Next: Critical Analysis of the Studies; Previous: Introduction; Up: Contents


Why twin studies?

Twins potentially provide us with one of the most powerful techniques available for separating the influences of genetic and environmental factors on psychological and behavioural traits. An identical (monozygotic) twin has exactly the same genes as his cotwin. Fraternal (dizygotic) twins, on the other hand, are no more closely related genetically than are normal siblings, i.e. roughly 50% of their genes are the same.

If index cases for a particular trait (e.g. homosexuality) are selected from twins in a given population, then the rate of concordance for that trait between the index case and his or her cotwin can be compared for monozygotic and dizygotic twins. It is assumed that the environment in which each member of a twin pair is reared is virtually identical, as twins are identical in age, and tend to have very similar experiences during childhood (e.g. they get similar attention from their parents, attend the same schools, etc.). Therefore, any difference in the concordance rates for the trait between monozygotic and dizygotic twins must be due to the former sharing all of their genes and the latter only sharing half of their genes.

In other words, the rationale behind twin studies of homosexuality is that if there is a difference between the concordance rate for homosexuality in monozygotic twins and dizygotic twins, then this is strong evidence that there is some genetic component to the aetiology of homosexuality. However, if the concordance rate in monozygotic twins is not 100%, then environmental factors must be exerting some influence.

Another technique involving twins is the study of identical twins reared apart. In such cases, we can look at the effect of varying the environment whilst keeping genetic factors constant (c.f. the previous method, where the environment that both twins experience is constant, but genetic factors are varied by comparing monozygotic with dizygotic twins). Looking at a particular trait in a sample of identical twins, the difference in variability of the trait observed among twins reared together and twins reared apart must be predominantly due to the variability in the rearing-environments experienced by twins reared apart. Obviously, the number of cases of homosexual identical twins who have been reared apart that come to the attention of researchers is minimal, but at least one report has been published on this subject (Eckert et al. 1986).

There now follows a summary of the methods and results of all of the major twin studies of homosexuality conducted to date, together with a summary of smaller studies and reports of individual twin pairs.

The major studies to date

Kallman, 1952

Kallman obtained 85 male index cases who were predominantly or exclusively homosexual, members of a twinship, and over the age of 20. Of these 85 index cases, 40 were monozygotic twins, and 45 were dizygotic twins, although no mention is made of tests administered to prove zygosity. The exact method of recruitment is unclear from the report, but it is stated that "the search for potential index cases was organized not only with the aid of psychiatric, correctional, and charitable agencies, but also through direct contacts with the clandestine homosexual world."

Each index case and his cotwin were given a Kinsey rating from 0 (exclusive heterosexuality) to 6 (exclusive homosexuality), although the precise method by which these ratings were obtained is, again, unclear.

For the dizygotic twins, Kallman found a concordance rate of 11.5% (3/26) for predominant homosexuality (Kinsey ratings 3-6), and 42.3% (11/26) for any homosexuality (Kinsey ratings 1-6). These figures exclude 19 of the 45 dizygotic index twins, 14 of whom had female twins, and 5 of whom had unclassified male twins (deceased or otherwise unavailable).

For the monozygotic twins, a very different picture emerged. Of the 37 index cases whose cotwins were classified, all were concordant for homosexuality (Kinsey ratings 3-6); a concordance rate of 100%. Kallman noted that "the majority of one-egg pairs not only are fully concordant as to the overt practice and quantitative rating of their aberrant sex pattern, but they even tend to be very similar in both the part taken in their individual sex activities and the visible extent of feminized appearance and behavior displayed by some of them" (Kallman's italics). He also pointed out that most of these twins had "developed their sexual tendencies independently and often far apart from each other, and that all of them deny categorically any history of mutuality in overt sex relations."

Heston & Shields, 1968

Heston and Shields reported on the male homosexual twins on the Maudsley Twin Register as of July 31, 1966. Such a sample appeared to be representative of Maudsley patients as a whole, not being selective regarding concordance or zygosity.

12 index cases were obtained. Zygosity was determined using blood samples and fingerprints, revealing that five were monozygotic, and the remaining seven were dizygotic. The report states that "the proband twins were examined as psychiatric patients. They and their cotwins were further assessed . . . through personal interviews and . . . by means of psychological tests of intelligence . . . and personality . . ."

