**4. Discussion**

The purpose of this study was to fill a gap in the literature regarding the associations between being bystanders of online hate, toxic online disinhibition, and online hate perpetration. To address this aim, data were gathered from a sample of 1480 German adolescents aged between 12 and 17 years old. Notably, 53.7% had observed at least one online hate incident, 11.3% reported having perpetrated at least one incident of online hate, and 16.9% reported being victimized at least once by online hate. Therefore, online hate appears to be a prevalent issue among adolescents that warrants further investigation in the future. Our finding that the majority of students report observing online hate parallels that of Räsänen et al. [1]. These findings also underscore the need for more research into the experiences of adolescents who are bystanders of online hate.

We found support for our prediction that being bystanders and perpetrators of online hate would correlate (Hypothesis 1), even after controlling for the effects of being victims of online hate. Thus, it seems to be important to limit online hate exposure among adolescents. More broadly, our findings align with those indicating that being bystanders and being perpetrators of online and offline aggression is correlated [7,8]. A possible explanation might be that adolescents who observe online hate or perceive that their peers accept it perceive online hate as normal and unexceptional behavior are therefore more likely to perpetrate online hate. Another explanation might be that some adolescents might become desensitized by observing online hate, making online hate seem like a potentially normative behavior.

The evidence showed that, as expected, higher levels of toxic online disinhibition were positively associated with online hate perpetration (Hypothesis 2). This result extends previous research that revealed positive associations between online disinhibition and cyberbullying perpetration, flaming, and trolling [17,20–23]. Nevertheless, personal features like impulsivity, repressed emotions, personal drives, and one's own experiences with exclusion and discrimination may also be important predictors of online hate perpetration, which need to be investigated in future research thoroughly.

We add to the literature that, consistent with expectations, toxic online disinhibition moderated the associations between being bystanders and being perpetrators of online hate (Hypothesis 3). Thus, the online disinhibition effect might be a key variable in understanding why adolescents who observe online hate also perpetrate online hate. More research is needed whether toxic online disinhibition might also moderate associations between other participating roles, such as between victims and perpetrators of online hate. Although this finding sheds light on possible contextual factors that explain the association between being bystanders and perpetrators of online hate, more research is needed to understand whether intra- and interpersonal factors (i.e., desensitization process, social norms within the peer group, popularity of the perpetrator, the subjective nature of the perceived severity of online hate) might also contribute to the correlation between being bystanders and being perpetrators of online hate. Finally, more research is needed to understand whether benign online disinhibition controversially might buffer the association between being bystanders and perpetrators of online hate.
