Measuring dynamic media bias

Proc Natl Acad Sci U S A. 2022 Aug 9;119(32):e2202197119. doi: 10.1073/pnas.2202197119. Epub 2022 Aug 1.

Abstract

Ideological media bias is increasingly central to the study of politics. Yet, past literature often assumes that the ideological bias of any outlet, at least in the short term, is static and exogenous to the political process. We challenge this assumption. We use longitudinal data from the Stanford Cable News Analyzer (2010 to 2021), which reports the screen time of various political actors on cable news, and quantify the partisan leaning of those actors using their past campaign donation behavior. Using one instantiation of media bias-the mean ideology of political actors on a channel, i.e., visibility bias-we examine weekly, within-day, and program-level estimates of media bias. We find that media bias is highly dynamic even in the short term and that the heightened polarization between TV channels over time was mostly driven by the prime-time shows.

Keywords: media bias; news; political communication.

MeSH terms

  • Longitudinal Studies
  • Mass Media* / statistics & numerical data
  • Politics*
  • Prejudice* / statistics & numerical data
  • Television* / statistics & numerical data