Gender and Tenure Diversity in GitHub Teams
Citation: Bogdan Vasilescu, Daryl Posnett, Baishakhi Ray, Mark G.J. van den Brand, Alexander Serebrenik, Premkumar Devanbu, Vladimir Filkov (2015) Gender and Tenure Diversity in GitHub Teams. CHI '15 Proceedings of the 33rd Annual ACM Conference on Human Factors in Computing Systems (RSS)
DOI (original publisher): 10.1145/2702123.2702549
Semantic Scholar (metadata): 10.1145/2702123.2702549
Sci-Hub (fulltext): 10.1145/2702123.2702549
Internet Archive Scholar (search for fulltext): Gender and Tenure Diversity in GitHub Teams
Download: https://pdfs.semanticscholar.org/5c02/6313a2bc5e61da4781bab1b5dfd21437e92c.pdf
Tagged: github (RSS), survey (RSS), gender (RSS), open source (RSS), diversity (RSS)
Summary
Background (quote):
- literature tends to agree on three analytical frameworks: similarity-attraction theory (SA) [12], social identity and social categorization theory (SIC) [55], and information-processing theory (IP) [52]. According to SA, people prefer working with others similar to them in terms of values, beliefs, and attitudes [66]. SIC postulates that people tend to categorize themselves into specific groups, and categorize others as outsiders. Members of one’s own group are then treated better than outsiders [55]. Due to greater perceived differences in values, norms, and communication styles between groups than within groups, SA and SIC explain why work-group heterogeneity can lead to confusion, stress, and conflict [31]. Both perspectives suggest negative effects of diversity on team outcomes. In contrast, IP treats diversity as positive: bringing to the table a mixture of cultural/educational backgrounds, and access to different networks and broader information can enhance a team’s creativity, adaptability, and problem solving skills
Sent survey (more fully described in Perceptions of Diversity on GitHub: A User Survey) to 4500 GitHub contributors stratified by gender (1/3 female) and number of projects contributed to. Achieved 19% response rate, of those 24% female indicating lower response rate for women.
Conclusions from survey:
- "participants recognize team-members as those who make any type of contribution to the project"
- "team members are quite aware of certain aspects of other team members, including gender and technical skill"
- "diversity appears to matter to contributors, but the perceived effects appear to vary"
Also extracted data from GHTorrent (recording of all public activity on GitHub) 2008-2013. For 80% that gender could be inferred for, found 91% male, 9% female. Dataset more fully described in A Data Set for Social Diversity Studies of GitHub Teams.
Response variables: productivity (commits in given quarter), turnover
Independent variables: gender diversity, tenure diversity
Control variables: team size, project forks, quarter index (90 day intervals; time moderates effects of diversity), overall project activity (total commit count), project age, tenure median, comments (reflects project social activity)
Hypotheses
H1. Gender diversity has a positive effect on productivity.
Confirmed.
H2. Gender diversity has a negative effect on turnover.
Confirmed.
H3. Tenure diversity has a positive effect on productivity.
Not confirmed.
H4. Tenure diversity has a positive effect on turnover.
Confirmed.
Theoretical and Practical Relevance
Quote from conclusion:
- Our study suggests that on a larger, economic and societal scale, added investments in educational and professional training efforts and outreach for female programmers will likely result in added overall value. Such efforts may have to be coordinated with efforts to remove existing barriers standing in the way of women’s equitable integration in programming teams.
