I admit I only read part but not all. That gets me beyond most posters so here we go..
Point is that if they have a conclusion that whether one knows if the code comes from a woman or not matters, two things seem necessary:
- That they figure out who are actually women. As you say, you have to use sources outside the ones that the people deciding whether to accept the code were presumably using. If they didn't have seperate sources, there would not be the two groups to distinguish between.
- That they figure out whether the people accepting the code believed them to be women. My assumption (wrong as another poster pointed out) is that in that sort of environment there would be no reason for the people accepting the code to know unless the person submitting the code somehow informed them.