COLUMBIA UNIVERSITY

WORKING PAPERS

OPPOSITE-SEX TWINS AND ADOLESCENT SAME-SEX ATTRACTION

Peter S. Bearman

Institute for Social and Economic Research and Policy

Columbia University

and

Hannah Brückner

Department of Sociology

Yale University

October 2001

ISERP WORKING PAPER 01-04

Data for this paper are drawn from the National Longitudinal Study of Adolescent Health (Add

Health), a program project designed by J. Richard Udry and Peter Bearman, and funded by a grant

HD31921 from the National Institute of Child Health and Human Development to the Carolina

Population Center, University of North Carolina at Chapel Hill, with cooperative funding

participation by the following agencies: The National Cancer Institute; The National Institute of

Alcohol Abuse and Alcoholism; the National Institute on Deafness and other Communication

Disorders; the National Institute on Drug Abuse; the National Institute of General Medical Sciences;

the National Institute of Mental health; the Office of AIDS Research, NIH; the Office of Director,

NIH; The National Center for Health Statistics, Centers for Disease Control and Prevention, HHS;

Office of Minority Health, Centers for Disease Control and prevention, HHS, Office of the Assistant

Secretary for Planning and Evaluation, HHS; and the National Science Foundation. We thank Ivan

Chase, Roger Gould, Michael Sobel, J. Richard Udry, Duncan Watts, and Harrison White for their

helpful comments. Authorship order is alphabetical. Address all correspondence to Peter Bearman,

Institute for Social and Economic Research and Policy, 814 IAB. 420 W 118th Street, Columbia

University, NY, NY. psb17@Columbia.edu.

Abstract

We consider social, genetic, evolutionary, and hormonal transfer hypotheses for same-sex romantic preferences of adolescent (N=5,552) sibling pairs drawn from a nationally

representative sample. We show that male but not female opposite-sex twins disproportionately report same-sex attraction; and that the pattern of concordance of same-sex preference among siblings is inconsistent with a simple genetic influence model. Our results provide substantial support for the role of social influences, reject the hormone transfer model, reject a speculative evolutionary theory, and are consistent with a general model that allows for genetic expression of same-sex attraction under specific, highly circumscribed, social conditions.

Keywords: Social influence, genetic influence, twin studies, same-sex preferences, and

socialization.

Three general frameworks compete for attention in the crowded field of understanding the etiology of human same-sex romantic attraction. The first account stresses social influences, the second genetic influences, and the third hormonal influences. The three foci seldom meet, net of those articles which, often with rhetorical flourish, call for either the elimination of one or the other approaches, or alternatively for broad integration of social and biological factors in the explanation of human behavior writ large1. In this article, we do something different: we empirically test social, genetic, evolutionary, and hormonal imbalance hypotheses for adolescent same-sex romantic preferences2. Adjudicating between these models requires unusual data structures, typically not available to researchers. We consider same-sex attraction for a large nationally representative sample of adolescent. The design allows us to test hypotheses about

genetic and intrauterine hormone transfer effects, and to consider the impact of social influence on adolescent same sex romantic attraction.

Social scientists have largely abandoned empirical work focusing on the individual determinants of same-sex attraction, in part because most studies have failed to provide evidence that supports the idea that social, psychological, or social-psychological factors play a role in shaping individual variation in sexual expression and attraction (Herdt 1996). In contrast, behavioral geneticists and biologists have recently embraced empirical studies on the etiology of same-sex romantic preference, in part because work in this tradition has appeared to be successful. Chapters on the biological causes of sexual orientation are now routinely included in textbooks on sexuality (D.Augelli and Patterson 2001; Davidson and Moore 2001, Cabaj and Stein 1996, Ellis and Ebertz 1997), and the general consensus is that .biology plays an important role in the

development of male and female sexual orientation. (Hershberger 1997:43). Oddly, despite the popularity of the idea, the evidence for genetic and/or hormonal effects on same-sex orientation is inconclusive at best. The most publicized genetic findings, for example, the discovery of a marker for homosexuality in men, (Hamer et al. 1993) has not been replicated, and studies purporting to establish a genetic or hormonal foundation to human sexual orientation tend to have serious methodological flaws (Stein 1999; Byne 1995; McGuire 1995).

