Single-sex schooling and adolescent risk behaviors: quasi-experimental evidence from South Korea.

This study examines the effects of single-sex versus coeducational middle schooling on adolescent risk behaviors using South Korea's lottery-based school assignment system. Exploiting quasi-random variation in exposure to all-boys, all-girls, and coeducational schools, I analyze nationally representative data from the Korean Youth Risk Behavior Web-based Survey and examine alcohol use, smoking, sexual experience, non-medical drug use, violence victimization, and composite measures of risk behavior. The results reveal pronounced gender heterogeneity. Relative to coeducational schools, attendance at all-girls schools is associated with reductions of 0.2-2.5 %e points across all reported risk-behavior outcomes, whereas attendance at all-boys schools is associated with higher reported risk-taking concentrated in specific behaviors: the effects are statistically significant for sexual experience, violence victimization, and an aggregate risk index, but not for drinking or smoking. Placebo tests using predetermined health conditions, together with covariate-balance and Oster (2019) coefficient-stability analyses, mitigate concerns about residual sorting, although the data do not contain exact school-district or school identifiers. The findings provide quasi-experimental evidence that single-sex school environments affect reported adolescent risk behaviors differently by gender. Single-sex schooling is associated with broad reductions in risk among girls but smaller, more domain-specific increases among boys. These results underscore the importance of incorporating gender-specific health effects into evaluations of school organization and peer-composition policies.