Exploring the Relationship Between Campaign Discourse on Facebook and the Public's Comments: A Case Study of Incivility During the 2016 US Presidential Election

Social media is now ubiquitously used by political campaigns, but less attention has been given to public discussions that take place on candidates' free public accounts on social media. Also unclear is whether there is a relationship between campaign messaging and the tone of public comments. To address this gap, this article analyzes public comments on Facebook accounts of candidates Trump and Clinton during the US election presidential debates in 2016. We hypothesize that attack messages posted by the candidates predict uncivil reactions by the public and that the public is more likely to be uncivil when attacking candidates. We use content analysis, supervised machine learning, and text mining to analyze candidates' posts and public comments. Our results suggest that Clinton was the target of substantially more uncivil comments. Negative messages by the candidates are not associated with incivility by the public, but comments are significantly more likely to be uncivil when the public is attacking candidates. These results suggest that the public discourse around political campaigns might be less affected by what campaigns post on social media than by the public's own perceptions and feelings toward the candidates.