Exploring the Impact of Gender Bias on Pair Programming

Pair programming, two partners working on a programming task together, is an effective tool for teaching computer science (CS), as measured by performance, confidence, and improved retention in CS programs [4]. These positive effects are especially impactful for women [9, 11]. In pair programming, mutual student engagement is key. But what makes a good pair? Gender affects the experience of (any pairing of) students due to phenomena such as implicit gender bias (e.g., assuming a woman will be less technically competent than a man). Previous work has found conflicting results about whether same-gender or mixedgender pairings are more effective [2, 3, 6, 8]. One explanation is that gender correlates with other dimensions that may affect collaboration, such as relative skill level, personality traits, or existing friendships [1–3, 5, 7, 10, 13]. However, it is not feasible to control for these other factors in a between-subject study design. We propose an IRB-approved within-subject methodology to gain insight into the effect of the perceived gender of a partner and the actual gender of a partner (Figure 1). This allows us to separate effects of factors such as implicit gender bias, which rely on perceived gender, and larger systemic factors, which affect people based on their actual gender. For perceived gender, we acknowledge the current study focuses on binary gender roles.