Women coders are rated better than men in blind studies. However, when the blindfolds are off, they're rated lower than their male counterparts for some reason — a new study details that reason.
A joint study from Cal Poly and North Carolina State University's computer science departments has revealed the gender bias associated with coders. They packaged their findings in a non-peer reviewed study entitled "Gender Bias in Open Source: Pull Request Acceptance of Women Versus Men."
It's the largest study on gender bias to date, the researchers have stated in the report in which they compared women and men coders in open-source community and code repository GitHub.
"Surprisingly, our results show that women's contributions tend to be accepted more often than men's," the study's report said [pdf]. "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 study examined more than 3 million code contributions, known as "pull requests," from men and women on GitHub. They found that pull request by women were accepted more than those made by men, about 78.6 percent to 74.6 percent, so they examined several theories as to why this may be the case.
They explored whether or not the results were skewed by submission from a handful of "highly successful and prolific women," and they looked into the role familiarity played in the findings. Do women typically only tackle projects in which they're highly versed or are they making pull requests that are more critical?
Well the researchers have found that, in general, women's pull request were less likely to be as urgent, their modifications were bigger, the high acceptance rates were true across all programming languages, their acceptance rates climb with experience and they are still more likely to have their requests accepted than males.
The best answer appears to be that women make more realistic assessments of their capabilities. While women tend to drop out of STEM field more frequently than men, those who stick around, and become good enough to contribute to the likes of GitHub, form a more talented pool than the group made of both prolific and mediocre ones.
Essentially, bad coders are dragging down the performance of males. And then there's just flat out gender bias. Women are often held to higher standards than men, which is why their approval rating drops when their gender is revealed.
