Who falls victim to cyberbullying? Insights from linear regression and machine learning model in a socioecological perspective.
Source: PubMed, NCBI / U.S. National Library of Medicine
Adolescent cyberbullying victimization has emerged as a pressing global concern, with complex risk factors spanning across multiple ecological levels. However, few studies have systematically examined multilevel determinants using large-scale, cross-national data. Guided by socioecological theory, this study examined associations between adolescent cyberbullying victimization and a range of individual, family, school, and community-level factors using both linear regression and machine learning approaches. The study drew on data from the 2019 Organization for Economic Co-operation and Development Survey on Social and Emotional Skills, involving 22,478 adolescents (= 15.45,= 0.55; 51.77% female) from 10 cities worldwide. Both analytical approaches consistently identified a set of variables associated with cyberbullying victimization, including gender, grade, smoking, drinking, sleep problems, family material well-being, family conflict, parent-child relationships, parental punishment, school bullying, school climate, school connectedness, teacher expectations, teacher support, peer deviance, friendship quality, neighborhood connectedness, and neighborhood safety. In addition, the Shapley additive explanation analysis highlighted several variables that were less prominent in the traditional regression models, including body mass index and school anxiety. Findings underscore the multifaceted and cross-system nature of cyberbullying victimization among adolescents. Effective prev
Abstract
Adolescent cyberbullying victimization has emerged as a pressing global concern, with complex risk factors spanning across multiple ecological levels. However, few studies have systematically examined multilevel determinants using large-scale, cross-national data. Guided by socioecological theory, this study examined associations between adolescent cyberbullying victimization and a range of individual, family, school, and community-level factors using both linear regression and machine learning approaches. The study drew on data from the 2019 Organization for Economic Co-operation and Development Survey on Social and Emotional Skills, involving 22,478 adolescents (= 15.45,= 0.55; 51.77% female) from 10 cities worldwide. Both analytical approaches consistently identified a set of variables associated with cyberbullying victimization, including gender, grade, smoking, drinking, sleep problems, family material well-being, family conflict, parent-child relationships, parental punishment, school bullying, school climate, school connectedness, teacher expectations, teacher support, peer deviance, friendship quality, neighborhood connectedness, and neighborhood safety. In addition, the Shapley additive explanation analysis highlighted several variables that were less prominent in the traditional regression models, including body mass index and school anxiety. Findings underscore the multifaceted and cross-system nature of cyberbullying victimization among adolescents. Effective prevention efforts should adopt a multilevel approach that addresses risk and protective factors across individual, family, school, and community contexts. (PsycInfo Database Record (c) 2026 APA, all rights reserved).