The family of one of the monozygotic twins was studied extensively, as it contained three sets of male identical twins, two of which were concordant for homosexuality; the other pair being concordant for heterosexuality. As in Kallman's study, Heston and Shields reported that the two sets of homosexual twins "denied any sexual contact or feeling for each other and reported the idea as distasteful."

Of the other probands, only one of the other four monozygotic twins had a cotwin clearly concordant for homosexuality, although one other cotwin had had delusions of sex change and other psychiatric complaints. Of the seven dizygotic twin pairs, one, again, was clearly concordant for homosexuality, and the cotwin of another dizygotic proband was suspected of latent homosexuality, although this was not proved.

There was evidence for other psychiatric disorders in many of the probands and their cotwins, but Heston and Shields concluded that, "in general, the presence or absence of other diagnosed psychiatric conditions in the MZ probands does not appear to account for the resemblance in homosexuality." They also commented that there "appears to be no good evidence from the present material or from other work for supposing that twins [per se] have a high risk of being homosexual."

It must be remembered that the conclusions of such a study, as Heston and Shields emphasized themselves, are limited due to the small numbers involved, and the fact that the sample cannot be considered representative of homosexuals as a whole. As the report says, "similar objections can be raised to most clinical studies of homosexuality."

Bailey & Pillard, 1991

In the most recent twin study of homosexuality, Bailey and Pillard report on a sample of 161 male homosexual probands, all over the age of 18 with a twin or adoptive brother. Recruitment was through advertisements placed in homophile publications in several cities in the Midwest and Southwest of the United States. For probands with adoptive brothers, it was stipulated that they must have entered a common rearing environment when both were no more than 2 years old.

Most of the probands were interviewed in person with informed consent. Questionnaires were sent to the cotwins and adoptive brothers, with a cover letter explaining that they were being asked to participate in a general study of personality, attitudes and behaviour. Five questions regarding sexual orientation were imbedded in over 100 other items about social attitudes, personality, and childhood behaviour.

Bailey and Pillard determined the zygosity of the subjects using a questionnaire containing items relating to physical similarity, past and present likelihood of twins being mistaken for each other, etc. They claim that "such questionnaires generally range in accuracy from 90% to 95%."

The distribution of sexual orientation among cotwins of the monozygotic probands appeared to be bimodal. In other words, most subjects classified themselves as heterosexual or homosexual, with very few giving evidence of significant bisexuality. This finding is in agreement with other recent studies (e.g. Buhrich et al. 1991).

According to their data, 52% (29/56) of monozygotic cotwins, 22% (12/54) of dizygotic cotwins, and 11% (6/57) of adoptive brothers were homosexual. Heritabilities of homosexuality were calculated using these results under a wide range of assumptions of the population base rate and ascertainment bias. Under all conditions considered, heritabilities were substantial (h2 was between .31 and .74 in all cases). However, "the rate of homosexuality among nontwin biological siblings, as reported by probands, 9.2% (13/142), was significantly lower than would be predicted by a simple genetic hypothesis and other published reports."

Bailey and Pillard suggested that this anomaly between the present data and previous reports might be due to the fact that they were looking at nontwin biological siblings of twin probands, whereas other reports had reported the concordance rate between pairs of nontwin siblings where, on the whole, neither sibling was a member of a twin pair. A difference in concordance rates may then occur "if the causes of homosexuality in twins and singletons were different, i.e., if a special twin environment contributes to the development of sexual orientation." Alternatively, the difference could just be due to sampling fluctuations, so the desirability of replicating the results was stressed.

The results of the three twin studies of homosexuality described above are summarised in table 1. It is notable that all three have concentrated on exclusively male samples; no large scale investigation of homosexuality in female twins has yet been conducted.

Table 1. Summary of the major twin studies of homosexuality.
Kallmann 1952  >20 37/37 (100%)  3*/26 (12%) Psychiatric, correctional and charitable agencies, plus direct contacts 
Heston & Shields 1968 20-52  3/7 (43%) 1/7 (14%)  Hospital Twin Register 
Bailey & Pillard 1991 19-65  29/56 (52%) 12/54 (22%)  Homophile publications 
* Concordance rate varies from 3/26 (Kinsey scale 3-6) to 11/26 (Kinsey scale 1-6)

N.B. The data in this table represent only those cases where the sexual orientation of each subject is 'fairly certain', and is either predominantly homosexual or predominantly heterosexual.

