Social Media Moderation and Content Generation: Evidence from User Bans

The rise of inappropriate content (e.g., misinformation, spam, hate speech, etc.) has become a major concern for social media platforms. To deal with such challenges, platforms adopt various strategies to moderate the content on their websites. This study focuses on user bans, a common but controversial moderation strategy that suspends rule-violating users from further participation on the platform for a predetermined period. Specifically, we investigate the impacts of user bans on banned users’ content-generating behavior (both quantity and quality). Leveraging the reactance theory, we formalize our hypotheses relating users’ behavioral reactions to such content moderation strategy. We implement multiple empirical designs to analyze data from a major social media platform. Our results show that users provide more answers on average after bans are lifted. In contrast, we find that the quality of the content (measured by the linguistic features as well as content appropriateness) decreases after user bans. Furthermore, we find that platform recognitions, such as badges and recommendations, alleviate individuals’ reactance toward bans. Specifically, users who have received platform recognition would reduce inappropriate postings and improve the quality of their content after bans. Lastly, we explore the heterogeneous effects of user bans for different banning causes and repeated bans. Our research is among one of the first to evaluate the effectiveness of user bans and could have important implications for content moderation on social media.