Your reasoning contains no fact specific to women. Thus, it should be valid for me to substitute any other group which is underrepresented in computing:

Given that few retards enter CS, and many who do leave, it would not be surprising to find that the remaining retards who have not been selected out somehow are in fact more skilled than the average non-retard engineer.

It also still doesn't make a >60% women class remotely likely - even if women make up 40% of the top engineers, the odds of recruiting at least 60% women in a 30 person class is only only 0.83%.

(PC note: I am only asserting that women are similar to retards in the sense that khuey's argument applies to both of them. No other similarity is asserted or implied.)