Journal of Personality and Social Psychology

We recruited twins systematically from the Australian Twin Registry and assessed their sexual orientation and 2 related traits: childhood gender nonconformity and continuous gender identity. Men and women differed in their distributions of sexual orientation, with women more likely to have slight-to-moderate degrees of homosexual attraction, and men more likely to have high degrees of homosexual attraction. Twin concordances for nonheterosexual orientation were lower than in prior studies. Univariate analyses showed that familial factors were important for all traits, but were less successful in distinguishing genetic from shared environmental influences. Only childhood gender nonconformity was significantly heritable for both men and women. Multivariate analyses suggested that the causal architecture differed between men and women, and, for women, provided significant evidence for the importance of genetic factors to the traits' covariation. The causes of sexual orientation have provoked intense scientific interest, inspiring both empirical work and theory ( ; ; ; ; ). This interest stems, in part, from the mostly mistaken belief that different etiological accounts of sexual orientation have different social and ethical implications ( ; ; ; ). But there are also legitimate and important scientific reasons for interest in the issue. Sexual orientation is a fundamental aspect of human sexuality, guaranteeing that for the large majority, men mate with women. Furthermore, sexual orientation is empirically closely linked to some aspects of gender roles, including childhood play behavior and gender identity ( ; ; ; ) and aspects of adult sex-typed behavior as well, particularly occupational and recreational interests ( ; ). Thus, illuminating the origins of sexual orientation could also shed light on the development of other important sex differences. Bell, Weinberg, & Hammersmith, 1981 Bem, 1996a Ellis & Ames, 1987 Hamer & Copeland, 1994 LeVay, 1996 Bem, 1996b Greenberg & Bailey, 1993 Schmalz, 1993 Stein, 1994 Bailey & Zucker, 1995 Bell et al., 1981 Green, 1987 Zuger, 1988 Bailey, Finkel, Blackwelder, & Bailey, 1996 Lippa, 1998 Empirical research about the origins of sexual orientation has been organized, generally, around the nature–nurture dichotomy, motivated by two different theoretical approaches. The first, sometimes called the “neurohormonal” or “neuroendocrine” theory ( ; ), examines the possibility Ellis & Ames, 1987 Meyer-Bahlburg, 1987 8/12/02 9:51 AM Ovid: Bailey: J Pers Soc Psychol, Volume 78(3).March 2000.524–536 Page 2 of 26 http://gateway2.ovid.com:80/ovidweb.cgi that homosexual people have been subject to atypical levels of hormones in development, thus causing sexatypical neural diffentiation. LeVay's finding that for one hypothalamic nucleus, gay men are more similar to heterosexual women than to heterosexual men is perhaps the most important finding motivated by this perspective ( ). The second approach, behavioral genetics, has focused on whether sexual orientation is familial, and if so, whether familial aggregation is attributable to genetic or shared environmental factors. It is important to emphasize that the two approaches are not competing theories but represent different levels of analysis. The present study is primarily an example of the second approach, although some of the variables considered (e.g., childhood gender nonconformity) are also highly relevant to the neurohormonal approach. LeVay, 1991 Distribution of Sexual Orientation In recent years a great deal of attention has been given to the prevalence of homosexuality (e.g., ; ). Media accounts have focused on the prevalence of homosexual behavior per se. Most current researchers, including us, define sexual orientation psychologically rather than behaviorally (e.g., ; , p. 45; ). Sexual orientation is one's degree of sexual attraction to men or women. Of course, sexual orientation should be closely related to sexual experience with one sex or the other, though many factors, especially social ones, could cause sexual orientation and sexual behavior to correlate less than perfectly. Diamond, 1993 Laumann, Gagnon, Michael, & Michaels, 1994 Money, 1988 LeVay, 1996 Zucker & Bradley, 1995 Kinsey's original surveys of sexual orientation suggested a rather marked sex difference (see , pp. 47–49). For men, sexual orientation appeared to be somewhat bimodal, but for women, it tapered gradually from strictly heterosexual to strictly homosexual orientation, with no elevation at the latter end of the sexual orientation scale. If this difference is true, it would suggest that sexual orientation development and phenomenology differ between the sexes. Unfortunately, Kinsey's assessment of sexual orientation confounded behavioral and psychological measures. Furthermore, his sampling scheme was notoriously haphazard ( , pp. 44–45). LeVay, 1996 Laumann et al., 1994 Familiality and Genetics of Sexual Orientation Both male and female homosexuality appear to run in families ( ; ; ; ; ). Studies of unseparated twins have suggested that this is primarily due to genetic rather than familial environmental influences ( ; ). Furthermore, there is some evidence that male sexual orientation is influenced by a gene on the X chromosome ( ; ), though some other studies have found contradictory evidence ( ; ). Bailey & Bell, 1993 Bailey & Benishay, 1993 Pattatucci & Hamer, 1995 Pillard, 1990 Pillard & Weinrich, 1986 Bailey & Pillard, 1991 Bailey, Pillard, Neale, & Agyei, 1993 Hamer, Hu, Magnuson, Hu, & Pattatucci, 1993 Hu et al., 1995 Bailey et al., 1999 Rice, Anderson, & Ebers, 1995 Although prior twin studies have been generally consistent in indicating a genetic contribution to male and female sexual orientation, they have also been rather consistent in their methodological limitations. Most importantly, all sizable twin studies of sexual orientation recruited probands by means of advertisements in homophile publications or by word of mouth ( ). Such sampling is likely to result in volunteer bias that affects twin concordances and heritability analyses ( ), though it is difficult to estimate the direction or magnitude of the bias from available information. Furthermore, respondents with exclusively homosexual orientations may be overrepresented, and those with modest levels of homosexual attraction, underrepresented, obscuring the potentially continuous nature of sexual orientation. Bailey & Pillard, 1995 Kendler & Eaves, 1989 Correlates of Sexual Orientation 8/12/02 9:51 AM Ovid: Bailey: J Pers Soc Psychol, Volume 78(3).March 2000.524–536 Page 3 of 26 http://gateway2.ovid.com:80/ovidweb.cgi Another limitation of prior twin studies of sexual orientation has been their frequent neglect of its correlates. Specifically, a great deal of evidence supports an association between childhood gender nonconformity and sexual orientation ( ). That is, gay men tend to recall having been feminine boys, and lesbians, masculine girls. Prospective studies have validated this association for men ( ; ); prospective studies of masculine girls have not yet been conducted. The association between childhood gender nonconformity and adult sexual orientation has been so well-documented and is so strong that remarked: “It is difficult to think of other individual differences (besides IQ or sex itself) that so reliably and so strongly predict socially significant outcomes across the life span, and for both sexes, too. Surely it must be true” (p. 323). Bem, and others, have recognized that developmental and etiological theories of sexual orientation must try to account for the association between the two traits, and Bem has proposed a theory intending to do precisely that. The theory, “exotic becomes erotic,” specifies that biological factors may cause childhood gender nonconformity and that gender-nonconforming children tend to feel different from other children of their sex, and as a result (through a rather complicated pathway), eroticize them. Bem's theory hypothesizes that childhood gender nonconformity is influenced by biological factors, perhaps including genetic ones. This hypothesis has been examined using a neurohormonal approach ( ), but it has not been tested through behavior genetics. Although some prior twin studies assessed (retrospective) childhood gender nonconformity ( ; ), their sampling of homosexual probands (as opposed to sampling from a population of unspecified sexual orientation) prevented either univariate or multivariate genetic analysis of traits other than sexual orientation. Bailey & Zucker, 1995 Green, 1987 Zuger, 1988