A reanalysis of five studies on sexual orientation and the relative length of the 2nd and 4th fingers (the 2D:4D ratio).

INTRODUCTION

The ratio of the lengths of the human index and ring fingers (the 2D:4D ratio) exhibits a substantial sex difference (for a review, see Manning, 2002). Males typically have smaller 2D:4D ratios than females. This difference is present at least as early as 2 years of age (Manning, 2002, Fig. 1.6). It has been suggested that this sex difference emerges during prenatal development and that it, like so many other sex differences in body, brain, and behavior, depends upon the degree to which a fetus is exposed to androgens (Manning, Scutt, Wilson, & Lewis-Jones, 1998). In addition, there is considerable evidence linking the 2D:4D ratio to various sex-dependent traits, including medical conditions such as breast cancer, heart disease, and autism (for a review, see Manning, 2002). Altered finger-length ratios also have been reported in children having a form of attention-deficit/hyperactivity disorder (McFadden, Westhafer, Pasanen, Tucker, & Carlson, 2003), which is much more common in males. Further, the 2D:4D ratio is negatively related to sensitivity to testosterone, as indicated by variation in the structure of the androgen receptor gene (Manning, Bundred, Newton, & Flanagan, 2003). Taken together, these various findings suggest that the 2D:4D ratio can serve as a marker for hormonal events occurring during early development that may affect later health and behavior.

Evidence of various sorts supports the idea that the magnitude of the 2D:4D ratio is negatively related to the degree of exposure to androgens prenatally. One important fact is the existence of the sex difference at a young age (Manning, 2002, Fig. 1.6). A second line of evidence comes from studies on people with congenital adrenal hyperplasia (CAH). In CAH, the adrenal gland produces abnormally high levels of androgens during prenatal development. Females with CAH are commonly born with masculinized genitalia and also have been shown to be masculinized on a number of behavioral measures, but the bodies and behaviors of males with CAH are less affected (see Collaer & Hines, 1995, for a review). Both Brown, Hines, Fane, and Breedlove (2002b) and Okten, Kalyoncu, and Yaris (2002) reported that the 2D:4D ratio was masculinized in CAH females and hypermasculinized in CAH males as compared to same-sex controls, in accord with the idea that prenatal androgen exposure reduces the 2D:4D ratio. (These differences were statistically significant for both sexes and both hands in the Okten et al. study, but were significant for only one hand for each sex in the Brown et al. study.) Although ultimately a correlational relationship, these findings do suggest strongly that excess androgen exposure can alter the relative lengths of the 2nd and 4th fingers. More recently, Buck, Williams, Hughes, and Acerini (2003) reported no difference in the 2D:4D ratio between girls with CAH and control girls. That study differed procedurally from the Brown et al. and Okten et al. studies in that only the left hand was measured, and the measurements were made from radiographs. Although the Ns were substantial in the Buck et al. study, it is not clear if their CAH sample was as ethnically homogeneous as their control samples and, if not, that may have contributed to the absence of a difference (see Manning, 2002, Fig. 1.7). Also, the radiographic procedure may have overlooked aspects of the digits (such as the fat pads at the fingertips) that contribute to the sex difference in relative length.

A third line of evidence in support of the prenatal-androgen exposure explanation is that the 2D:4D ratio in 2-year olds exhibited a negative correlation with the ratio of testosterone to estradiol measured from samples of amniotic fluid obtained during the prenatal development of those same children (Lutchmaya, Baron-Cohen, Raggatt, Knickmeyer, & Manning, 2004).

Partly because of the implication that the 2D:4D ratio can serve as a window on to hormonal mechanisms operating early in development, several studies have investigated how the 2D:4D ratio varies with sexual orientation. Although differences between homosexuals and heterosexuals have been reported, the direction of the difference has not been consistent across all studies. For example, some studies have found that the 2D:4D ratio in homosexual males was shifted toward the value found in heterosexual females (a hypomasculinization), but others have found the ratio to be shifted away from that of heterosexual females (a hypermasculinization). Similar discrepancies across studies exist for homosexual females. These inconsistencies across studies provided the chief motivation for the work reported here.

The authors of the various studies on the 2D:4D ratio and sexual orientation agreed to share their raw data so that detailed comparisons could be made at all stages of analysis, not just the final ratios. Thus, this paper reports not just a meta-analysis of some published studies but a true reanalysis of the original data. The hope was that these detailed comparisons would reveal the origins of the discrepancies and thereby provide insight both into the factors contributing to those discrepancies and into the nature of studies that might clarify them. Although some insights were obtained, some mysteries remain. The studies of interest were Lippa (2003), McFadden and Shubel (2002), Rahman and Wilson (2003a), Robinson and Manning (2000), and Williams et al. (2000).

Prior to explaining the analyses performed on the combined data sets, and discussing some of the issues addressed, we briefly summarize the findings of the studies as they were reported originally. We begin with the comparison between heterosexual males and heterosexual females and then consider the comparison between heterosexual and homosexual individuals for each sex separately. We confine ourselves to data from the second and fourth digits, although one study included other digits and length ratios as well (McFadden & Shubel, 2002; also see McFadden et al., 2003). For completeness, we note that Martin and Nguyen (2004) have reported that various skeletal measurements, including dimensions of the hand, suggest a hypomasculinization of homosexual males and a masculinization of homosexual females.