This presentation introduces a novel approach to estimating the ideological positions expressed in newspapers by combining automated topic modeling with LLM-enabled stance classification. Applying this approach to German newspapers, we identify the positions taken in news coverage across political issues and link these patterns to survey data on individuals’ political preferences and media use. This allows us to examine how the ideological profiles of newspapers relate to those of their audiences. In particular, this line of research addresses three questions: 1) whether newspapers are more or less polarized than their audiences, 2) whether this relationship is symmetric across the ideological left and right, and 3) whether the match between newspaper ideology and audience ideology is stable over time. More broadly, the project offers a new way to study the alignment between media content and audience preferences at scale, while translating the distinction between mass and elite polarization from political science to the study of journalism.
This event has passed.