Bullying is a growing public health concern for South Korean adolescents. An important micro-level factor highly influential in explaining bullying experiences among adolescents is a child’s school environment. The school environment, or social milieu, includes a child’s peers and teachers who can provide supportive relationships that can potentially mitigate bullying’s harmful effects. Yet, one feature of school environments remains less well understood: the gender composition of schools. Importantly, whether the gender context of schools matters for bullying is an unresolved question lacking solid empirical footing. This is critical to understand given Korea’s national school assignment policy known as the equalization policy (EP) that randomly assigns children to either coeducational or single sex schools. Importantly, if a school’s gender composition impacts bullying, then Korea’s school assignment policy may have far-reaching unintended negative or positive consequences, particularly on the types of bullying-related behaviors adolescents engage in and/or experience.

In our quantitative investigation, we analyze the frequency with which Korean adolescents in single-sex versus coeducational schools experience and engage in four bullying behaviors (verbal, indirect/exclusion, theft, and physical). We use two nationally representative datasets, the 2011 Trends in International Mathematics and Science Study (TIMSS) and the 2005 Korea Education Longitudinal Study (KELS). Although under the EP, students are randomly assigned to different schools, it is important to note that random assignment occurs within districts (also known as enrollment zones) with varying distributions of coeducational and single-sex schools. Often, districts offer only one type of school per district. Given this, we believe that assignment into school type is not completely exogenous, limiting our ability to make causal inferences about the impact of school gender composition on bullying outcomes. Therefore, we rely on propensity score matching (PSM) to further reduce the threat of selection bias. Since our bullying outcome measures are ordered categories with arbitrary distances between categories, we estimated ordered logistic regression models. Using data weighted with the estimated propensity score, we fitted models separately for our sample of boys and girls, regressing each of our bullying outcomes on our main predictor variable—the dichotomous variable indicating assignment into single sex or coeducational school.

We find strong evidence consistent with the extant empirical literature on school gender composition and bullying: girls in all-girls schools are less likely to experience verbal bullying, theft and physical bullying compared to girls in coeducational environments; at the same time they are also less likely to be perpetrators of these same bullying behaviors when in single-sex environments. This lends credible evidence supporting prior findings that all-girls environments tend to lack a strong “culture of bullying” and, in fact, can mitigate both the incidence and perpetration of bullying. For boys, reported frequency of bullying victimization (particularly for verbal bullying, theft, and physical bullying) is higher in all-boys schools versus coeducational schools; furthermore, with the exception of verbal bullying, perpetration does not significantly vary between boys in all-boys schools versus coeducational schools.