tl;dr: are the results really as surprising as the text suggests?

You state

prior work on gender bias in hiring – that women tend to have resumes less favorably evaluated than men (5) – suggests that this hypothesis may be true.

and immediately after

The hypothesis is not only false, but it is in the opposite direction than expected; women tend to have their pull requests accepted at a higher rate than men!

As I understand it, the reference (5) (Davison & Burke 2000) surveys primarily studies where gender was visible to the evaluators. In contrast to studies like Futoran & Wyer 1986, where ambiguous gender was explicitly included, where the gender bias was shown to decrease.

A common observation in feminist circles, or «Feminism 101» type texts is that women (especially in male-dominated technical fields) have to work much harder than men in the same fields to receive remotely similar acceptance.

For open source development (for which GitHub is a common platform), these two factors seem to naturally interact; anyone still in the game enough to send out pull requests would have maintained their interest in programming through the added exhaustion of having to work harder for acceptance (if not under their gender-neutral alias, commonly in their day-to-day activities), and in the face of pervasive sexist attitudes. At the same time, there are already reason to believe that this gender bias decreases with a gender neutral presentation.

Since your design was to pull gender markers from alternative sources, it seems like a very reasonable result that pull requests from women may well be of better quality (because of the pressure to perform flawlessly), and at the same time escaping gender bias (whenever women have a hard-to-gender identifier).

. : . . : . . : . . : . . : . . : . . : . . : .

Is your result really as dramatically unexpected as your text on page 7 would indicate?