It is not so difficult as all that: high-stakes tests are conducted all the time and gender is routinely recorded. I refer you to the WP article for how stereotype threat evaporates the moment it would ever matter.

Which studies there are you referring to as being relevant? Note by the way that the study in question isn't quite the same as stereotype threat in the classical sense anyways.

What do you find is bizaare about this and how do you think that undermines the study?

Their results make no sense in almost any causal model of how stereotype threat would work. What sort of stereotype threat has no effect on attitudes and body images and increases interest in co-workers, and how would you expect this to support the argument you made with regard to co-workers in the real world?

I'm not completely sure what model would actually do this but it could be something that causes them to think of themselves less as people doing math and more as people who are socially or sexually interested in others. But the fact that it didn't have an impact on body image is strange, and needs further investigation. In the short version though, that should suggest that this study actually is more reliable: one of the most common criticism of psychology as a discipline is that the studies have way too high a confirmation of hypotheses rate. That's been discussed on Less Wrong before. In this case, the fact that part of the study went against the intuitve hypothesis and went against what the authors explicitly hypothesized is a reason to pay more attention to it.

That's a reason to be skeptical of the results, not a reason to a priori throw them out.

Indeed. So why did you cling to a weak reed?

Because this is one of the very few studies that have looked at how sexualization impacts performance. There are a lot of stereotype threat studies (as you noted) but they don't generally look at this. I'd be happy to rely on something else or change my opinion here if there were that many more studies.

Do you have any additional point beyond which you said to Kaj there?

No. I stand by the sum of my comments: that it is blatant post hoc rationalizations which contradict any theory a feminist would have made before seeing the actual data, which clearly supports an economic rather than pure bias account, and makes false claims about new CS students as well.

So in your view, what precisely is the reason for the fact that the percentage of female CS students was consistently rising and then took a sharp drop-off? Also, why do you think Kaj disagreed with your position?