Inconsistencies of prevalence rates of (cyber)bullying and (cyber)victimization in large-scale cross-national datasets: Country trumps definition

Abstract

Previous research comparing different large-scale cross-national surveys has shown substantial discrepancies in (cyber)bullying and (cyber)victimization rates within countries. This study hypothesized that systematic differences in bullying definitions might explain these discrepancies. Prevalence rates of bullying, victimization, cyberbullying and cybervictimization of two large-scale surveys, EU Kids Online (EUKO) and Health Behaviour in School-aged Children (HBSC) that used two types of bullying definitions were compared using data from 123,206 students (aged 10–16 years) in 19 overlapping countries. Contrary to our hypotheses, multilevel Bayesian logistic regression analyses in Mplus and Bayes factor calculations in R showed no evidence of systematic differences in prevalence rates between EUKO and HBSC within countries, either before or after statistically controlling for sample characteristics and study features. Instead, prevalence rates generated by EUKO and HBSC varied unsystematically within countries, highlighting that within country differences produced by EUKO and HBSC should be interpreted with great caution.