I’ll start with a puzzle. What do the following variables have in common?
Studying a STEM field
Levels of altruism, trust, positive & negative reciprocity, risk taking & patience in economic games
Emotionality & other personality traits
Use of gendered names
Playing chess
Gender of 1st author on academic papers
Subjective well-being of adolescents
To answer the question, they have all been subjects of papers purporting to show a gender equality paradox (GEP). The GEP is the claim that more gender equal societies produce larger sex differences in all of these areas. To take just the first variable above, the idea is that gender equal societies like Sweden have larger male/female gaps in the study of STEM fields, while less gender equal ones like Iran have smaller gaps, they are closer to balance.
This finding has been a relatively hot one in social science. Just witness the papers above, all published since 2018 and with relatively high numbers of citations despite their recency. Yet, the finding strikes me as peculiar in a number of ways that I’ll describe below. I’m still trying to figure out whether this is a case of something real and substantial or just a handful of random correlations. I’ll try to work through my thinking below.
Underdeveloped theory
There is a theory behind the GEP, but it is relatively underdeveloped. The basic idea, sometimes referred to as the evolutionary perspective, argues that more gender equal societies allow the two sexes more freedom to indulge their inherent proclivities. Reference is also made to Inglehart’s postmaterialism thesis which posits that when societies reach a certain degree of prosperity, individuals come to care less about material concerns and more about their various identities.
One of the worries about this theory is that the contrast is to traditional or agricultural societies which tend to be much more patriarchal and are thus likely to have much more gender segregation in most areas of life (see Evans on the great gender divergence). It is true that economic pressures are greater in such societies and thus there is less space for individual choice of activities or general behavior and preferences, but there are also greater social and cultural pressures for conformity to traditional gender roles. That such societies would have smaller sex differences seems at least to me unlikely.
Equality causes inequality?
One of the most striking complications of the gender equality paradox is that at its heart it argues that gender equality creates gender inequality (and presumably vice versa). This comes close to one of Hirschman’s classic reactionary arguments - in particular, his jeopardy argument where a progressive change purportedly undermines other progressive values, though typically the undermined values lie in a different domain.1 Here, somewhat tautologically, some types of equality produce other types of inequality.
Indeed, the GEP starts with a set of gender equalities that define a gender equal society. These typically include relative equality in education, labor force participation, and even politics (these of course are the main dimensions of the UN’s Gender Inequality Index which is often used as the independent variable in GEP studies). Yet, these equalities coexist with and even lead to other inequalities in the areas listed above - i.e., studying STEM fields, playing chess, acting altruistically, etc. It certainly could be the case that certain equalities are causing other inequalities, but it is somewhat peculiar.
Proper domain
Indeed, the question I raised at the start is what distinguishes areas where there is equality from ones where there is inequality. Or more briefly, where should we find the GEP? My sense is that the areas discussed above are mostly activities and behaviors that are stereotypically masculine. This fits things like studying STEM fields, playing chess, risk taking, (non)emotionality, and academic publication. Is this the reason that scholars chose these areas? It is not clear.
Should scholars continue to look at stereotypically masculine behaviors/beliefs? I decided to choose one - participation in the Olympics - and found the opposite of a gender equality paradox. In fact, more gender equal countries sent more sex balanced delegations and appeared to have more sex balanced medal counts. Where should we look next? One place might be existing literature where indices of gender equality are used to predict a variety of other variables. My sense is that in most of these gender equality will be associated with smaller sex differences.
Cross-national data
Most of the GEP results are based on associations with cross-national data. Such studies suffer from well-known methodological problems especially in cases like this one where the key variables - here gender norms and gender inequality - have changed rapidly over time. It is hard to infer causality from such research designs and easy to conflate time trends and cross-national differences. A time-series design might help, but again changes in gender equality are caught up in many broad trends. Ideally, one would consider exogenous changes in gender equality, a sort of natural experiment, but are there any? Perhaps wars as in the case of Rwanda?
Appropriate controls
Many of the GEP papers have been subjected to methodological criticism. In a few cases, there are concerns about the dependent variable (for example, the measure of who is studying STEM fields). Another is that studies do not include the appropriate controls. Berggren and Bergh, for example, claim that the studies suffer from a Simpson’s paradox. That is to say, the GEP is a composition effect and it disappears when including controls for different cultural regions. Within regions, gender equality is negatively associated with sex differences, but across regions the association is positive. The question is then what are the appropriate controls to include in these studies.
Conclusion
What should we make of the GEP? There are relatively few new and counterintuitive findings in social science and the GEP counts as one of them. It is possible that researchers simply stumbled on this one and were excited by its potential without thinking too deeply about the issues noted above. It would be possible to ascribe more nefarious motives - i.e., that if the GEP were true, then gender equality is not all it is cracked up to be - but I am hesitant to go that far. I think the motivation of counterintuitiveness and potentially a desk drawer problem (only positive results are published) are enough to get us to where we are today.
What is necessary going forward are better designed studies (including appropriate controls and potentially exogenous changes) across a much broader range of potential outcomes that would better map the domain of the GEP if it in fact exists. Cowen says that a just released meta-study proves that the GEP is real, at least for personality. Even if this is right, I’d still like to know why it exists.
The other reactionary arguments are perversity (progressive changes actually make things worse) and futility (progressive changes actually have no effect).