Add a new academic study to the growing body of evidence that gender bias is rampant in the world of computer programming.
Computer scientists from California Polytechnic State University and North Carolina State University gathered publicly available information about roughly four million GitHub users who logged in to the service on April 1, 2015. GitHub, a repository for programming code, doesn’t require that users reveal their gender, but the researchers were able to use other information and what they call a “novel gender-linking technique” to identify the gender of just over 35 percent of those users, or around 1.4 million.
An analysis of pull requests, or users’ submissions of new code to the projects of other software developers, revealed that code written by women was accepted 78.6 percent of the time. For men, the figure was 74.6 percent. But when female coders did indicate their gender, they were far less likely to have their code accepted, with their approval rate plummeting to 62.5 percent.
The whole study is quite thorough, and worth a closer look.
The data set contained a disproportionate number of pull requests from men, which is in line with the gender imbalance that pervades computer programming. But the authors say the difference in the rate of accepted requests is statistically significant. They go on to examine a range of hypotheses that might explain what they observed.