Leveraging Cross-Platform Social Media Data to Study Conflict and Polarization

ABSTRACT

The explosion in online data availability has made the internet a ripe source of new data as well as fertile grounds for inquiry regarding the ways in which it has impacted traditional forms of political expression and attitude formation. However, as the use of online data in general and social media data in particular proliferates, it is also important that researchers understand the biases present in using any single source. This thesis uses cross-platform social media data to both consider differences between sources and explore the connection between online data and offline events. In the first chapter, I find that actors in Syria use Twitter, Telegram, and Facebook in different ways to communicate their civil war narratives. In the second chapter, I build on this work in the Syrian context to explore how conversation varies in areas with and without refugee return. Finally, the third chapter explores segregation in content viewership between Republicans and Democrats in the U.S. on web browsing, YouTube, and audio apps. Though there is support for higher levels of segregation at lower levels of aggregation, such as URLs versus domains, other results indicate a need to be cautious about small sample bias when considering these smaller groups. Taken together, these papers add to the literature on how social media and browsing data can be used to understand offline events as well as the necessity for cross-platform research in the social sciences.

COMMITTEE

Fotini Christia (chair, supervisor), Dean Eckles, Kiran Garimella (Rutgers)

EVENT INFORMATION

Hybrid event. To attend virtually, please contact the IDSS Academic Office (idss_academic_office@mit.edu) for connection information.