A new study has found that suggestions for improving open-source software code submitted by women programmers are accepted more often than those submitted by men – unless it’s obvious the suggestion came from a woman.
When the gender is clear, the women’s suggestions end up being rejected more often, according to the study conducted by researchers at N.C. State University and California Polytechnic State University.
The peer-reviewed study, published online Monday by PeerJ Computer Science, confirms earlier studies that have shown that gender bias prevails in the open-source community, said Emerson Murphy-Hill, corresponding author of the paper and an associate professor of computer science at N.C. State University.
Open-source software is free software developed by a community of developers who make the source code publicly available and then make improvements based on suggestions that are submitted.
The study looked at more than 4 million “pull requests,” or suggestions, submitted to GitHub, which is the largest open source developer community. Of those suggestions, the researchers were able to use e-mail addresses and social media profiles to identify the gender of those who submitted 35 percent of the suggestions, or more than 1.4 million suggestions.
Overall, suggestions made by women were accepted 78.7 percent of the time, versus a 74.6 percent acceptance rate for suggestions made by men. That, according to the study, is a statistically significant difference.
But the situation was reversed for suggestions from “outsider” programmers – that is, programmers who weren’t involved in overseeing the initial coding – whose gender easily could be identified based on their names or profile pictures. Suggestions from outsider programmers easily identified as women had a 58 percent acceptance rate, versus a 61.2 percent rate for those easily identified as men.
The study also found that the suggestions from women programmers that were accepted involved bigger changes. Women’s suggestions led to a median of 29 lines added and 5 lines removed, versus a median of 20 lines added and 4 lines removed for suggestions from men.
Murphy-Hill said the study shouldn’t be interpreted as concluding that women programmers are inherently better. Rather, their superior record on suggestions – after bias is removed from the equation – likely is a result of other factors, such as less-competent women being more likely to change fields than their less-competent male counterparts.