Tackling (misleading) incivility online: a user-centric evaluation of different comment moderation strategies

Uncivil online comments, e.g., in the form of insults or misinformation, come with severe consequences for platforms, users, and democratic processes – making effective moderation essential. However, there is a limited understanding of user preferences regarding the moderation of such comments that violate different communication norms while remaining protected by freedom of speech laws. This study addresses this gap by conducting a factorial survey experiment with a representative sample of Austrian Internet users (N = 572). It investigates how various factors – i.e., the type of incivility (relation vs. information norm violation), the source of moderation (human vs. AI), the level of transparency (explanation: yes vs. no), and the level of interactivity (deletion, warning labels, guideline reminders, counterspeech) – affect moderation acceptance. The study also considers individual differences, such as users’ views on censorship and political identity. The results reveal that users generally favor comment deletion for insults but prefer more interactive moderation methods, such as counterspeech, to address misinformation. Moreover, moderation acceptance is higher for human moderators than AI. Users’ opposition to censorship and political identity further significantly impact their moderation preferences. Overall, these findings provide evidence-based recommendations for policymakers, platforms, and moderators for developing moderation strategies that are in line with users’ expectations.