Don’t @ Me: Experimentally Reducing Partisan Incivility on Twitter

Abstract

I conduct an experiment which examines the impact of moral suasion on partisans engaged in uncivil arguments. Partisans often respond in vitriolic ways to politicians they disagree with, and this can engender hateful responses from partisans from the other side. This phenomenon was especially common during the contentious 2016 US Presidential Election. Using Twitter accounts that I controlled, I sanctioned people engaged partisan incivility in October 2016. I found that messages containing moral suasion were more effective at reducing incivility than were messages with no moral content in the first week post-treatment. There were no significant treatment effects in the first day post-treatment, emphasizing the need for research designs that measure effect duration. The type of moral suasion employed, however, did not have the expected differential effect on either Republicans or Democrats. These effects were significantly moderated by the anonymity of the subjects.