Gennaro Gloria, Derksen Laurenz, Abdelrahman Aya, Broggini Emma, Green Mariya Alexandra, Haerter Victoria Andrea, Heer Elia, Heidler Isabel, Kauer Fiona, Kim Han-Nuri, Landry Benjamin, Levis Alessio, Li Jiazhen, Şimşir Şevval, Srbinovska Iva, Vital Robin Anna, Donnay Karsten, Gilardi Fabrizio, Hangartner Dominik
Department of Political Science, University College London, London, WC1H9QU, UK.
Center for Comparative and International Studies, ETH Zurich, 8092, Zurich, Switzerland.
Sci Rep. 2025 Jul 1;15(1):22018. doi: 10.1038/s41598-025-05041-w.
Online intergroup hostility is a pervasive and troubling issue, yet experimental evidence on how to curb it remains scarce. This study examines counterspeech as a user-driven strategy to reduce hate speech. Drawing on theories that suggest adopting the perspective of minority groups can reduce prejudice, we randomized four counterspeech strategies across the senders of 2102 xenophobic Twitter messages. Compared to a passive control group, we find that the pooled effect of the three perspective-centered strategies-traditional perspective-taking, analogical perspective-taking, and perspective getting-increased the likelihood that the sender deleted their xenophobic message by +0.14 SD ([Formula: see text]), decreased the number of likes the xenophobic message received by others (- 0.133 SD, [Formula: see text]), but yielded a limited and not statistically significant estimate for the share of xenophobic messages the sender posted over the following four weeks (- 0.084 SD, [Formula: see text]). Differences between the three perspective-centered strategies were generally small and not statistically significant, though analogical perspective-taking-encouraging senders to compare their own experiences of being attacked online with their discriminatory behavior toward outgroups-appears to have slightly larger effects across multiple outcomes. Disapproval messages without a perspective shift produced smaller and non-significant estimates. These findings advance our theoretical understanding of how counterspeech works and provide actionable insights for how users can contribute to reducing intergroup hostility and its amplification online-especially at a time when many platforms are scaling back content moderation.
网络群体间的敌意是一个普遍且棘手的问题，但关于如何遏制它的实验证据仍然很少。本研究考察了反驳言论作为一种由用户驱动的减少仇恨言论的策略。借鉴那些表明采纳少数群体视角可以减少偏见的理论，我们在2102条仇外推特信息的发送者中随机分配了四种反驳言论策略。与被动对照组相比，我们发现三种以视角为中心的策略——传统的换位思考、类比换位思考和视角获取——的综合效果使发送者删除其仇外信息的可能性增加了+0.14标准差（[公式：见原文]），减少了仇外信息获得的他人点赞数（-0.133标准差，[公式：见原文]），但对于发送者在接下来四周内发布的仇外信息所占比例的估计有限且无统计学意义（-0.084标准差，[公式：见原文]）。三种以视角为中心的策略之间的差异通常较小且无统计学意义，不过类比换位思考——鼓励发送者将自己在网上受到攻击的经历与他们对外群体的歧视行为进行比较——在多个结果上似乎有稍大的影响。没有视角转变的反对信息产生的估计值更小且无显著性。这些发现推进了我们对反驳言论如何起作用的理论理解，并为用户如何有助于减少群体间敌意及其在网上的扩散提供了可操作的见解——特别是在许多平台正在缩减内容审核的时候。
