Persistent Classmates: How Familiarity with Peers Protects from Disruptive School Transitions

Students' social networks are deeply disrupted during school transitions and students start in a classroom environment where almost all their peers are new. In this study, we investigate the consequences of keeping partly the same classmates during the transition to high school. To overcome the issue of endogenous selection across classes, we exploit rare natural experiment settings in which students are plausibly randomly allocated to classes within high schools. Two estimation strategies are presented and provide the same results. We find that each classmate who was already in a student's class in the last grade of middle school reduces substantially the risk of grade retention in 10th grade, but also in following grades. For low-ability students, the effect amounts to minus 1 percentage point per "persistent classmate", without increasing the risk of dropping out. A number of robustness checks are provided. By analyzing the distribution of the effect, we show that it is the strongest for students who are the most likely to experience a difficult transition, i.e. low-ability, low-SES students from low-quality middle schools. The underlying mechanisms are examined. Our results suggest that grouping students who already know each other during school transitions would constitute an efficient, no-cost policy lever to improve overall achievement and equality in high schools.