Introduction 2007 1996 2004 1988 2006 1996 1998 2005 2004 1996 1988 2002 2006 1996 2004 1985 2003 2004 2003 2003 2003 2000 2001 1996 1998 1999 Accordingly, the aim of the present study is to determine the contribution of both genes and shared environment to individual differences in attitudes to homosexuality, using data from a large twin sample. Findings will result in further insight into the aetiology of variation in homophobia. Methodology Participants All participants were drawn from the Australian National Health and Medical Research Council Twin Registry (ATR). Subjects for this study were recruited in two phases from a large twin-family study of alcohol use and abuse. N 2000 In total, 1,907 complete pairs and 1,090 twins from incomplete pairs completed the questionnaire (1,824 male and 3,080 female individuals). However, due to missing data and ambiguous zygosities the effective study sample used for the homophobia analyses is lower than 4,904, as described in the result section. Zygosity 1990 1975 Measures 1976 1989 The structure of masculinity and femininity self ratings 1989 Accordingly, in the present study homophobia was measured using the 10 items of this Attitudes to Homosexuality scale, which comprised statements like ‘Homosexuality is obscene and vulgar’, and ‘Homosexual men should be allowed to work as schoolteachers’ (see Appendix A for the total scale). Twins were asked to either tick YES if they agreed with the statement or NO if they disagreed. If a participant filled out both answers on an item (both YES and NO), the item was coded as missing. To enhance unbiased answering, a preamble to the section explained that there are no right or wrong answers and no trick questions and they were encouraged to work quickly and not to think too long about the exact meaning of questions. 1999 N N As the distribution of the homophobia data showed significant skewness and kurtosis, scores were grouped into four ordinal categories with roughly equal sample sizes. Individuals scoring 0 on the continuous scale were given an ordinal score of 0, while individuals scoring 1 or 2 were given score 1, individuals scoring 3–6 score 2, and individuals scoring 7–10 score 3. To determine the participants’ sexual orientation, twins were asked the following question in another part of the questionnaire, called ‘Sexual Feelings and Behaviour’: Do you consider yourself to be heterosexual (straight), bisexual or homosexual (lesbian or gay)? For the present study, sexual orientation was only subdivided into heterosexual and non-heterosexual. The latter includes both bisexual and homosexual individuals. Data analysis 1989 2003 2 To determine the effects of age, sexual orientation and sex on the thresholds, different (nested) submodels are tested against a full model, in which all parameters were estimated. By looking for differences in model fit after constraining these effects to zero in a stepwise manner, the significance of parameters that contribute to explaining differences in thresholds can be determined. Heterogeneity of twin pair correlations is tested by determining the change in model fit when twin pair correlations (for sex and/or zygosity) were equated. 1978 Results Descriptive statistics 1 Table 1 Percentage of agreement (yes), disagreement (no) and missing answers on the 10 homophobia statements for males and females separately  N N Yes No Missing Yes No Missing Homosexuality is not immoral 50.9 48.2 0.9 64.0 35.1 1.0 Allow as school teachers 53.0 46.9 0.1 69.4 30.5 0.0 Allow as court judges 65.2 34.7 0.1 78.4 21.4 0.2 Allow as ministers 56.9 43.0 0.1 67.4 32.2 0.4 Allow as medical doctors 58.0 41.9 0.1 70.1 29.8 0.1 Allow as government officials 70.8 29.1 0.1 81.8 18.1 0.1 a 34.4 63.4 2.2 20.3 77.8 1.9 a 48.0 51.3 0.7 30.3 68.6 1.1 a 29.2 69.3 1.5 19.9 77.7 2.4 Allow to dance with each other in public 49.1 49.2 1.7 62.2 35.2 2.6 a 2 Table 2 Number and percentage of males and females, heterosexuals and non-heterosexuals, for each category of the ordinal homophobia scale (0 is least, 3 is most homophobic) Homophobia category 0 1 2 3 N 368 (22.6%) 314 (19.3%) 392 (24.1%) 553 (34.0%) N 77 (69.4%) 18 (16.2%) 13 (11.7%) 3 (2.7%) N a 450 (25.6%) 338 (19.2%) 408 (23.2%) 562 (32.0%) N 1,011 (36.1%) 692 (24.7%) 580 (20.7%) 519 (18.5%) N 78 (78.0%) 9 (9.0%) 10 (10.0%) 3 (3%) N a 1,098 (37.5%) 705 (24.1%) 600 (20.5%) 527 (18.0%) a 2 Hypotheses concerning differences in threshold parsimony 12 2 P 3 2 P 1 2 P 1 2 P Heterogeneity of twin pair correlations 3 1 2 P Table 3 Polychoric twin pair correlations (95% confidence intervals) for the homophobia score for each zygosity group, estimated in Mx  MZM 295 pairs DZM 182 pairs MZF 634 pairs DZF 354 pairs DZOS 357 pairs Correlation 0.61 (0.52–0.69) 0.42 (0.37–0.55) 0.51 (0.43–0.57) 0.39 (0.28–0.50) 0.29 (0.17–0.41) 2 2 P The inability to find significant differences between the MZ twin pair correlations for males and females, and between the DZ same sex and opposite sex twin pair correlations may, however, be due to a lack of power to detect qualitative and quantitative differences between the groups. 1 2 P Genetic modelling 4 1992 Table 4 Genetic model fitting: Comparison of different genetic models  Model −2ll d.f. Vs 2 Δd.f. P-value DZOS 1 General sex-limitation model 12,199.93 4672     0.09 2 Common effects sex-limitation model 12,201.36 4673 1 1.43 1 0.23 0.50 3 General ACE model 12,205.88 4675 2 4.52 2 0.10 0.50 1 2 P 2 2 P 5 Table 5 Proportions of variance (95% confidence intervals) attributable to A, C and E effects for both sexes for the general sex-limitation model, the common effects sex-limitation model and the general ACE model  males males males females females females DZOS General sex-limitation model 0.38 (0.08–0.68) 0.23 (0.00–0.49) 0.39 (0.31–0.44) 0.23 (0.00–0.49) 0.28 (0.04–0.49) 0.49 (0.43–0.57) 0.09 (0.00–1.00) Common effects sex-limitation model 0.55 (0.16–0.69) 0.07 (0.00–0.47) 0.38 (0.30–0.47) 0.20 (0.00–0.54) 0.30 (0.00–0.51) 0.50 (0.43–0.58) 0.50 General ACE model 0.36 (0.19–0.54) 0.18 (0.02–0.32) 0.46 (0.41–0.52) 0.36 (0.19–0.54) 0.18 (0.02–0.32) 0.46 (0.41–0.52) 0.50 1994 5 1 Fig. 1 a b \documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym} 
