New research from Cal Poly San Luis Obispo and North Carolina State University suggests that women may be better at writing code than men.
The study looked at data from GitHub and found that code changes suggested by women are more likely to be accepted than changes suggested by men, but only when the woman’s gender was not publicly available.
In cases where a user’s gender was publicly available, code suggestions made by women were more likely to be rejected.
The authors of the study wrote that the data suggests that a bias against women exists despite the overall competency of women programmers on GitHub.
“Surprisingly, 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.”
BBC News reports that the study is still awaiting peer review.
GitHub is a publicly available, web-based Git repository where members are able to upload their code. Other users are able to view the code and suggest revisions, which are known as pull requests. Although anyone can suggest a pull request, the owner of the project has to accept, or merge, the pull request for it to go through.
Students at Cal Poly San Luis Obispo and North Carolina State University looked at pull request data from GitHub to determine whether any gender bias could be detected, in a PeerJ preprint that has not yet been peer reviewed.
The 12 million users of GitHub are not required to provide a gender when making an account, but the authors of the study used various means to determine the genders of 1.4 million users.
According to the Daily Dot, user profiles from GitHub were cross-referenced with Google+ and other sources to determine gender, but information they found on individual women will not be released due to safety concerns.
With gender information for 1.4 million GitHub users, the study looked at 3 million pull requests. According to that data, pull requests from women had a 78.6 percent merge rate, while requests from men had a 74.6 percent merge rate.
According to the Daily Dot, the researchers attempted to explain why women were more likely to have suggestions accepted on GitHub by looking closely at the actual pull requests. They checked to see if women were making more urgent requests than men, or if women were submitting more minor changes, but they found no such correlation.
In fact, they found that women are likely to have larger pull requests than men and are less likely “to serve an immediate project need.”
When the researchers looked at requests where the user’s gender was easily identifiable on GitHub, the numbers changed.
According to the study, the acceptance rate was much lower in cases where a woman’s gender was identifiable via their GitHub profile and the woman was an outsider, meaning that she was not involved with the project in question. In these cases, women only had a 62.5 percent pull request merge rate.
“Our results suggest that although women on GitHub may be more competent overall, bias against them exists nonetheless,” the student researchers wrote.
Both men and women in the outsider group showed decreased acceptance rates when their gender was available, but the drop off was sharper for women.
The authors of the study do admit that their results could be skewed if any of the users they looked at were actually robots, or if men and women misrepresent their genders at different rates. For instance, their study could have mislabeled men as women on GitHub, and vice versa, based on their Google+ profiles or other outside data.
If the findings of the study are valid, they appear to suggest that women are as competent as their male colleagues, in the realm of coding, although they likely face an inherent gender bias in regard to their work.
According to BBC News, women are underrepresented in the field, with women accounting for only 16 percent of Facebook’s and 18 percent of Google’s tech staffs as of 2015.
Do you think women really are better at coding than men, or is there another explanation that the authors of the GitHub study didn’t explore?
[Image via Shutterstock/Lucky Business]