Social media, sentiment and public opinions: Evidence from #Brexit and #USElection

This paper studies information diffusion in social media and the role of information dissemination in shaping public opinions. Using Twitter data on the 2016 EU Referendum and the 2016 US Presidential Election, we find that information about these two events is spread quickly on Twitter, most likely within 1-2 hours. There are also interactions among different types of Twitter agents in spreading information with a considerable spillover from bot to human tweeting activities. However, the degree of influence depends on whether bots provide consistent information with humans' priors. This finding lends support to the "echo chambers" effect on Twitter that Twitter users are more likely to expose to information supporting their own views while ignore the opposite information. Further examination shows that sentiment matters in information acquiring and sharing. Overall, our results suggest that the aggressive use of Twitter bots, coupled by the fragmentation of social media and the role of sentiment, increases the polarization of public opinions about the EU Referendum and the US Election.