[hntrader]

Summers did, read his exact words. He's not only talking about gender balance in science but about top performers specifically.

Also now that you mention it, there are mean level differences at a young age in interests (people vs things), which is also a plausible explanation for gender disparity in STEM, whether that difference is genetic or cultural or both.

[pseudalopex]

The documented median and variability differences are modest. They're enough to explain 20% women at the +4 SD level if I recall correctly. But there are too many people in STEM fields for them to be so selective.

That seems to be in line with what Summers said. Pinker talked about gender balance specifically though.

[hntrader]

> They're enough to explain 20% women at the +4 SD level if I recall correctly

That's right if we're looking at univariable distributions.

> But there are too many people in STEM fields for them to be so selective.

Actually I agree now that the variability hypothesis is insufficient to explain why there's so many more men than women that self-select into STEM.

I think a more plausible explanation is mean differences in interests (which may or may not be genetic).

The variability hypothesis can possibly help to explain things like why most chess champions are men, but it can't explain why most people that play chess in the first place are men.