Women Coders Surpass Male Peers In GitHub Blind Test But Gender Bias Is Clear

If you're not affected by it, and don't know anyone else who's affected by it, it's easy to brush the issue of gender bias under the rug. The reality, however, is that it's an issue that's proven over and over again to be a serious problem. Apparently, it's even a problem in the programming world.

Woman On Notebook

Six researchers from Cal Poly and North Carolina State University banded together to see the effects of gender bias on GitHub, the world's leading open source code-hosting website. GitHub's structure allows project members to submit "pull" requests from the project leader to gain permission to overwrite existing code or add some new code. A simple feature, and one that you wouldn't think would be that affected by gender bias. Code is code, right? Yes - but that doesn't matter if someone is sexist.

GitHub

Looking closer at profiles on GitHub that reflected the gender of the developer, it was found that women submitters are accepted much less often than males. That in itself could be arguably discredited as a fluke, but in an effort to determine if it was an anomaly, there was an additional test condition the research team put forth.

For the second test, the researchers sought out accounts of those who didn't specify a gender but made a lot of commits. Based on simple research, the genders of these accounts were noted (based on the profile picture among other things). Accounts where no gender could be assigned without a benefit of a doubt were left alone. Ultimately, what was found was that in this particular case, female submissions were accepted more than male submissions - 78.6% vs. 74.6%.

The researchers make clear the fact that women submissions do not differ from male submissions in terms of quality or the amount of work that went into each one. Based on the sheer amount of information poured over, it's hard to disagree with the conclusion: if you are a woman making a pull request on GitHub, your chance of it being accepted is apparently automatically lower.

This appears to be a sad fact (at least in terms of this study) and it certainly doesn't seem right at all. 


Via:  PeerJ (PDF)
Show comments blog comments powered by Disqus