I will stop beating this into the ground after this post. I promise.
But I finally figured out what’s been bugging me about that study: It looks at the difference (rather, lack thereof) in math scores, but ignores differences in visuo-spatial ability. (One way to measure visuo-spatial ability is to test a subject’s ability to rotate an image in his or her head). A decade ago, the American Psychological Association noted a gap of about a standard deviation between the male and female average in visuo-spatial skills.
On PBC, we’ve already documented that while average math scores are the same, men are more likely to score very high or very low, so a college hiring very-high-scoring-on-math professors will disproportionately hire men. My point here is that if a department also requires very high visuo-spatial skills (physics, engineering), that will cut against women as well.