Many psychologists embrace the “gender similarities hypothesis” – the view that we tend to overstate the differences between men and women. My colleague Garett Jones pointed out a thorough meta-analysis by Janet Hyde, claiming that – with the exceptions of motor performance and sexuality – gender differences are quite small.

I’m not convinced. The evidence on cognitive abilities seems OK. But personality and preference measures, in contrast, are probably biased downards by reference group effects. If you ask men and woman how “emotional” they are, for example, at least some respondents will interpret it as “emotional compared to other members of my gender.” If everyone did this, a big gap would look like no gap at all; if half of all people did this, a big gap would appear to be half of its actual size. (Yes, you can say the same for “happiness” too, though there are ways of coping with the problem).

Another problem is that Hyde’s meta-analysis includes few studies the compare the genders on work, family, and interest attitudes. Steven Rhoads shows that these are very big.

The biggest problem, though, is that there is so little empirical work on gender identity per se. How often do you wear a dress? Go shopping with your mother? Say “Mad… what makes you think I’m mad?”? When most people say that men and women are very different, I suspect that these are some of the main differences they have in mind. And if you don’t think they’re important, I suggest you try ignoring them for a week, and see what happens!