Abstract: Aggressive behavior plays an important role in the social, emotional and psychological adjustment of children and adolescents. Children and adolescents with aggressive behavior are prone to violate social moral norms, and even commit crimes in serious cases. It is noted that peer victimization is an important predictor of aggressive behavior. Peer victimization refers to that individuals have experienced attacks by peers, such as physical and verbal victimization, attacks on property and social manipulation. Children and adolescents who have been victimized by peers will characterize the attacker as hostility, which will be generalized to the whole peer group, thus showing more aggressive behavior in future interpersonal communication. Several previous studies have examined the relationship between peer victimization and aggressive behavior in children and adolescents. However, due to differences in research design and inconsistent results, there is still some uncertainty about the relationship between peer victimization and aggressive behavior in children and adolescents, and the moderating effect on this relationship is not fully clear. Therefore, the present study employed a three-level meta-analysis to quantitatively synthesize the results of original literatures to obtain reliable estimates of effect sizes and examined a range of moderators (sample, publication, study design, outcome, and assessment characteristics). Through the retrieval of articles published before October 2020, the current meta-analysis identified 40 studies, with 25,605 participants (range of mean age: 6 ~ 19 years) and 333 independent effect sizes. The funnel plot and Egger's test results suggested an absence of publication bias in current meta-analysis. Analysis revealed a significant positive association between peer victimization and aggressive behavior in children and adolescents (r=0.295, p < 0.001), implying that children and adolescents showed a higher level of aggressive behavior when they experienced higher level of peer victimization. In addition, the present study found a significant moderating effect of peer victimization variable. Compared with physical victimization (r=0.219, p=0.005), the association between relational victimization (r=0.298, p=0.005) and aggressive behavior in children and adolescents was stronger. Moreover, the overall association was influenced by region. Aggressive behavior in children and adolescents was more strongly associated with peer victimization in Asia (r=0.351, p=0.006) than in South America (r=0.149, p=0.006). Study design was also a significant moderator. The association between peer victimization and aggressive behavior in children and adolescents was smaller in longitudinal (r=0.234, p=0.014) than in cross-sectional studies (r=0.339, p=0.014). Finally, the moderator analyses also showed that the informant of peer victimization was a significant moderator. The strength of the association between peer victimization and aggressive behavior in children and adolescents was significantly stronger when peer victimization was reported by teachers (r=0.476, p=0.023) than by peers (r=0.290, p=0.023). In addition, the present study found that the overall association between peer victimization and aggressive behavior in children and adolescents did not differ in strength across gender, age, publication, variable and informant of aggressive behavior. Moreover, to eliminate the multicollinearity between moderators, the current study built a multivariate model by including all significant moderators that had been identified in the bivariate models. The result indicated that at least one of the regression coefficients of moderators significantly deviates from zero (F (12, 316)=3.973, p < 0.001). In sum, the results of the current meta-analysis contribute to a better understanding of the association between peer victimization and aggressive behavior in children and adolescents. These results also provide a reference for future empirical studies on explaining aggressive behavior in children and adolescents.

Key words: children and adolescents, peer victimization, aggressive behavior, three-level meta-analysis, moderate effect