Violent Peers at School: Effects on Achievement, Behavior, and Mechanisms

This paper examines how exposure to new middle-school peers who attended a different primary school and previously reported involvement in physical fights affects students’ academic outcomes, and explores potential mechanisms. Measuring peers by their own reported prior behavior differs from most earlier studies, which proxy for disruptive peers using family background or diagnoses. For identification, we exploit idiosyncratic variation in classroom composition among students who share the same primary-to-middle-school transition path. We find that a higher proportion of new violent peers lowers test scores across subjects, with larger effects in mathematics and English than in Hebrew and science. A one-standard-deviation increase in the share of such classmates lowers average test scores by about 0.06 standard deviations. The effects are larger for girls than for boys, although the difference is not precisely estimated. Exposure also worsens students’ reports of the classroom learning environment and increases the likelihood that other students, particularly girls, report involvement in physical fights. These results remain after adjusting for multiple hypothesis testing. We find weaker, suggestive evidence of reduced homework time among girls and students from low-SES families, no detectable effect on private tutoring, and non-causal, imprecise associations with teacher outcomes.