Learning and Peer Effects

Abstract

The paper documents the results from an experiment on peer effects in learning processes. The experimental approach eliminates endogeneity and selection problems which typically restrict the measurement of peer effects with administrative or survey data. The results provide evidence on the mechanisms of peer interaction and on optimal learning group composition. Learning with a partner is more successful than learning without one, though differences between gender and age groups are observable. High ability students provide benefits to other high ability students. Subjects who are member in a club provide a benefit to partners who are not.