Peer Effects in Education: How Might They Work, How Big Are They and How Much Do We Know Thus Far?

This chapter summarizes the recent literature on peer effects in student outcomes at the elementary, secondary, and post-secondary levels. Linear-in-means models find modest sized and statistically significant peer effects in test scores. But the linear-in-means model masks considerable heterogeneity in the effects experienced by different types of students. Using nonlinear models, one prevalent finding is larger peer effects in which high ability students benefit from the presence of other high ability students. Studies that stratify students by race and ability often find that students are affected both by the racial composition of their peers and by the achievement of their same-race peers. At the university level, several studies find modest sized effects from dormmate and roommate background on own academic performance. For both university and high school students, the measured peer effects on "social" outcomes such as drinking are larger than the effects on academic outcomes. Many authors find substantial peer effects in drinking, drug use, and criminal behavior. This chapter suggest areas for future investigation and data collection.