A study published this week has found that code written by women via GitHub is in fact more likely to be approved and accepted by their coding peers than code written by men. That is, at least when their reviewing peers don’t know they’re female…
The results came as a surprise to the student researchers who had expected their research to confirm that gender bias in software development was overwhelmingly prejudiced against code written by women. While the study paper has yet to be peer reviewed, the results are still intriguing.
The authors examined “pull requests” on GitHub, one of the largest and most well-known open source software development communities in the world. They looked specifically at the behaviour of coders. The researchers studied how women programmers are treated on GitHub, using a variety of different methods.
The researchers found that despite 92% of software developers being men, the small number of female coders were proportionally more likely to have their work approved and viewed favourably on GitHub, as long as they didn’t actually identify themselves as women.
The study’s authors wrote:
“Our results show that women’s contributions tend to be accepted more often than men’s…However, when a woman’s gender is identifiable, they are rejected more often.
Our results suggest that although women on GitHub may be more competent overall, bias against them exists nonetheless.”
The researchers looking, at around 3 million individual pull requests on GitHub, discovered code written by women was approved at a rate of 78.6%. Code written by men however had a rate of 74.6%.
But this figure was found to only be true when female coders gender was not “identifiable.” The acceptance rate for women coders when their gender was noted in their profile, was worse than men’s, dropping to 62.5%.
The paper concludes:
“Women have a higher acceptance rate of pull requests overall, but when they’re outsiders and their gender is identifiable, they have a lower acceptance rate than men.”
The full report of the study can be read, here.