The tendency that female developers have a higher contribution rate than men was found from GitHub data analysis

ByGokhan Altintas

For women's written code it is harder to review than the code written by men, but of the ones reviewed, the percentage of women's written is higher for code written for women Shared web service "GitHub"Data analysis revealed.

Gender bias in open source: Pull request acceptance of women versus men [PeerJ PrePrints]
https://peerj.com/preprints/1733v1/

Data analysis of GitHub contributions reveals unexpected gender bias | Ars Technica
http://arstechnica.com/information-technology/2016/02/data-analysis-of-github-contributions-reveals-unexpected-gender-bias/

In analyzing data, GitHub does not disclose the user's gender, so researchers first started with a survey to locate the user's sex. The research team focused on people using Gmail among all the users and said that they identified gender by using the Google + profile. In this way, it seems that 35.3% of the approximately 4 million users were identified.

Although GitHub has a function called "pull request" that a project leader reviews code written by himself, the researchers said that "women are less salary than men and have to prove their own values ​​more Based on the results of past research, "We expect that few people will accept contributions by reviewing women's pull requests at project leaders." As a result of survey, it is exactly true for open source projects, and female coda has very little contribution when only looking at numbers.

ByHackNY.org

However, in terms of the percentage of "merge" that reflects the content of pull requests sent by the project leader in the project, the merger rate of the code written by men is 74.4%, whereas the code written by the woman The merging rate of 78.6% was recorded high numbers.


In addition, 25% of females showed 100% pull request approval rate, while for men 100% pull request approval rate was 13.5% as a whole, which also turned out to be significant difference.


For this reason, the initial researchersSurvival biasIt seems he thought that it was not. However, if the survival bias, the survivor female coda should participate in the open source project more than men, but he did not necessarily get such a result.

Researchers finally think that 'male bias' exists in open source projects. So it was investigated whether the acceptance rate will change depending on whether the leader of the project knew the sex of the person who sent the pull request.

The result is as follows. The graph on the left is when the person who sent the pull request was a person who knew of the project leader, when the right was a person who knew nothing at all. Looking at the time when the sender of the pull request was known, when there is no gender unknown, there is hardly any difference between males and females, and on the contrary when the gender is known, the female approval rate is a little higher. On the other hand, if a request is sent from a person who does not know at all, the female approval rate is high when the sex is unknown, but if you know that it is a woman you will find that the approval rate is low.


In addition, it is indicated that the higher the approval rate of women as a whole, it is when the leader does not know the sex of the partner or trusts the other woman.

In the merger rate, researchers commented on why females were better than men, "Researchers said" In the harsh environment, women who are not excellent were culled and only excellent women remained . On the other hand, women who participate in open source projects have a master's degree or doctorate degree compared to men, while mentioning survival bias as one of the possibilities that men are still not excellent people remain Many people,Self-selective biasIt also shows the possibility of existence.

As genetically it is certain that some social bias is working because the fact that "women are superior to men in computer science" has not been discovered, male "excellent women recognize Ars Technica of the news site commented that the fact that there was no gender bias named "There is no hope" is a hopeful light.