\usepackage{amsfonts} 
\usepackage{amssymb} 
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$ {\left( {D\hat{Z}OS} \right)} $$\end{document}  Discussion 1996 2004 2002 2004 The aim of the present study was to determine the contribution of genes and environment to individual differences in attitudes toward homosexuality. Based on the general ACE model, estimates of the influence of A, C and E on the variance in homophobia scores are: 36%, 18% and 46%. However, although no significant qualitative and quantitative differences in sources of variation between males and females were found, twin pair correlations do suggest possible differences between the sexes. Under a general sex-limitation model, estimates of the influence of A and C on variation in homophobia scores are 38% and 23% for males and 23% and 28% for females, respectively. The additive genetic correlation for the DZ opposite sex twins is estimated at 0.09, considerably lower than the 0.5 assumed for DZ same sex twins. This finding would suggest that genes influencing homophobia in males are to a large extent different from those in females. However, since both A and C effects are present in both sexes, the parameterisation of sex limitation as genetic in origin is arbitrary. If it were parameterised as a lower common environmental correlation between males and females, the same model fit would be obtained. Moreover, there could be sex limitation of both A and C influences, but since there is only one degree of freedom to estimate them—obtained from the DZ opposite sex twin pair correlation—the two are completely confounded. In either case, differences in saliency of homosexuality between males and females make it unsurprising that there are distinct causes of variation in homophobia in the two sexes. 1999 2007 1986 2007 1978 1 After correction, the contribution of shared environment on individual differences in homophobia scores falls from 0.18 to 0.03, while the genetic estimate increases to about 0.51. This finding suggests that variation in attitudes toward homosexuality is substantially inherited, while social environmental influences are relatively minor. When applying the assortative mating correction to the results of the general sex-limitation model, the contribution of shared environment on individual differences in homophobia scores for males falls from 0.23 to 0.04 and for females from 0.28 to 0.24, while the genetic estimate increases to about 0.57 for males and 0.27 for females. As such, based on the general sex-limitation model familial aggregation for homophobia scores for males might be almost totally accounted for by genetic effects, while for females both genes and shared environment seem to have about equal effects. If this would be the case, and there is little or no C influence on homophobia scores for males, sex limitation for homophobia scores can only have a genetic origin. P P 2000 2000 There are a few important methodological limitations to be considered for this study. As already mentioned, the most important limitation of this study was a lack of statistical power. Although the sample size is fairly large, due to the fact that the homophobia scale had to be analysed as ordinal data, it is still insufficient. Neale et al. (2004) showed that for a threshold study at least three times the sample size needed for equivalent power to a study using continuous data is required. 1993 2002 1996 1999 1998 1996 1997 Another limitation to consider is the possibility of socially desirable responding. Although it is not possible to exclude social desirability effects, twins were encouraged to be honest by guaranteeing anonymity and by asking them to fill out the questionnaire while alone with sufficient privacy. In summary, this study concerning the aetiology of homophobic attitudes revealed that familial aggregation in attitudes toward homosexuality is accounted for by genes as well as by shared environmental factors. However, when the plausible effect of assortative mating on our estimates is taken into account, familial aggregation for homophobia scores might be almost totally accounted for by genetic effects. More research is necessary to further unravel the sources of variance in homophobia and to determine whether these differ between sexes. This future research will need greater sample sizes—preferably including spouses—and should analyse attitudes toward male and female homosexuals separately. It will also be of interest to compare prejudice toward homosexuals with prejudice toward other targets (like racism and sexism), to see to what extent similar results will be obtained and whether the same genetic and environmental sources explain individual differences in prejudice toward different out-groups.