Scientific American just published the article Dakota McCoy and I wrote arguing against a misinterpretation of some surprising data. (We included some re-analysis of the data, too.)
Short version: there are more women in STEM in countries with low gender equality (like Jordan) than there are in countries with high gender equality (like Sweden). Some people call it the "gender-equality paradox" and claim it exists because women are naturally less interested in science, so when you let them choose freely, they freely choose other things. We think that's wrong. Tons of research shows that gendered expectations and experiences push young girls away from STEM paths, even in supposedly gender-egalitarian places.
What gender-equality paradox research really shows is that the same gender bias that keeps women out of government/business/etc. is not the same gender bias that keeps them out of physics departments. That's because bias isn't one big boogeyman who causes every bad thing; it's a nasty rat nest of implicit and explicit attitudes that can form and change independently of each other.
I think the broader point is even more important. Some people––who fortunately are a tiny minority––think we've done away with all bias and discrimination, so group differences that remain must be due to innate differences in abilities or interests. That's wrong, and believing it risks making those differences acceptable and permanent. And that's bad for everybody, not just the folks we traditionally think of as disadvantaged. For instance, did you know that girls have gotten better grades than boys for over 100 years? If we assume that's because girls are just smarter than boys, then we start expecting boys to perform worse, and eventually our expectations *cause* boys to perform worse. And that's bad, just like gender disparities in STEM are bad.
(Okay not exactly the short version! But this is complicated.)