A quantitative analysis of news consumption on Facebook

The advent of social media and microblogging platforms hasradically changed the way we consume information and formopinions. Despite the importance given by the scientific communityto these changes in news production and consumption,not much is certain about the spreading and consumptionpatterns through social media platforms. This thesis doesan in-depth quantitative analysis of news consumption andpolarization on Facebook, taking into account factors such astrust in news, confirmation bias and cognitive dissonance.First, we explore the anatomy of the English news by characterizingon a global scale the news consumption patterns of376 million users over a time span of six years. Second, wecompare the selective exposure and polarization of France,Germany, Italy and Spain (86M users), and present a modelof selective exposure that considers trust in the emergence ofcommunities. Finally, we analyze the polarization on the vaccinationdebate over time (2.6M users).Our findings show that confirmation bias and homophily arepresent in the users’ consumption of news, which leads topolarization and the creation of homogeneous communities.We also find that the preferences of users and news providersdiffer. By tracking how Facebook pages like each other andexamining their geolocation, we find that users have a morecosmopolitan perspective of the information space than newsproviders. In addition, we devised two simple models ofselective exposure that reproduce the observed connectivitypatterns.