Abstract
A social dominance explanation of the well-established sex difference in anti-gay male prejudice, whereby heterosexual men (vs. women) are more negative toward gay men, is proposed. We test a mediation model whereby respondent sex (male vs. female) predicts prejudice toward gay men indirectly through social dominance orientation and sexism (itself predicted by social dominance orientation (SDO). The model is also considered for anti-lesbian prejudice. Across three samples the model is supported, with the relation between participant sex and anti-gay male prejudice mediated (explained). The model was not consistently supported for anti-lesbian prejudice. Implications are discussed.
Notes
1. Given our interest in explaining sex differences, we did not include right-wing authoritarianism (RWA, a construct representing conventionality, submission to authority and aggression toward norm violators, Altemeyer, Citation1998) in our proposed model. Nevertheless, given established associations between RWA and prejudice toward homosexuals (see Haddock & Zanna, Citation1998; Stones, Citation2006; Whitley, Citation1999) in ancillary analyses we tested samples B and C models statistically controlling for RWA (Altemeyer, Citation1996), observing results consistent with those reported. We also examined models replacing SDO with RWA, but RWA did not mediate (because sex was not associated with RWA in Sample B or C).
2. Maximum likelihood estimation was used to test all models.
3. For each sample, we also tested a single model including both prejudice toward gay men and prejudice toward lesbians as endogenous variables (allowed to correlate). Across samples, path estimates were identical to those presented in – and model fit was excellent. To test whether the key path from sex to prejudice was equivalent across prejudice targets, we constrained paths from sex to prejudice towards gay men and from sex to prejudice towards lesbians to equivalency. The Akaike Information Criterion (AIC) measure of comparative fit was larger for the constrained models in samples A and B (AIC = 44.72 and 39.04, respectively) than in the unconstrained models (40.01 and 36.11, respectively). This suggests that the unconstrained models better fit the data, and that the paths are not equivalent. For Sample C, however, the constrained model (AIC = 41.36) fitted the data better than the unconstrained model (AIC = 43.40), suggesting equivalency in Sample C. Thus, results are consistent employing this approach. For clarity, separate models are presented for prejudice toward gay men and prejudice toward lesbians.
4. We also tested an alternative model where the positions of SDO and prejudice were switched. In all samples, fit indices were identical to those reported above, and indirect effects from sex to sexism were significant. However, in all models the key path from sex to sexism remained significant (ps < 0.05). This path was not reduced to non-significance as in the original proposed model, and thus the alternative model does explain sex differences. Furthermore, in all samples, the path from sex to prejudice toward lesbians was non-significant in this model. An alternative model switching the positions of SDO and sexism produced results largely equivalent to our proposed model. We argue, however, that SDO has temporal priority over sexism, given previous theory specifying a causal chain (Sidanius & Pratto, Citation1999). Regardless, SDO and sexism consistently mediate the relationship between sex and prejudice toward gay men.
5. Notably, the indirect effect of sex on prejudice toward gay men was stronger than the indirect effect of sex on prejudice toward lesbians in samples A (z = 4.73, p < 0.001) and C (z = 1.90, p < 0.05), but not in Sample B (z = −0.85, p > 0.05).
6. We thank an anonymous reviewer for suggesting this point.
Additional information
Funding
Notes on contributors
Cara C. MacInnis
Cara C. MacInnis received her PhD from Brock University in 2013. She is currently a postdoctoral fellow at the University of Toronto. Her research interests include prejudice and discrimination, political ideology, personality and individual differences, prejudice-reduction strategies and sexuality.
Gordon Hodson
Gordon Hodson is a Professor of Psychology at Brock University. His research interests include stereotyping, prejudice and discrimination, with a focus on ideology, emotions (disgust, empathy), intergroup contact/friendship and dehumanisation.