Ilmu Komputer & AI editorial
Open AccessOA2022
Gender Differences in Collaboration Patterns in Computer Science
This study analyzes gender differences in collaboration patterns across computer science subfields, finding that the gender gap varies from 6% to 42% female authors and correlates with lower female productivity, larger coauthor groups, and higher gender homophily.
Josh Yamamoto; Eitan Frachtenbergยท Publ.ยท 2022ยท DOI 10.3390/publications10010010
The core problem
Computer science (CS) exhibits a well-documented gender disparity, with women comprising approximately 15โ30% of publishing researchers. Prior research has explored various factors contributing to this gap, including differences in collaboration patterns. This study extends that knowledge by examining collaboration patterns specific to subfields of CS. The authors aim to quantify the gender gap across subfields and relate it to five collaboration metrics: female productivity, coauthor group size, gender homophily, single-author papers, and others. The central hypothesis is that subfields with larger gender gaps may also exhibit less favorable collaboration patterns for women.
Innovation
The authors curated a dataset of nearly 20,000 unique authors from approximately 7,000 top conference papers published in a single year. Each conference was manually assigned to a field and subfield, and gender was assigned to most researchers (likely via name-based inference). The gender gap was measured as the proportion of female authors in each subfield. Five collaboration metrics were computed per subfield: (1) female productivity (average papers per female author), (2) average coauthor group size, (3) gender homophily (tendency to collaborate with same-gender authors), (4) proportion of single-author papers, and (5) possibly other measures. The relationships between the gender gap and these metrics were analyzed across subfields.
Introduction
Computer science (CS) exhibits a well-documented gender disparity, with women comprising approximately 15โ30% of publishing researchers. Prior research has explored various factors contributing to this gap, including differences in collaboration patterns. This study extends that knowledge by examining collaboration patterns specific to subfields of CS. The authors aim to quantify the gender gap across subfields and relate it to five collaboration metrics: female productivity, coauthor group size, gender homophily, single-author papers, and others. The central hypothesis is that subfields with larger gender gaps may also exhibit less favorable collaboration patterns for women.
Why it matters
The results indicate that the gender gap in CS is not uniform; subfields like theoretical CS have extremely low female representation (6%), while CS education approaches parity (42%). The correlation between the gender gap and collaboration metrics suggests that subfields with fewer women may have environments less conducive to female productivity and more prone to homophily, potentially reinforcing the gap. However, the lack of association between single-author papers and female ratio implies that solo authorship is not a primary driver of the disparity. The authors discuss possible explanations, such as differences in subfield culture, resource allocation, and networking opportunities. They also acknowledge limitations, including the use of name-based gender inference and the focus on a single year. Future research should explore longitudinal trends and causal mechanisms.
Who should read this
Pembaca Research Digest dan praktisi rumpun ini.
Opening member contentโฆ