, R. A., Ben-Ezra, L. A., & Hurley, M. E. (1992). Content analysis and methodological critique of articles concerning lesbian and gay male issues in counseling journals. Journal of Counseling Psychology, 39, 91-99. Clift, S. M. (1988). Lesbian and gay issues in education: A study of the attitudes of first year students in a college of higher education. British Educational Research Journal, 14, 31- 50. Croteau, J. M., & Kusek, M. T. (1992). Gay and lesbian speaker panels: Implementation and research. Journal of Counseling and Development, 70, 396-401. Croteau, J. M., & Morgan, S. (1989). Combating homophobia in AIDS education. Journal of Counseling and Development, 68, 86- 91. D'Augelli, A. R. (1992). Lesbian and gay male undergraduates’ experiences of harassment and fear on campus. Journal of Interpersonal Violence, 7, 383-395. Eagly, A., & Chaiken, S. (1993). The psychology of attitudes. New York: Harcourt Brace Jovanovich. Ernulf, K. E., & Innala, S. M. (1987). The relationship between affective and cognitive components of homophobic reaction. Archives of Sexual Behavior, 16, 501-509. Forstein, M. (1988). Homophobia: An overview. Psychiatric Annals, 18, 33-36. Gonsiorek, J. C. (1988). Mental health issues of gay and lesbian adolescents. Journal of Adolescent Health Care, 9, 114-122. Gould, M. (1985). Teaching about men and masculinity: Method and meaning. Teaching Sociology, 12, 285-298. Grayson, D. A. (1987). Emerging equity issues related to homosexuality in education. Peabody Journal of Education, 64, 132-145 [published 1989]. Herek, G. M. (1984). Attitudes toward lesbians and gay men: A factor-analytic study. Journal of Homosexuality, 10, 39-51. Herek, G. M., & Berrill, K. T. (1990). Anti-gay violence and mental health: Setting an agenda for research. Journal of Interpersonal Violence, 5, 414-423. Holtzen, D. W., & Agresti, A. A. (1990). Parental responses to gay and lesbian children: Differences in homophobia, self- esteem, and sex-role stereotypes. Journal of Social and Clinical Psychology, 9, 390-399. Innala, S. M., & Ernulf, K. E. (1992). The relationship between affective and cognitive components of homophobic reaction: Three cross-national replications (Göteborg Psychological Reports, Vol. 22 No. 2). Göteborg, Sweden: University of Göteborg, Department of Psychology. Kurdek, L. A. (1988). Correlates of negative attitudes toward homosexuals in heterosexual college students. Sex Roles, 18, 727-738. Martin, A. D., & Hetrick, E. S. (1988). The stigmatization of the gay and lesbian adolescent. Journal of Homosexuality, 15, 163-186. Morin, S. F., & Garfinkle, E. M. (1978). Male homophobia. Journal of Social Issues, 34, 29-47. NSW Department of School Education. (1991). Resources for teaching against violence. Sydney: Author. Plasek, J. W., & Allard, J. A. (1984). Misconceptions of homophobia. Journal of Homosexuality, 10, 23-37. Price, J. H. (1982). High school students’ attitudes toward homosexuality. Journal of School Health, 52, 469-474. Rofes, E. (1989). Opening up the classroom closet: Responding to the educational needs of gay and lesbian youth. Harvard Educational Review, 59, 444-453. Sears, J. T. (1991). Helping students understand and accept sexual diversity. Educational Leadership, 49, 54-56. Squirrell, G. (1989). Teachers and issues of sexual orientation. Gender and Education, 1, 17-34. Wells, J. W. (1991). The effects of homophobia and sexism on heterosexual sexual relationships. Journal of Sex Education and Therapy, 17, 185-195.