New Twin Study shows little Genetic / Utero Hormonal causation of homosexuality

From here:

Twin studies are favorites of mine because of the potential light they throw on the origin of SSA. The latest one (Santtila et al., 2008) is three times larger than any previous one – in fact bigger than all the rest put together. Does it change anything? Quick answer: No. It confirms the best recent studies: genetic factors are minor; non-genetic factors are major. That is, critical idiosyncratic reactions predominate.

The paper’s title was "Potential for homosexual response is prevalent and genetic." This implies to the average reader that homosexuality is perhaps hidden, but very common, and is predominantly genetic. We shall see this is not representative of their best conclusions.

This is the fifth systematically sampled twin study to look at SSA independently in men and women. Of the four previous studies, two were from Australia (Buhrich, Bailey & Martin, 1991, Bailey, Dunne & Martin, 2000 and two from the USA (Hershberger, 1997, Bearman & Bruckner, 2002).

This latest study is from Finland. Using the very centralized records typical of Scandinavian states, they assembled a large genuinely random sample of twins, (6,001 individual females and 3,152 males) primarily for a study on aggression. With that constraint they were permitted only two questions about SSA: "What same-sex sexual contact have you had in the last year?" and (in essence) "If there was no prospect of anyone finding out, and you were sexually propositioned by someone of the same sex you liked, what would be your chances of accepting?"

Before we go further let’s address one small difficulty. Unfortunately, different studies use different measures for SSA. Some ask for total number of partners – this one asked only the frequency of contacts in the last year. Other surveys ask how frequent same-sex fantasy was – this one asked respondents to fantasize (perhaps for the first time) about what sex might be like. The authors then call this "potential homosexuality," but you and I would probably conclude that such a measure is fairly clearly mostly indicating something other than SSA.

It obviously includes bisexual people and casts the net so wide that it could well be testing for something like novelty, curiosity, or sensation-seeking, rather than actual sexual orientation. 32.8% of men and 65.4% of women replied "yes" to that question in contrast to 3.1% of men and 1.2% of women who were actually homosexually active.

The results were:

Genetics Shared Environment Nonshared Environment
Men 27%(2.7-38) 0%(0-18) 73%(62-85)
Women 16%(8.3-24) 0%(0-3.6) 84%(76-91)

Men 37%(12-47) 0%(0-19) 63%(53-73)
Women 46%(32-52) 0%(0-11) 54%(48-60)

Table 1. Relative influence of various factors for the Santtila and Sandnabba (2008) data. Error ranges in parentheses are the 95% (2 sigma) error range.

The table shows the genetic effects are minor rather than predominant, and that non-shared factors (i.e. environmental factors particular to the individual) predominate. Given that the trend of twin study methodology will still be to overestimate the genetic content, we must say (considering the activity figures alone) that genetic effects are weak, and conclude the title of the paper is rather misleading. Even 27% in the above table is a maximum and has only indirect effects.

Twin study researchers usually involve the siblings of twin subjects as much as possible, because they are genetically related to the same degree as fraternal twins. This sibling/twin comparison is very interesting because it tests for any special twin environment. In this case, the siblings were tested along with the identical twins and fraternal twins and the results were meaningless, i.e. they did not yield results compatible with any plausible hypothesis of genetic influence in SSA. Although the authors do not specify exactly what the problem was it must have been severe ("…attempts at fitting univariate and bivariate extended-family scripts for categorical data were not successful….") This would usually be enough to knock such a study on the head, but rather incredibly the authors simply and blithely ignore the siblings for the rest of the paper and use the twins only.

As usual in these studies, family upbringing ("shared environment") was consistent with zero, but I contend again that many family factors are hidden in the "non-shared environment," and highly individualistic to the people concerned.

The results by my calculations do reinforce one conclusion drawn from previous studies. That is if one identical twin, male or female has SSA, the chances are only about 10 percent that the co-twin also has, i.e. identical twins usually differ for SSA.

In spite of my criticisms, some useful points emerge from the study. The sample is probably the least biased so far. The authors believe uterine hormonal influence theories do not hold because they should lead to greater similarly between identical twins, not less similarity as we see above. The final errors are less than in previous studies. Also we see a continuation of an already established trend; the later and better the study, the smaller the detected genetic influence on SSA.

In the meantime one should continue to assume genetic effects on SSA are minor, and other very individualist factors predominate.

Reference List

Bailey, J.M., Dunne, M.P., & Martin, N.G. (2000). Genetic and Environmental influences on sexual orientation and its correlates in an Australian twin sample. Journal of Personality and Social Psychology, 78, 524-536.

Bearman, P.S., & Bruckner, H. (2002). Opposite-sex twins and adolescent same-sex attraction. American Journal of Sociology, 107, 1179-1205.

Buhrich, N., Bailey, J.M., & Martin, N.G. (1991). Sexual orientation, sexual identity, and sex-dimorphic behaviors in male twins. Behavior Genetics, 21, 75-96.

Hershberger, S.L. (1997). A twin registry study of male and female sexual orientation. Journal of Sex Research, 34, 212-222.

Santtila, P., Sandnabba, N.K., Harlaar, N., Varjonen, M., Alanko, K., & von der Pahlen, B. (2008). Potential for homosexual response is prevalent and genetic. Biological Psychology, 77(1), 102-105.