If they have given up the empirical focus on individual variation in sexual orientation, social scientists have not surrendered empirical focus on cross-cultural variation in sexuality. Indeed, social scientists tend to argue that sexual preference is socially constructed, pointing towards ethnographic studies which show significant variation across time and cultures in what is regarded as homosexual behavior, who engages in it, and how this behavior is normatively regulated (Herdt 1996, Risman and Schwartz 1988, Troiden 1988). The empirical evidence for these ideas is both consistent and striking; so much so that if one could say that studies of the determinants of individual variation in sexual orientation are largely absent, there has been a veritable growth industry in studies of the constructed nature of sexuality. In contrast, biologists,

behavioral geneticists and evolutionists have had a difficult time operating empirically at the macro-level. Few non-social scientists try to make sense out of the welter of forms of sexual expression found across human societies.

Even though social scientists have basically ceded individual variation in sexual preference to the biologists and retreated to the macro-level, and biologists have not aspired to explaining macro-

________________________________________________

1 The most recent expression of this debate can be found in the August 2001 American Sociological Review, on the biological limits of gender construction, cf. commentary and response by Miller and Costello, Kennelly, Merz, and Lorber, Risman, and replies by Udry and Firebaugh, pp. 592-621.

2 The term .preferences. is often interpreted as signifying .choice., as versus .orientation. which is often interpreted as signifying a fixed characteristic. In this article we use these two words interchangeably. Fireworks aside, it is a false debate and it is not our intention to signal through word-choice a position on the .choice. versus .constraint. debate on the etiology of same-sex orientation, preference, attraction, or behavior.

Strana 1

level variation, social scientists and geneticists alike stress the obvious point that neither genes, nor hormones, nor specific social situations determine sexual behavior by themselves. Rather, the extent to which same-sex and opposite-sex desires are expressed in the individual is seen to be a complex interplay of biological, social, and situational factors (McGuire 1995; Parker and De Cecco 1995; Risman and Schwartz 1988). This is easy enough to say, and one could hardly disagree, but in this article we show how social science can be meaningfully brought back into the empirical debate about the etiology of same-sex preferences at the individual-level.

Specifically, we identify a specific social structure that posits limits to a specific aspect of gender socialization and thus allows for a more frequent expression of same-sex preference.

To anticipate the main findings reported below, we show that adolescent male opposite-sex (hereafter, OS) twins are twice as likely as expected to report same-sex attraction; and that the pattern of concordance (similarity across pairs) of same-sex preference for sibling pairs does not suggest genetic influence independent of social context. Our data falsify the hormone transfer hypothesis, by isolating a single condition that eliminates the OS twin effect we observe – the presence of an older same-sex sibling. We also consider and reject a speculative evolutionary theory that rests on observing birth-order effects on same-sex orientation. In contrast, our results support the hypothesis that less-gendered socialization in early childhood and pre-adolescence shapes subsequent same-sex romantic preferences.

Below, we review the four main theoretical models for same-sex romantic preference, and the evidence for each model. We identify the findings that would support or allow us to reject each hypothesis. We then describe our data, before turning to presentation of results. Social influences on sexual orientation It is commonly accepted that sexual expression varies from society to society; and that sexual socialization, as with culinary, dress, ritual and linguistic socialization, varies across cultures (Parker and Easton 1998, Herdt 1996). Consequently, sexual preference is seen as the product of specific social and historical forces that link in different ways diverse social processes organizing

gender identity, desire, scripted behaviors, and other cognitive and affective elements into a single framework. Because sexual expression varies so remarkably across cultures, it is obvious that what is considered erotic, the expression of erotic desires, and the organization of erotic practices, is the consequence of specific socialization experiences. It follows that variation within a society with respect to sexual preference (for example, same-sex preference in a society organized around opposite-sex eroticism) is seen by social scientists as the consequence of differential socialization experiences.

