Do Peers Matter for Adolescents’ Risky Behaviors? Peer Group Effects from Random Assignment in Korean High Schools

On the basis of the unique setting in Seoul, Korea where middle school graduates are randomly assigned into high schools within school districts and also where the assignment of high school students into homeroom classes is likely to be random, we address whether peers matter for adolescents’ risky behaviors (drinking and smoking). These procedures of student assignment provide an exceptional opportunity to estimate the effects of peer groups, which are not contaminated by an adolescent’s selection of peer groups. Using the sample of Seoul high school students from a nationally representative survey of Korean high school students in 10th – 12th grades conducted every year since 2005 to 2011, we model an adolescent’s initiation of drinking or smoking in time t conditional on no initiation though through t-1, by the average percentage of her (his) current classmates who undertook the risky behavior p years earlier (i.e., a lagged group variable), after taking into account a variety of individual-level variables. Our school-fixed effect models, furthermore, control for unobserved school environments that may simultaneously affect the adolescent and her (his) classmates. Our study significantly improves estimates of the effects of peer groups on adolescents’ risky behaviors.