Personality and Sexual Orientation: Extension to Asexuality and the HEXACO Model

Abstract

Previous research has examined self-identified sexual orientation in relation to self-reports on the Big Five personality factors. Here we extend this research by including asexuality as a fourth sexual orientation and by assessing the HEXACO personality factors, using self-report data from a large anonymous online sample (N ≈ 100,000). A large overlap was observed among all sexual orientation groups on all six HEXACO dimensions, but notable group differences also emerged. All nonheterosexual groups, particularly bisexual people, averaged higher in openness to experience than did heterosexuals. Heterosexual men averaged much lower in emotionality than heterosexual women, but only a small difference was observed between gay men and lesbians on this dimension. Asexual men and women averaged lower in extraversion and in some facets of emotionality (specifically, dependence and sentimentality) than did the other sexual orientation groups of the same sex. Results are discussed in relation to the different social realities that each sexual orientation group often experiences, along with the (gender-related) developmental influences hypothesized to underlie in part the origins of sexual orientation.

Acknowledgments

We thank Carolyn Hafer for comments on a previous draft of this manuscript.

Notes

¹ Note that, although the emotionality scale of the HEXACO inventories shows fairly large sex differences, that scale does not contain any items assessing gender-related interests (e.g., in occupations or recreational activities). Although interests represent important psychological individual difference variables, they are generally considered to define their own domain of individual differences, separate from the domain of personality traits, and thus are generally assessed by interest inventories rather than by personality inventories. One noteworthy exception is the 16PF Sensitivity Scale (e.g., Cattell & Schuerger, 2003), which contains several items that ask a respondent’s preference for a stereotypically more masculine or stereotypically more feminine subject of interest, such as poems versus football, love stories versus action movies, math classes versus English classes, or building things versus reading. This item content likely contributes to the very large sex differences observed on this scale (see, e.g., Del Giudice, Booth, & Irwing, 2012). In the present study, we did not assess gender-related interests, because we believe that large sex differences and large sexual orientation differences have been well established for this variable.

² With respect to the scalar invariance model, the CFI change from the metric equivalence model (i.e., .939 to .923) is somewhat greater than the .01 value conventionally used to support invariance constraints (Chen, 2007). The subsequent analyses showed that the decrease in CFI occurred mainly because the intercepts of the openness facets were constrained to be equal across sex, whereas in fact the inquisitiveness and aesthetic appreciation facets differed greatly in their sex differences: Men averaged more than half a standard deviation unit higher than women in inquisitiveness, but women averaged nearly one-third of a standard deviation unit higher than men in aesthetic appreciation. (See Marsh et al., 2010, for a similar finding observed for NEO-PI-R openness to experience.) Some researchers might interpret this result to suggest that it would not be meaningful to compare men and women on the broad openness to experience scale. As pointed out by McCrae (2015, pp. 107–108), however, it is entirely plausible that different facets of openness would show different patterns of sex differences, and it would still be meaningful to combine those facets and examine sex differences in an overall openness score. As noted later in this article, sex differences in the overall openness to experience scale were in any case quite small.

³ For comparisons of sexual orientation groups within each sex, we computed d statistics using the standard deviation of the heterosexual group rather than a pooled standard deviation for the pair of groups being compared. In this way, any sexual orientation groups having the same mean will obtain the same d value in comparison with the heterosexual group.

⁴ An anonymous reviewer suggested the calculation of multivariate Mahalanobis D statistics as indices of the overall differences between the sexes and sexual orientation groups (see, e.g., Del Giudice, 2009). The overall difference between heterosexual men and heterosexual women on the six factor-level scales was D = 1.11, and this value increased to D = 1.40 when computed on the 25 facet-level scales. Differences between sexual orientation groups were smaller: Using the six factor-level scales, we obtained Ds of 0.61, 0.54, and 0.88 for heterosexual men in comparison with gay, bisexual, and asexual men, respectively, and Ds of 0.41, 0.73, and 1.05 for heterosexual women in comparison with lesbian, bisexual, and asexual women, respectively; the corresponding values using the 25 facet-level scales were 0.80, 0.65, and 1.10, and 0.65, 0.88, and 1.18. These multivariate indices thus indicate substantial differences but also heavy overlap between the sexes and sexual orientation groups.

⁵ In relation to the above point, we note that VanderLaan, Gothreau, Bartlett, and Vasey (2011) reviewed findings that “preandrophilic” boys (i.e., boys who grow up to become androphilic or gay men) on average show relatively high levels of separation anxiety. They also suggested that this separation anxiety reflects a tendency toward kin-directed altruism, and that such a tendency has helped in part to maintain across generations the genes that promote an androphilic orientation in men. The present findings of higher mean levels of emotionality in gay men than in heterosexual men are consistent with the suggestion of VanderLaan et al. (2011), given that emotionality—which is defined in part by traits relevant to separation anxiety—is interpreted as the personality dimension underlying tendencies relevant to kin altruism (see, e.g., Ashton & Lee, 2007).

⁶ An anonymous reviewer suggested we address to what extent our study has implications for viewing asexuality as a disorder (e.g., as a diagnosable disorder in the Diagnostic and Statistical Manual of Mental Disorders [DSM]). The complex issue of whether asexuality is a disorder has been addressed in recent reviews of the literature (e.g., Bogaert, 2006, 2012, 2015; Brotto & Yule, 2016). Generally, the conclusion drawn by the authors of these reviews has been that asexuality is not per se a disorder. Our study can be argued as potentially adding another view against a widespread pathologizing of asexuality, as asexual people have a significant overlap in personality with sexual people and the majority of asexuals are well within the typical range of personality variation. For example, even on extraversion, on which asexual persons averaged nearly three-quarters of a standard deviation unit lower than heterosexual persons, for a given asexual person and heterosexual person selected at random from our sample, there would nevertheless be roughly a 30% chance that the asexual person would have a higher extraversion score than the heterosexual person. It is also noteworthy—as pointed out to us by an anonymous reviewer—that on most dimensions asexual persons showed slightly greater variability than did heterosexual persons (see the standard deviations of Tables 1 to 6).