Women Write Better Code Than Men, Say Statistics
But they get their code rejected more on GitHub when their gender is known, a study finds.
As if we needed any further evidence of gender bias in the tech community, a group of researchers has now quantified it.
The findings of a study out of Cal Poly and North Carolina State University suggest that female coders are better at their jobs than their male counterparts. By looking through a large subset of the 4 million members of open-source coding community GitHub who logged in on April 1, 2015, researchers identified that pull requests — suggested changes in someone’s code — are more likely to be accepted from women than from men. Statistically, 78.6 percent of pull requests by women coders were accepted, versus 74.6 percent for men.
When open-source code nerds identify a way to solve a problem more elegantly or efficiently than the original author, they file a pull request as a suggested improvement. It’s up to the original author if those changes are implemented or not, and thus we have the paradigm studied by the researchers.
Ah, but then the rub: Those female-authored pull requests are even more likely to be accepted when it’s not known that they are coming from a lady. Github doesn’t require members to identify their genders, but these researchers were able to identify the gender of 1.4 million of the people active on the site that day by either looking at their GitHub profiles or by matching their email addresses to their Google+ profiles. Hold your “meddlesome researcher” privacy concerns — this info is publicly available. The researchers gathered it en masse, but they will certainly not be publishing it in aggregate.
What they did publish, however, is far from heartening. According to their findings, a woman using a gender-neutral GitHub profile will see her pull requests accepted 71.8 percent of the time. If her profile announces her gender, this figure drops to 62.5 percent. The researchers write that there is “a similar drop for men, but the effect is not as strong.”
Put another way, on an equal playing field where skill and ability are the only things considered, women were better coders. On a playing field that more closely resembles the real world, where identity and gender are considered, women’s changes to code were accepted less often.
We only need look as far as tech giants like Google and Facebook to see the technology sector echo this back to us. Google employs a technical workforce of 18 percent of women. Facebook’s technical staff is just 16 percent female.
The GitHub research paper is currently awaiting peer review. In the meantime, how does one file a pull request with reality?
