Evidence of Class-size Effects on Bullying in Swedish Schools

In this paper we estimate the effect of class-size on the prevalence of physical and verbal bullying in Swedish schools. We use self-reported individual level data from approx. 3 100 Swedish adolescents in the 9th grade (aged 15-16) regarding their experience of bullying in the school environment. The data covers 40 schools containing 159 classes. We run probit regressions, school fixed-effects probit regressions controlling for between-school endogeneity, as well as using an instrumental variable approach controlling for between- and within-school endogeneity. The results indicate, giving the same conclusion in all specifications, that bullying is not less or more prevalent in smaller classes. However, there are some results indicating that in smaller classes there is a higher probability that an adolescent self-identifies as a bully.