Long Form

Github Study Shows Gender Bias

A group of student researchers recently released a study of code hosted on Github.com, a popular website that computer programmers collaborate on, that aimed to understand the gender biases that surround coding. The researchers expected that there would be bias against code written by women, given the fact that the industry’s gender balance is heavily tilted towards men.1

Interestingly, the researcher’s hypothesis did not hold, and they found that code written by women was in fact more likely to be approved by community leaders on Github.com than code written by men. Specifically, the study found that code written by women was accepted 78.6% of the time compared to code written by men which was accepted 74.6% of the time.2 However, this finding only held true when the researchers removed any gender markings from the profiles and code that was submitted. When gender markers were added back in, a gender bias against women was documented. This led to the overall conclusion that women may be more competent at coding but their code faces discrimination anyway.

The researchers were sure to investigate other possible sources for this disparity. Code written by women was accepted more than code written by men across all the top programming languages. Additionally, women were not found to be writing shorter code than their male counterparts, so their contributions were not more likely to be accepted because of their snippet’s length. Women also were not benefitting from reverse bias, whereby coders promote the work of women on purpose to help foster a more inclusive coding community.

The findings highlight the struggles that women face in an industry with a pronounced gender imbalance. This issue, of course, extends into the real world, where women report that they face discrimination in interviews and on the job, with reports of women who feel that they have to work twice as hard as the men at their firm. Anecdotally, hiring managers have told women that they are the most qualified for the position but that they would have to fight to get them on the team because upper management was expecting a man. It is this uncalled-for misogyny that makes even women advise other women against getting into the field because they know the obstacles and boundaries that they will face. In an occupation that is known for its high pay, availability of new jobs and its ability to work from home, this is truly unfortunate. It is widely accepted that diversity is important in creating the best product, whether it is on Github.com or at a company in the real world.

I’ve heard some in the industry say that there is a pipeline issue—that too few women are studying in the field—and while this is true, this research suggests that even at the professional level biases still exist. Companies and communities clearly must do more work to eliminate this discrimination, which ranges from the unconscious prejudice to forthright sexism. While it will not happen overnight, I’m hopeful that studies like this will help raise the public’s awareness of the gender imbalance in the technology field and that the industry will be able to realize improvements in the coming years.