Toxic disinhibition makes a versatile rulebreaker in cyberspace: Mapping online deviance with exploratory graph analysis and structural equation modeling

Deviant behaviors in online spaces are rising at an alarming rate, causing profound psychological and/or financial harm to victims. Previous research suggests that various manifestations of cyberdeviant actions tend to co-occur and may fall into a distinct online problem behavior syndrome. Gaining a deeper understanding of the co-occurrence of these behaviors and their predictors can meaningfully guide future prevention and intervention efforts. The present work explores the co-occurrence of cyberbullying, trolling, and cyberfraud through two complementary statistical approaches: exploratory graph analysis (EGA) and structural equation modeling (SEM). Moreover, the study investigates the role of online disinhibition in predicting the shared variance among these behaviors. Using cross-sectional data from a large adult sample (N = 1321), an initial item-level EGA on the full item pool indicated a three-dimensional structure. However, when applied to a refined item set composed of locally independent items identified through unique variable analysis, EGA revealed a unidimensional network structure encompassing all three behaviors. SEM confirmed their convergence onto a single higher-order latent factor. Toxic online disinhibition emerged as a strong predictor of this factor, accounting for over 60 % of its variance alongside age and gender as control variables. These findings highlight the interconnected nature of deviant online behaviors and suggest that integrated prevention strategies targeting multiple forms of cyberdeviance may offer a promising path forward.