Social Approval Seeking and Misperceived Legitimization: The Roots of Toxicity in Online Communities

Abstract

Online toxicity has become a major global concern. However, we know little about how monetary and non-monetary incentives give rise to such toxicity and how online platform designs shape these incentives. This paper investigates whether mutual rating systems, illustrated by a ``like'' button, induce approval-seeking behavior and lead to the spread of toxicity. For this purpose, we draw on two natural experiments from a major Japanese Q\&A platform: (i) the introduction of a like-button feature and (ii) the subsequent integration of these ratings with monetary incentives. Using a difference-in-differences approach combined with machine learning-based toxicity measures, we compare changes in toxicity levels before and after these interventions among users who are particularly sensitive to social approval versus those who are not. We find that individuals driven by social approval become significantly more toxic once these features are introduced, indicating that public affirmation can intensify harmful behaviors. Further analysis reveals a “misperceived legitimization” mechanism: approval-seeking users initially conceal their harmful tendencies, but interpret visible endorsements of toxic content as an indication of wider community acceptance, causing them to unleash their latent toxic behavior. These results suggest that platform features that promote social approval seeking can originate toxicity in online communities.