If that is the study I remember, I don't think it shows women are deemed less capable. It might as well be possible that women are deemed less likely to follow through with their studies (for example because they might drop out to have kids), making them a worse investment.

That's of course still a problem, but a different one than you make it to be.

I think if there is an actual issue like that (if women are a less promising investment), rather than crying "sexism", society should compensate for it somehow. Why should it be the burden of employers to insure against women's life choices?

There is even stronger evidence that there is a 2:1 hiring advantage for women in STEM fields[1]. That study first looked at real-world hiring rates, which showed this 2:1 advantage and then tried to see whether the advantage was due to the women applicants being generally more qualified (which would obviously explain the empirical finding). The study went on to show that superior qualifications were unlikely to be the cause for the real-world hiring advantages.

Note also that the departments selected in the study you cite were Psychology, Biology, Chemistry and Physics, not tech (Computer Science, Electrical Engineering, Engineering in general). In Biology women are the majority, in Chemistry the ratio is close to 50/50. Physics is close to Tech at around 20-25% and in Psychology women are the vast majority of students and practitioners at around 75%.

Therefore, this study shows, if anything, that there is no link (causal or otherwise) between the discrimination it purported to find and participation rates.

Also note that the study you cite looks at sending the exact same resume and changing the name. Another study (that I can't find right now) showed that if the qualifications are noticeably different, gender and other factors play no role. People are competent at spotting differences in qualification and will select on merit if possible. So whatever discrimination there may be (in whichever direction) apparently only plays a factor if the resumes are carefully calibrated to be the same, an unlikely scenario in real hiring situations.

Another issue I had with the study you cite is that they write that the resumes were identical apart from the name, so not real-world resumes. I could imagine that male and female resumes are different in practice, and a female name on a "male-style" resume would look odd, as would a male name on a "female-style" resume.

Some people want to believe that tech and STEM are this maelstrom of sexism, but the evidence doesn't support that...

[1] http://www.pnas.org/content/112/17/5360.abstract