A minor quibble here, but I think a more relevant natural experiment on revealed employment preference is contemporary Scandinavia, where there's still a lot of (ostensibly voluntary) sex-based employment segregation.

As a more general response to the article and replies, harping on "women-in-STEM" in a US context is sort of odd since women overtook men on earned doctorates in the life sciences 15-20 years ago and in the chemical sciences around a decade ago. There are definitely plenty of problems to go around, but I think the evidence that men with low educational attainment are in a particular bind right now is more compelling than the evidence that there's a similarly acute economic problem facing women. That being said, Mr. Xu's arguments in favor of greater interpersonal connection between men and women are well taken, and increasing sex-based partisan polarization is troubling.