Bayesian network analysis of individual-level factors associated with bullying among high school students

This study endeavors to establish a comprehensive network elucidating the multifaceted factors associated with bullying among high school students, elucidating their intricate interrelationships. We aim to develop a theoretical framework that informs the design of targeted and efficacious interventions to mitigate bullying behaviors.

A cross-sectional survey with a structured questionnaire was administered to 1,401 high school students in Yunnan Province to assess their levels of stress, core self-evaluations, dual mode of self-control, depressive symptoms, and bullying experiences. The data were analyzed using chi-square tests, multifactor logistic regression, and Bayesian network models.

Gender, stress perception, core self-evaluation, the dual mode of self-control, and depressive disorders were statistically significant factors associated with variations in bullying detection rates. Within the directed network analysis, depression severity emerged as a key predictor of relational bullying victimization, while both relational and verbal bullying predicted subsequent physical bullying. In the regularized partial correlation network, core self-evaluation exerted the strongest influence, acting as an indirect mediator in predicting relational and physical bullying victimization by modulating individual depression levels in the directed network.

Gender, stress perception, core self-evaluation, the dual mode of self-control, and depression were significantly correlated with bullying victimization. Depressive disorders emerged as a potential direct predictor of relational bullying, whereas relational and verbal bullying predicted physical bullying. Furthermore, core self-evaluation served as a key mediator within this complex network of factors.