Woman at a computer.

Women outperform men in coding, but their success is still stifled by gender bias, according to a new study. Nullplus/Getty Stock

Women are considered better computer coders than men, but only when they hide their gender.

At least that's the essence of a new study published this week.

Researchers analyzed data from GitHub, a San Francisco-based open source software community with more than 12 million users who collaborate on coding projects by suggesting solutions to various problems. By tracking users using social networks and Google, the researchers were able to obtain the gender of 1.4 million users, which allowed them to assemble the largest scale study of gender bias to date.

The team examined whether men and women were equally likely to have their coding suggestions, known as a pull request, accepted on GitHub. They hypothesized that pull requests made by women were less likely to be accepted than those made by men. After all, the number of women in computer science are dwarfed by the number of men – according to the U.S. Bureau of Labor Statistics, 73 percent of the field is comprised of men.

But they found the exact opposite: Women actually have their requests accepted at a higher rate than men, 78.6 percent of the time compared to 74.6 percent of the time for men.

Histogram of mean acceptance rate per developer for women and men.

Courtesy of Peer J Terrell J, Kofink A, Middleton J, Rainear C, Murphy-Hill E, Parnin C. (2016)

And overall, women's acceptance rates dominated over men's for every programming language in the top 10.

Pull request acceptance rate by file type, for programming languages (top) and non-programming languages (bottom), for men and women.

Terrell J, Kofink A, Middleton J, Rainear C, Murphy-Hill E, Parnin C. (2016)

However, they found significant gender bias when their coding requests were made to outsiders, as in users who they had not been working with directly prior to making their request.

If women made those requests to an outsider and their user profiles did not identify them as women, the requests were accepted 71.8 percent of the time. But if they made the requests using a profile that identified them as women, the acceptance rate dropped to 62.5 percent.

Pull request acceptance rate by gender and perceived gender, with 95% Clopper-Pearson confidence intervals, for insiders (left) and outsiders (right).

Terrell J, Kofink A, Middleton J, Rainear C, Murphy-Hill E, Parnin C. (2016)

A similar drop occurs for men, but the effect is not nearly as strong.

The researchers were quick to note that their study does not definitively provide that differences between gendered interactions are caused by bias among individuals, but the trend is "troubling" nonetheless, they said, especially in a field like open source software, which is often praised for being a meritocracy.

Biases against women in the workplace have been documented in several studies.

Indeed, a separate study also published earlier this month showed a similar gender bias: Among 1,700 students in undergraduate biology classes at the University of Washington, men overwhelmingly chose men when asked to nominate the best students in the class, revealing a gender bias among male students that was 19 times the size of that of female students.

The Obama administration has taken a keen interest in getting more women and young girls involved in computer science and coding. The president's fiscal 2017 budget, for example, included a $4 billion proposal to make computer science and coding classes a "basic skills" in K-12 schools and to increase access to such programs for female and minority students.

According to data in the 2015 U.S. News/Raytheon STEM Index, male students expressed far more interest in pursuing engineering and technology than females, which shows students’ reported interest in different STEM fields. And after high school, with few exceptions, women lag behind men in the number of STEM degrees granted, exam scores and general interest in the STEM fields.​

Of course, the push by the administration is as much about economics as it is about erasing gender stereotypes. The Bureau of Labor Statistics projects a 37.5 percent growth by 2022 in the "computer systems design and related services" industry – from 1.6 million jobs in 2012 to more than 2.2 million in 2022.

But only 1 in 10 schools nationwide currently are teaching computer science classes, according to the Information Technology and Innovation Foundation. And Code.org, a computer science advocacy group, reports that 9 out of 10 schools don't offer computer programming coursework.

Tags: gender bias, STEM

Lauren Camera Staff Writer