Empathy Can Reduce Hate Speech and Its Amplification on Social Media

• In Event: Emotions and the Psychology of Immigration Attitudes

Abstract

Online hate speech is a pervasive and troubling issue, yet experimental evidence on the effectiveness of different strategies to curb it remains scarce. This study focuses on counterspeech as a means for users to actively contribute to reducing racist hate speech beyond platforms' content moderation practices. Informed by theories from social psychology about the role of empathy in overcoming intergroup prejudice, we design the following strategies: traditional perspective-taking, which encourages the sender of hate speech to adopt the perspective of the outgroup; analogical perspective-taking, which encourages the sender to draw an analogy between their own experiences of being attacked online and their discriminatory behavior towards the outgroup; and perspective-getting, which offers the experience of a discriminated outgroup member to the sender. We randomize these counterspeech strategies across 2,102 recently sent racist or xenophobic messages from US-based Twitter users. Compared to the control group, the three empathy-based strategies increase, on average, the propensity that the sender deletes the xenophobic message by 70\% (a 3.6 percentage points increase), reduce the share of xenophobic messages over the following four weeks by 50\% (0.3 percentage points), and decrease the xenophobic message's number of likes and retweets by 54\% (1 fewer like) and 98\% (0.14 fewer retweets), respectively. Among these strategies, analogical perspective-taking is particularly effective. In contrast, disapproval messages not seeking to induce empathy have weaker effects. These findings deepen our theoretical understanding of how to reduce intergroup hostility and its amplification in the online space and provide actionable insights for designing scalable counterspeech strategies.

Authors

• Gloria Gennaro, UCL

• Laurenz Derksen, ETH Zürich

• Dominik Hangartner, ETH Zurich