Cyberbullying in context: Direct and indirect effects by low self-control across 25 European countries

Abstract

Random samples of at least 1,000 youth, ages 9 to 16 years, from 25 European countries (N = 25,142) were used to test the salience of low self-control on cyberbullying perpetration and victimization (direct and indirect effects), framed by a cross-cultural developmental approach. Path models, which provided evidence of invariance by sex, tested the hypothesized links among low self-control as well as known correlates, including offline perpetration and victimization, and externalizing behaviours. Results showed positive associations between online and offline bullying behaviours (perpetration and victimization), and, more interestingly, both direct but mostly indirect effects by low self-control on cyberbullying perpetration and victimization; externalizing behaviours had little additional explanatory power. Importantly, multi-group tests by country samples provided evidence of quite modest differences in the tested links across the 25 developmental contexts, despite some observed differences in the amount of variance explained in the dependent measures.

Keywords:

• Cyberbullying
• Self-control
• Deviance
• Problem behaviours
• Cross-cultural

Acknowledgments

Data collection of the “EU Kids Online” network was funded by the EC (DG Information Society) Safer Internet Plus Programme (project code SIP-KEP-321803); this work was supported by Masaryk University and by a Fulbright-Masaryk Distinguished Chair fellowship to the first author to spend the fall semester (2010) in the Department of Psychology and the Institute for Research on Children, Youth and Family at Masaryk University in Brno (Czech Republic). The remaining authors were supported by the Czech Ministry of Education, Youth and Sports (MSM0021622406), the Czech Science Foundation (P407/11/0585), and the Faculty of Social Studies, Masaryk University. We would like to thank Pan Chen at the University of Chicago for her assistance with conducting the multi-level model test.

Notes

¹ We should note that we tested potential age effects on the two dependent measures by adding age as a predictor of them; although it was statistically significant, the size of the effects were very modest (βs = 0.05 and 0.03, respectively), and, more importantly, the addition of age did not materially affect the remainder of the model tested (fit or parameter estimates). Thus, for parsimony, we omitted age in model tests.

² In a final exploratory analytic step, we wanted to understand to what extent our findings might be a function of sample nesting effects; in other words, do we find level 2 variability in the dependent measures, and to what extent do model predictors explain some of this variance. Thus, we examined both an unconditional and a conditional multi-level model in SPSS. Two findings require mention. First, there was very little between-country variability (ICC), namely 0.5% for cyberbullying perpetration and 1% for cyberbullying victimization. Second, we found both fixed and random effects by low self-control on the two dependent measures, and, very importantly, the between-country variance became non-significant once low self-control was added into a conditional model, thus indicating that the very modest level 2 variance in cyberbullying measures was fully explained by low self-control.