Keywords: Social media, big data, sentiment analysis, and emerging technologies

Although more and more political information is distributed via social media sites, research on the effects of social media on users’ selective exposure to political information and opinion formation is scarce, especially regarding the question of whether those platforms foster or attenuate partisan selective exposure. It is still a matter of debate, whether social media sites facilitate exposure to like-minded viewpoints - and therefore help bolster pre-existing attitudes - or increase the likelihood of encountering attitude-challenging viewpoints (Brundidge, 2010; Pariser, 2010). To further our understanding of how social media sites affect information selection, the present study therefore aims to investigate under which circumstances social media increase exposure to attitude-consistent information and under which circumstances they may help diversify users’ political worldviews by bursting the shield of partisan selective exposure.
Previous research has identified a wide range of variables moderating the impact of attitude-consistency on selective exposure in general, such as content cues (e. g., argument strength, issue relevance) and individual characteristics (e. g., dogmatism, political tolerance). A rather new feature that potentially affects partisan selective exposure especially on social media platforms are social endorsements: Posts in a newsfeed are augmented with social cues, such as the number of likes and reactions, recommendations from personal contacts, and aggregated popularity measures. So far, however, only few studies have examined the impact of social endorsements on partisan selective exposure. Although first experimental studies indicate that social endorsements may - at least to a certain extent - play a role as potential moderators in partisan selective exposure (Dvir-Gvirsman, 2017; Messing & Westwood, 2014), little is known if this impact holds in a more realistic setting. The present study thus aims to add to this line of research by investigating the moderating role of social endorsements during the federal election campaign in Germany in 2017. This campaign was characterized by a strengthening of the new right-wing populist party AfD (Alternative für Deutschland), whose supporters spread their ideology mostly via social media sites and led to an increased polarization among the German electorate.
Against this backdrop, the study examined partisan selective exposure and the moderating role of social endorsement cues with a special focus on AfD supporters and opponents. We conducted an innovative online survey-experiment among German citizens (n = 209) who were asked to browse a closely mimicked Facebook newsfeed. It contained genuine current affairs and entertainment news, posted by the 20 most successful German social media pages on the day of the survey. While the content of the posts was held constant for participants, the type and amount of social endorsements were randomly manipulated. To investigate user selectivity among AfD supporters and opponents, three selective exposure measures were employed: attention to the posts, clicking on the posts, and the time spent with linked full-text articles. Statistical analyses will focus on the moderating effects of social endorsement cues on partisan selective exposure to news supporting or criticizing the AfD, controlling for issue relevance, political tolerance, and dogmatism, among others.