Student Performance, Peer Effects, and Friend Networks: Evidence from a Randomized Peer Intervention

We estimate the effects of an educational peer intervention in which previously high-and low-achieving students are randomly paired as deskmates in elementary schools in China. Our treatment boosts the mathematics scores of the low-achieving students. Moreover, the treatment enhances the extraversion and agreeableness of high -and low-achieving students, thereby providing evidence that peers can influence personality traits. The positive treatment effects can be attributed to the deskmate-level peer effects. We document friend network structure changes and present additional evidence on how friendship ties help us better understand the mechanisms behind peer effects. (JEL I21, I26, J24, O15, P36, Z13)