With the growing prevalence of the Internet among adolescents, online aggressive behavior poses a growing threat to adolescents' mental health. Existing studies suggest that online aggressive behavior is influenced by multiple factors rather than a single factor. However, prior research has relied on cross-sectional designs, examining only the correlation between cumulative ecological risk and adolescents' online aggressive behavior. The longitudinal relationship between these two factors remains insufficiently understood and how the interactions of various risk factors influence online aggressive behavior has yet to be examined. Additionally, these studies often lack a strong theoretical framework. To address this gap, the present study adopts a longitudinal cross-lagged design to explore the longitudinal relationships among these factors, and established a three-dimensional interaction theory to explain adolescents' cyber aggression.

We conducted a two-year longitudinal study. A total of 1,763 adolescents were assessed on macro-social, meso-environmental, and micro-individual factors at two time points. Macro-social factors were assessed with the Social Security Questionnaire and the Violence Exposure Questionnaire. Meso-environmental factors were assessed with the Child Psychological Maltreatment Scale, the Harsh Parenting Scale, the Bullied Scale, and the Cyber-ostracism Questionnaire. Micro-individual factors were assessed with the Moral Disengagement Questionnaire, the Trait Anger Scale, and the Dirty Dozen Scale. Online aggressive behavior was assessed with the Online Aggressive Behavior Scale. Cumulative ecological risk is the sum of the standardized scores of the three indicators. All scales demonstrated strong reliability and validity. Data were processed using SPSS25 and Mplus8.3, and analyses included ANOVA, correlation analysis, and cross-lagged analysis.

The results showed: (1) All variables were significantly correlated with each other at both measurement points. (2) T1 cumulative ecological risk significantly predicted T2 adolescents' online aggressive behavior. (3) T1 macro-social, meso-environmental, and micro-individual factors each significantly predicted T2 online aggressive behavior. (4) The interaction between T1 macro-social risk factors and T1 micro-individual risk factors significantly predicted T2 online aggressive behavior. (5) The three-way interaction among T1 macro-social, meso-environmental, and micro-individual risk factors significantly predicted T2 online aggressive behavior. Common method bias was examined using the single-method latent variable approach. For both measurement occasions, the model without a common method factor showed good fit, whereas model fit deteriorated when the common method factor was added. This indicates that common method bias is not a serious concern in this study.

In summary, based on the findings, we propose a three-dimensional interaction theory of adolescents' online aggressive behavior. We posit that online aggressive behaviors are primarily influenced by additive and interactive effects across three ecological systems: the macro-level social system, the meso-level environmental system (encompassing family, school, and online environments), and the micro-level individual system. This theoretical framework offers new explanations and theoretical support for interventions targeting adolescent maladaptive behaviors and contributes to the maintenance of adolescent mental health.