Two-phase testing assesses individual performance (phase 1) and then allows collaborative learning within small groups (phase 2). While groups typically outperform individuals, less is known about the social decision schemes that influence member collaboration. In a classroom setting, we compared individual and group performance on a standard test versus a two-phase test (Study 1 n = 180, Study 2 n = 60) to determine the impact of collaborative testing on recalling factual and application based test items. We also assessed various aspects of group dynamics (e.g. liking). Both studies showed several benefits to collaborative testing with few costs: students enjoyed collaborative testing and the performance benefits were stronger for lower scoring students and integrative versus factual test items.