In contemporary American society, sex-role socialization in early childhood and pre-adolescence has been hypothesized to be associated with adolescent and adult romantic sex-preferences. As noted above, the traditional social science model . the attempt to discover specific aspects of childhood socialization that impacts same-sex preferences has been largely discredited, and consequently, abandoned (Risman and Schwartz 1988; Terry 1999). These studies did show that in contexts with strong sanctions against same-sex preference, gender socialization is deeply entangled with heterosexual orientation (Sedgwick 1991; Terry 1999). Specifically, social biases

for heterosexual erotic expression lead parents and others in interaction with children to subtly encourage gender-appropriate behaviors and to negatively sanction gender-inappropriate behaviors through reliance on gendered socialization scripts that shape response to children.s imaginative play, dress, and interactive style (Huston 1983).

Strana 2

The precise mechanisms for how such socialization affects later expression of sexual preferences are poorly specified in the literature. Strongly held norms against same-sex erotic interest are thought to induce parents to sanction behavior that is culturally associated with homosexuality. Although children of both genders are encouraged to behave in a sex-typical manner, stereotypically masculine behavior from girls is more often accepted than feminine behavior from boys3. Even at very young ages, peers ostracize or ignore males who prefer female-typed toys or games. Female .tomboys,. on the other hand, may be more accepted by both other girls and boys and even acquire leadership roles in play groups (Huston 1983). While poorly understood theoretically, it is possible to test for gender socialization effects. In this article we consider the hypothesis that parents. (and other socialization agents.) interactions with OS twins are less scripted with respect to gender socialization. Specifically, because OS twins are

similar, except for gender, parents are hypothesized to treat them similarly. Less gendered upbringing, should, if the social influence hypothesis is correct, be associated with increased rates of same-sex erotic preference, especially for males. On the other hand, the socialization hypothesis is falsified if OS twins. same-sex romantic preference rates are comparable to other populations. One simple model is that through subtle interactions, social norms work to limit the organization and articulation of same-sex erotic preferences. In the absence of social structural (or cultural) constraint, same-sex erotic preference rates revert to a .baseline rate., in the same way that objects placed in a vacuum become weightless. Here, this weightlessness, whether expressing itself as higher or lower rates of same sex erotic preference, may be seen as the .residual genetic. effect, given that the social shaping processes are eliminated. More plausible is the idea that genetic expression is activated only under strongly circumscribed social structural conditions4. In contrast to other theories considered below, we assume that the close connection between gender identity and sexual identity is socially constructed.

Genetic influence on sexual orientation A number of previous studies, most prominently, twin studies, have argued that there is a genetic component to same-sex romantic preferences. In this context it is helpful to understand what support for the genetic influence hypothesis would look like. Fundamentally, such support depends on concordance rates for same-sex preferences across sibling pairs. Monozygotic (MZ)

twins concordance should be higher than dizygotic (DZ) twins. Likewise, DZ twins concordance should be comparable to full-siblings (who are genetically similar, except for age). Concordance for DZ twins and full-siblings should be greater than unrelated pairs or stepsiblings. A summary statement is that if concordance rates do not parallel degree of genetic similarity, a simple genetic influence model should be rejected.

Against this background, most family studies report findings that support a general genetic  influence model, that is, they show that monozygotic twins report higher concordance for homosexuality than dizygotic twins, that brothers of homosexual subjects are more likely to be homosexual than brothers of heterosexual subjects, and that concordance rates for sibling pairs

3 In the 1980 Diagnostic and Statistics Manual published by the American Psychiatric Association, the first issue that did not include an entry for homosexuality and the first to include an entry for Gender Identity Disorder (p. 265-6), girls are diagnosed with this disorder only if they (mistakenly) insist on being anatomically male. In contrast, boys having a preference for cross-dressing or a .compelling desire to participate in the games and pastimes of girls. Are considered to have the disorder (Sedgwick 1991). While the development of gender identity and sexual preference may well happen independent from each other in the course of childhood and adolescence (Savin-Williams 1998, Sedgwick 1991, Whisman 1996), in the minds of parents, therapists, and peers, they go together. For boys and for their social

environment, heterosexuality symbolizes masculinity.

4 The distinction between these two formulations may seem opaque. The first assumes a baseline predisposition, the second a pure interaction effect, that is, no main effect for genes.

Strana 3