The Analysis of Topic Model Network (ANTMN) method was developed by myself and my colleague Dror Walter to equip researchers with an inductive, data-driven approach for the identification of media frames. Media framing is a core journalistic practice consisting of the selective emphasis of specific features of events and people, at the expense of others. Though a journalistic practice at its core, framing is being used by others, from politicians to laypersons on social media. In this talk I will go through the three steps of ANTMN (topic modeling, network analysis, community detection) and demonstrate its applicability to various contexts, and its flexibility when combined with other social science methods, such as time series analysis, surveys, and experiments. I will demonstrate our work with ANTMN using examples ranging from the way the media talk about epidemics and its effects on the public’s trust in and compliance with CDC, to the ways politicians frame and are being framed during electoral races and its impact on their success, to the analysis of Russian trolls and bots who interfered with American online discourse during the years leading to the 2016 elections and what it means for our democracy.

See other DSSN presentations.