Partner effects and collaborative learning

Individuals interact within small, and often unobservable, groups embedded in broader social contexts, which significantly shape peer dynamics and influence outcomes. Although peer effects are widely studied, their direction and impact are highly variable and context-dependent. This lack of consensus presents difficulties for decision-making in education, workplaces, and other settings, underscoring the importance of addressing subtle aspects of the interaction process effectively. We investigate the impact of partner interactions on learning outcomes in a setting combining collaborative and individual tasks. We conduct a field experiment using the simplest group structure possible—pairs—thus avoiding the endogenous formation of smaller peer groups. In particular, university students were randomly and repeatedly paired to sit together and collaborate during classroom activities. Our findings reveal a negative partner ability effect: an increase in partner ability reduces one’s academic outcomes. This effect is particularly pronounced when paired with more capable counterparts and for male students. Results remain robust across various specifications and sample definitions, among others. We explore potential mechanisms behind these results such as free-riding and discouragement.