Small scale studies and individual case-studies

In addition to the three studies just described, there have also been numerous smaller scale investigations and reports, some being more useful scientifically than others. Many of these reports have included cases where the sexual orientation of one or more of the subjects was by no means certain (e.g. Parker, 1964; Holden, 1965; Green & Stoller, 1987). Table 2 summarises the reported cases where there was some degree of certainty regarding an individual's orientation, and where such orientation is predominantly homosexual or heterosexual (rather than significantly bisexual).

Table 2. Summary of small scale reports and individual case studies of homosexuality in twins.
Lange 1929  24 &  

mid 20s 

1/2  Criminal and psychiatric records 
Sanders 1934  ND 5/6  0/1  Homosexual probands 
Habel 1950  ND 3/5  0/5 Prison Population 
Rainer, Mesnikoff, Kolb & Carr  1960 29 & ND  0/1 0/1  Psychoanalytic agency 
Klintworth 1962  20 0/1  Hospital 
Mesnikoff, Rainer, Kolb & Carr  1963 ND  0/2*  0/1 Psychoanalytic and psychotherapeutic patients 
Koch 1965  >25 0/1 0/1  Follow-up on non-psychiatric, non-delinquent twins 
Pardes, Stinberg & Simons  1967 36  1/1 Psychiatric clinic 
Green & Stoller 1971 24  0/1 Referral to Gender Identity Research and Treatment Programme 
Perkins 1973  45 1/1 ND
Zuger 1976  23 0/1  Reported by family doctor 
Friedman, Wolleson & Tendler  1976 25  0/1 Volunteers 
McConaghy & Blaszczynski  1980 21  0/1 Psychiatric centre 
Myers 1982  27 1/1  Psychiatric centre 
Eckert, Bouchard, Bohlen & Heston  1986 25-48  1/1  0/3  Minnesota study of MZ twins reared apart 
Buhrich, Bailey & Martin  1991 19-40  4/9 0/2 Australian NH & MRC Twin Registry 
ND No Data

* Report also included the two MZ twins reported by Rainer et al. (1960).

N.B. The data in this table represent only those cases where the sexual orientation of each subject is 'fairly certain', and is either predominantly homosexual or predominantly heterosexual.

It is also apparent that some reports are describing twins who have already been reported by other authors. Mesnikoff et al. (1963) report five twins, two of which were already reported by Rainer et al. (1960). Parker (1964) and Heston & Shields (1968) both obtained probands through the Maudsley Twin Register, so that the two male twin pairs reported by Parker are also included in Heston & Shields' study. In both of these cases, the twins are only recorded once in table 2.

Because of the limited information regarding the relative importance of environmental and genetic factors that may be obtained from such reports, most will not be described in detail.

Many of the cases are drawn from prison populations (Lange, 1929; Habel, 1950) or through psychiatric clinics (e.g. Rainer et al., 1960; Mesnikoff et al., 1963; Pardes et al., 1967; McConaghy & Blaszczynski, 1980; Myers, 1982). Samples drawn from such sources are unlikely to be representative of a larger population. In some of the reports, especially the earlier ones (e.g. Lange, 1929; Sanders, 1934), cotwins of some probands were not explicitly asked about their sexual orientation, which was just inferred from other sources of information.

After Kallmann's 1952 report of a 100% concordance rate for homosexuality in male twins, many subsequent investigations have been biased towards searching for discordant pairs (e.g. Rainer et al., 1960; Klintworth, 1962).

In Habel's 1950 report of homosexuality in twins from a German prison population, he drew the distinction between 'genuine' homosexuality and 'pseudohomosexuality', important for prison populations in particular. In fact, other reports (e.g. Slater, 1962; Parker, 1964) have also suggested that there are various different 'types' of homosexuality, all with a different combination of environmental and genetic aetiological influences. If this is true, then estimations of heritability for homosexuality in a given population can only be seen as referring to an 'average' of all of the different 'types'. Individual types may be much more strongly heritable than the overall estimate, or much less so.

