There are myriad competing theories about just why men and women think and behave differently from each other. From the evolutionary, to the physical and chemical, to the social, explanations for gender differences in behavior, personality, and ability are nearly as varied as they are ubiquitous.
There’s just one problem: all too often, no one bothers to check if those differences actually exist first.
Countless pop-science articles and books (as well as numerous scientific studies) exist that purport to categorize, measure, or explain differences in the way that men and women view the world, react to stimuli, behave in relationships, and so on. The same sort of thinking is applied to works attempting to prove, justify or explain differences in male and female intellectual achievement, particularly in areas like science and mathematics. What I have for you today is some evidence, in the form of an analysis of a number of large datasets of personality traits, suggesting that all these articles, books and studies are putting the cart before the horse, and that most observed psychological differences between men and women are either much less marked than is commonly believed or show considerable overlap and cross-over between the sexes, pointing at causes much less fixed and immutable than many people would suggest.
Researchers at the University of Rochester analyzed data from 13 studies (comprising 13,301 individuals in total), and performed a variety of statistical analyses on each, attempting to determine whether any of the 122 indicators studied (and if so, which) formed such distinct groupings between male and female subjects that they could be used, either individually or in concert, to predict with a high degree of confidence whether a person was male or female – and by association, whether that person shared other such sex-linked traits with other members of their group.
The authors contrast their approach with one that notes an average difference between sexes (as is common in a lot of pop-science writing, among other things) and then attempts to explain it or treats it as an innate difference that distinguishes men and women. The latter approach is less useful because it masks some very common features of these sorts of datasets – the amount of overlap between groups and the amount of variation within each group, which can point at the differences being studied as much less significant than they are widely assumed to be.
To start, the researchers demonstrated the use of their analysis on datasets that would be largely noncontroversially accepted as broadly different between male and female subjects – size, physical strength, and “traditionally sex stereotyped” leisure activities, such as boxing and cosmetics. They showed that these variables could be used to accurately sort men and women into distinct groups, with very few cross-overs. This result, as the researchers note, is unsurprising – sex-based differences in size and physical strength are well-established, and the leisure activities used in the validation phase were selected for being overwhelmingly preferred by one gender over the other. However, the results demonstrate that their analyses are capable of finding sex-based differences in a given dataset.
Moving on to the meat of the analysis, however, the researchers took on a slew of other personality and behavioral attributes, grouped broadly into four categories:
- Sexuality and Mating (comprised of data about sexual behavior, attitudes, and partner or mate selection)
- Interpersonal Orientation (broken into two subsets related to empathy and relational interdependence)
- Gender-Related Disposition (among other things, measures of masulinity, femininity, and inclination towards science), and
- Intimacy (both with romantic partners and in non-romantic contexts, such as with a close friend)
They analyzed these sets of variables separately, and then all together, looking for evidence that they could be used to predict whether a given subject was male or female. And what they found, with a very few exceptions, was…nothing.
For all but a handful of the variables studied, the results showed that rather than displaying distinct psychological differences between men and women, each variable instead fell along a continuum, with considerable overlap between the sexes. Furthermore, they found that having an especially “male” or “female” score in one trait was not predictive of a subject scoring similarly in other traits. In other words, if you simply take the average of the sexes for each trait, you’ll find that men are more aggressive and women are more talkative – but finding that a given person is especially aggressive doesn’t let you predict how talkative they’ll be, or vice versa. This continuum of responses makes it less likely that the analyzed traits are sex-linked in a biological sense, also, since we’d expect them to be much more nearly universal within the sexes (as, for example, having breasts is among women, or facial hair among men) if they were.
This has implications for how we write characters of both sexes – because a character who falls into neatly delineated sex-based categories in all of their personality traits is not only likely to be less interesting than a more nuanced one, they’re also demonstrably unrealistic – and just as it makes for a better story when our characters’ physical worlds are well-developed and realistic, so, too, do we benefit when their inner worlds reflect the same kind of nuance.