them to prefer females as romantic partners. Because the expression of same-sex erotic attraction appears to be independent of traits governing the expression of culturally induced images of femininity and masculinity, the mechanism linking hormone imbalance to same sex preference appears extremely weak.

One version of the hormone transfer hypothesis focuses on the .masculinization. of females and predicts increased levels of same-sex attraction among female, but not male opposite-sex twins.

We test this hypothesis and find no support for it. A less restrictive version, which allows for a .feminization. of males, appears at first glance to be inseparable from the socialization hypothesis. This is not the case. In this article, we design a test to isolate the socialization effect.

Specifically, we consider same-sex romantic attraction for OS twins with an older same-sex sibling. The socialization hypothesis suggests that if a same-sex older sibling is present in the household, parents and other socialization agents would have already established scripts for gendered upbringing (Huston 1983). These scripts provide models for interaction with the twin of the same sex. The hormone transfer hypothesis is falsified if we show that OS twins with an older same-sex sibling do not report same-sex preferences that are different from the other sibling pairs, since the effects of hormone transfers should be insensitive to birth-order. This is the case in our data. Table 1 summarizes the predictions arising from the models reported above.

Table 1: Social Influence, Genetic Influence, Hormone Transfer, and Evolutionary Dynamics Predictions for Adolescent Same-Sex Romantic Attraction.

Data and Design

Data for this study were drawn from the National Longitudinal Study of Adolescent Health, a nationally representative, school-based study of adolescents in the 7th .12th grades. Data from the first wave, a self-administered in-school survey conducted in 1994-1995, were used to generate a nationally representative core sample of students and several special samples for a second and third wave of interviewing. A genetic sample of twins and siblings living in the same household was drawn from the 90,118 respondents of the first in-school survey. The genetic sample comprises 5512 persons making up 3139 pairs of siblings. Both siblings in each pair were interviewed in a second, in-home survey. The genetic sample consists of 289 pairs of MZ twins,

495 pairs of DZ twins, 1251 pairs of full siblings, 442 pairs of half siblings, and 662 pairs of nonrelated siblings. The majority of same-sex twins were determined to be MZ or DZ based on their self-reported confusability of appearance (averaged over both twins. self-report). When selfreport data on appearance was missing, MZ or DZ classification was made from the mother.s report of confusability of appearance, or on the basis of molecular genetic markers12.

Including the special samples, 20,745 adolescents were interviewed in the second wave in 1994- 95, which solicited information on socio-economic background, demographic variables, health status and health risk behavior, self-esteem and depression, sexual activity, romantic relationships and friendships, as well as academics, expectations for the future, and employment. Over 79% of eligible respondents completed the follow-up second wave interviews. With the exception of seniors, all respondents were eligible for a follow-up survey in 1996, resulting in 14,738 interviews. The response rate for the third wave was over 80%.

_______________________________________________

12 For details on the design of the National Longitudinal Study for Adolescent Health, see P.S. Bearman, J. Jones, J.R. Udry. The National Longitudinal Study of Adolescent Health: Research Design. URL:

http://www.cpc.unc.edu/addhealth

Strana 8

For the present study, same-sex romantic attraction was based on the question: .Have you ever had a romantic attraction to a female (male)?. Both in-home surveys used ACASI technology (Audio computer-assisted self-administered interviewing) for sensitive questions about sexual and non-normative behavior, including the question about same-sex romantic attraction. ACASI technology has been shown to yield more accurate responses to sensitive questions than standard interview technique (Turner et al. 1998). The proportion of adolescents reporting a same-sex relationship or homosexual activity is small in this sample (3.4 % and 0.84 % respectively).

Consequently, we focus on same-sex romantic attraction.13

Results

Table 2 reports the proportion of adolescents reporting same-sex attraction, by gender and sibling classification. Overall, 8.6 % of the full sample reported a same-sex romantic attraction in the first and/or second in-home survey, 7.8 % for female adolescents and 9.4 % for males (N=20,729; 26 respondents had missing data on both waves). We first consider support for the social influence model that hypothesizes that opposite-sex twins should be subject to a different socialization regime than same-sex twins or opposite-sex full siblings. To allow for unambiguous classification, rows 1-7 exclude respondents who contribute more than one pair to the genetic sample (94 % of persons in the genetic sample are part of only one pair). Row 9 reports prevalence for a comparable group in the non-genetic sample, and row 8 for the entire nongenetic sample. The genetic and non-genetic samples are not significantly different from each other with respect to prevalence of same-sex attraction.

Strana 9