It seems like you did not even read the article...

> He ignores that the effect remains in graduate-level math studies

I'm not sure what "effect" you're talking about, but he points out that graduate-level math programs have about the same gender disparity as CS (explained by his theory because like you say, no one gets a PhD to teach and thus the studies shift from being about people to being about things.)

I don't know about biochem, but he explains increased enrollment in biology over other STEM fields because it's about people and/or animals, which the "women like people, men like things" theory predicts.