Do more gender equal countries have greater sex differences?

So I’ve seen a few articles arguing this point: that sex differences increase in more gender equal countries. This has been found for personality, school performance, a many individual differences and occupation choice. Even as someone who studies evolutionary and biological influences on behaviour, this is counter-intuitive for me. I’d expect that in a culture of gender equality, those who don’t want to follow a sex-typical life would do so, blurring the lines between the sexes. But that’s not what the data seem to suggest.

The Theoretical issue

The papers that find this relationship tends to make a similar argument: more gender-equality reduces cultural pressures, which means that any biological ones, no matter how small, come to the surface. This makes sense. However, this is not exactly what studies find. The evolutionary biological theory does not predict that sex differences would increase in gender-equal societies. All that explanation predicts is that countries with greater gender equality would converge on the true effect of biology in the absence of culture on outcomes (whatever that means). The effect of biology can be exaggerated or quashed by a culture of gender inequality, so it could increase or decrease when removed. It does not predict a straightforward linear relationship between sex differences and gender equality.

To illustrate the problem as I see it, imagine two countries: one has very high gender equality and one has very little. In the gender equal country, there are no cultural forces propelling people towards gender typical careers or pressuring them to develop personalities in a certain way. We can assume that the majority of gender differences seen here are due to biological factors. Now we consider a country with little gender equality. Here, biology may play a relatively small role, and culture will largely force people into gender-typical roles, and to display gender-typical personality. So on the one side, we have biology, and the other we have biology plus a culture of gender typicality. Assume biology is equal in both societies. For us to find that gender-equal countries show greater sex differences than gender unequal ones, Biology alone must create greater gender differences than the same biology plus cultural factors. For this to be true, cultural factors forcing people into gender-typical roles must reduce gender differences in outcomes.This seems very unlikely, though I may have missed something important here.

Potential statistical issues

Something that may shed more light on this is going beyond comparing means. Showing that sex differences in means of an outcome vary across countries is fine, but may obscure all sorts of interesting dynamics. For many of these distributions, the mean may not be an appropriate summary of what’s going on, especially if they are skewed. Distributions might have the same mean but look very different:

Or two distributions could have the bulk of their observations in the same place but have a decent mean difference, as below:

So if we dig down into these mean gender differences in traits or outcomes, they may look very different between countries.

Possible publication bias?

So these findings do seem to replicate, which is fantastic, and I applaud authors and editors alike for conducting and publishing replications of these phenomena. But then, there are tons of replications (conceptual ones at least) for many findings that we now think are probably not real, like Ego Depletion, Stereotype Threat and Facial Feedback. There is always the possibility that this effect is due to publication bias. The fact that it’s surprising should set spidey senses tingling here: surprising findings are often more likely to get published.

However, given how this result is mainly counterintuitive to progressives, who dominate academia, you might think that this work would be less likely to be published, considering it kinda contradicts the narrative. I wouldn’t go as far as to say that research that contradicts the progressive narrative is censored, but we critique things more if they contradict our worldview, so the research may be less likely to be published. Overall though, I don’t think this would outweigh the publication advantage this work gets by being sexy and counterintuitive. I can’t say for sure, but perhaps publication bias is making this finding appear more robust than it actually is.

Third variables

Another thing to bear in mind is that all these data are correlational, so already causality is out the window. But also there’s the possibility that other variables could account for or mediate the relationship. For example in a review of the subject Costa et al’s found that sex differences in personality were positively correlated with GDP (r = .47), women’s life expectancy (r = .57), and negatively associated with fertility rate (r=-.56), women’s illiteracy rate (r = -.46) and women’s illiteracy rate relative to men’s (r = -.48). It’s possible that these variables might cause the increase in sex differences in gender equal societies. So a variety of factors related to country development seem relevant here. We really shouldn’t be doing simple studies where we correlate differences to development across countries.

Conclusions

I’m not going to pretend that this blog post has ‘debunked’ anything, but I feel I have given a variety of reasons why this finding may require a bit more attention. I think researchers interested in this phenomenon should look a little deeper, present richer statistics (e.g. either reduce or justify focus on mean based stats), do a bit more model building (e.g. third variables, mediators etc) and investigate if publication bias is present in this literature. Finally, I’d like to reiterate that I’m not against this hypothesis being true, but given how counterintuitive it is and it’s pretty large social implications, I think it would benefit us all if it had a bit more rigour.