Political behavior in the social media age

Subtitle:a data mining approach

The storming of the United States Capitol in January 2021 dramatically illustrates the impact of social media on society and political outcomes. The new reciprocal relationships afforded by digital media have reshaped the way political information is produced and consumed, and challenge some of the established theoretical insights in political communication. At the same time, the digital revolution also offers new opportunities for empirical research. By leveraging the information captured in digital traces, we can expand our understanding of political behavior in a way that was simply unimaginable a mere decade ago. In this PhD thesis, I collect and analyze social media data to explore the opportunities of data mining, text mining, and network analysis techniques for political research. The first part studies elite polarization with a large-scale comparison of political Twitter networks in 12 countries. In addition, a more in-depth study of party communication in Belgium is performed. In the second part, I analyze political polarization in non-political domains using Facebook-like-data in Belgium and the United States. I find that political polarization and partisanship are dependent upon the social network, institutional context, and individual characteristics. To mitigate polarization, I suggest lifestyle domains in which most cross-cutting interactions are present. This thesis shows that social media data provide a unique and rich source of online behavior but also come with ethical, technical, and methodological challenges. Therefore, joint efforts between social and computer scientists are needed to convert the enormous empirical potential into valuable insights.

Publication year:2021

Keywords:Doctoral thesis

Accessibility:Closed