Emotions and Information Diffusion on Social Networks: Evidence from German Political Twitter

Social media gained strong influence on multiple aspects of human behaviour, ranging from individual consumption to political participation. It is thus important to understand factors that make online content more successful in attracting viewership and user engagement. In this study, we address a particular mechanism reinforcing information diffusion, namely, the use of emotional words. We explore a unique data set of 171 thousand tweets of the German Bundestag members between August 2017 and October 2018 and we measure emotions in the tweets using the dictionaries of emotional words for the German language. For identification of the effects, we develop an instrumental variable approach, exploiting the exogenous variation in weather conditions as an instrument for the use of emotional words in the politicians’ tweets. We find that the use of negative emotional words, especially “actively” negative ones, in posts increases the number of retweets. Our results indicate, that policies aimed at reducing hate speech on social networks may indeed restrain the diffusion of controversial messages.