Facebook Commenting during the 2016 U.S. Presidential Debates

The second day at AoIR 2017 starts with a panel on the U.S. elections in 2016, and Patrícia Rossini is the first speaker. She notes the limited focus in the past on how voters interact with election campaigns; much of the research has paid attention simply to the campaigning strategies themselves. But there is also evidence that users encounter a good deal of campaigning in their social networks, though they do not necessarily like doing so – in part because the discourse can be heated, emotional, and uncivil. Further, reactions to some discourse differ based on whether users agree or disagree with the uncivil statements being made.

Such uncivil discourse can also be driven by the ephemeral features of online discourse; it is easily possible to make an uncivil statement and then disconnect from any further discussion, for example. Overall, uncivil discourse can include harmful statements and attacks on others in the conversation, as well as pejorative statements towards candidates and other political actors. The present project examined this especially in the context of the major presidential debates during the 2016 campaign, where many viewers are now second-screening by following the debate on TV and commenting on it through social media at the same time. It draws on Facebook comments on the candidates' own posts, examining some 43,000 comments in total.

The candidates' posts were categorised as attacks, advocacy, calls to action, informative statements, and endorsements. Some 32% of all comments were uncivil, and this was equally the case for comments on Trump and Clinton posts, despite the differences between the two candidates. The candidates' messages influenced incivility: attack and call to action posts actually produced less incivility, and more so when posts were engaging in advocacy; further, messages posted by Donald Trump were somewhat more likely to receive uncivil responses (though not necessarily directed at him).

Incivility is therefore an inherent characteristic of online public discourse, and occurs frequently in comments on candidates' posts. Candidates' posts can influence the level of uncivil discourse, through the correlations here seem somewhat counter-intuitive. The plan is now to expand this analysis to the full dataset of all comments over the course of the election campaign.