The discordant male monozygotic twins studied by Davison et al. (1971) are peculiar among all of the cases mentioned in that the homosexual twin, Paul, was given aversion therapy at age 18, and claimed to have lost all homosexual desires after this. He subsequently became engaged to a girl, but broke this off before marriage. Two years later he still claimed to be free from homosexual desires. Bearing in mind the work of Kinsey (1948) and others, some doubt is cast onto the relevance of any report that deals with the sexual orientation of people under 25 years old.

Throughout the 1960s and 70s we see a spate of reports, each of one or two concordant or discordant twin pairs, usually written from a psychoanalytic point of view.

Such psychoanalytic reports often investigate inter-family environmental influences in the childhood of the probands that appear to have been influential in the subsequent development of sexual orientation. Such factors will be discussed at a later stage.

Two recent reports (Eckert et al. 1986, Buhrich et al. 1991) have investigated sexual orientation in relatively large series of twins (55 and 158 twin pairs respectively) unselected for homosexuality.

Eckert et al. utilised data from the Minnesota Study of Twins Reared Apart. "That the twins are highly selected cannot be doubted; they are not representative of twins or homosexuals," the authors warn. "Nevertheless, study of them has yielded clues which warrant description." Six pairs of monozygotic twins were found in which at least one member was homosexual (in five cases) or bisexual (in one case).

Of the four female twin pairs, three were discordant for homosexual behaviour. The fourth case contained the bisexual index case, who had had a prolonged and intense homosexual affair in her late twenties, but regarded herself as exclusively heterosexual at the time of the study. Her cotwin was heterosexual. One of the male twin pairs reported was clearly concordant for homosexuality, and after their reunion at age 25 became sexual partners. In the other male twin pair, one member was exclusively homosexual, and reported being sexually attracted to his cotwin after reunion. The cotwin regarded himself as exclusively heterosexual, although he had had a homosexual affair with an older man between ages 15-18. For the male twins, Eckert et al. conclude that, "despite problems of ascertainment and diagnosis, it is hard to deny genetic factors an aetiological role." As for the females, the "pattern of findings suggests that female homosexuality is a trait acquired after conception, most likely after birth, but before menarche . . . Our evidence, though based on a small sample, implicates environmental factors as the major determinant of female homosexuality." Similar findings have been reported elsewhere (e.g. Bell et al. 1981)

In the study by Buhrich et al., 95 pairs of male monozygotic twins and 63 pairs of dizygotic twins were assessed for present and childhood sexual identity, sex-dimorphic behaviours and sexual orientation. A significantly higher rate of adult homosexuality was found among the monozygotic than the dizygotic twins; the precise figures of concordance for predominant homosexuality (ratings 4-6) were 44% (4/9) for monozygotic twins, and 0% (0/2) for dizygotic twins. These figures were not explicitly quoted in the report, whose authors warned against ignoring the evidence "that sexual feelings and sexual orientation have a continuous (if skewed) distribution."

A model-fitting approach was employed to test for genetic and environmental influences on variation for each trait singly and on the covariation among all traits. The results of multivariate analyses suggested that "over half the variance in Adult Sexual Orientation was attributable to the additive effects of genes. Only about 1% of the variance was attributable to the influence of environmental factors shared by siblings." However, because of the restrictive nature of some of the models, and because the sample showed different rates of orientation by zygosity, Buhrich et al. concluded that their results "are best considered tentative but do suggest that further biometrically oriented studies of sexual orientation and its correlates would be worthwhile."

This concludes the résumé of twin studies of homosexuality conducted to date. Clearly, many of these studies have (sometimes severe) methodological shortcomings. As Parker (1964) notes, "as long as there is an associated stigma attached to any deviation from normal sexual behaviour, difficulty will be experienced in obtaining co-operative twin pairs, and this has been repeatedly emphasized in the literature." Even if we just consider the less selective reports, there is still considerable debate as to how much the results of any twin study can tell us of the relative strengths of genetic and environmental factors. These problems are discussed in the next section.

Next: Critical Analysis of the Studies; Previous: Introduction; Up: Contents
This document was converted to HTML on 16th December 1997 by Tim Taylor