Selective exposure and echo chambers in partisan television consumption: Evidence from linked viewership, administrative, and survey data

Influential theories doubt that partisan television's audience is sufficiently large, moderate, or isolated from cross-cutting sources for it to meaningfully influence public opinion. However, limitations of survey-based television consumption measures leave these questions unresolved. We argue that nonpolitical attributes of partisan channels can attract voters to form habits for watching channels with slants they do not fully share. We report findings from three novel datasets which each link behavioral measures of television consumption to political administrative or survey data. We find that approximately 15% of Americans consume over 8 hours/month of partisan television. Additionally, weak partisans, independents, and outpartisans comprise over half of partisan channels' audiences. Finally, partisan television consumers largely consume only one partisan channel and remain loyal to it over time, consistent with "echo chambers." These findings support our argument and suggest partisan television's potential to influence public opinion cannot be dismissed.