A Balanced Information Diet, Not Selective Exposure: Evidence from Erie to Arbitron

Abstract

Do citizens seek out political information in a biased and selective manner, that is do they primarily watch, listen and read information supporting their side? Or, are they balanced- information seekers, giving equal time to both sides of the argument? To adjudicate between these competing models of political exposure, I measure political exposure bias across two media environments using novel individual level longitudinal datasets. First, I attempt to replicate the large selective exposure effects found in Lazarsfeld’s 1940 panel study – designed to extensively measure voter’s exposure to the universe of campaign communications – political speeches, newspaper stories, radio newscasts, editorials, columns, magazine articles over the course of the 1940 U.S. presidential election. The second study leverages new behavioral data measuring selective exposure to TV and Radio newscasts during the 2006 Midterm Election. These data combine survey measures of political views with media tracking data that continuously and automatically log panelists’ 24/7 exposure to all radio and television programming over a 90 day period – (panelists’ cell phones pick up radio and TV sound which is matched against a database of programming) – resulting in 541, 892 observations for 920 panelists. Results are clear across media environment and platform: less than 5% of the electorate has a tendency to read, listen or watch to one’s own side. Rather the vast majority of citizens, consume centrist media, while tuning out news from ideological sources altogether. Of those who do turn to ideological sources, most have a balanced information diet – in that they consume news programming from both ends of the political spectrum. Contrary to recent claims depicting an electorate where powerful partisan loyalties bind citizens to their preferred ideological news sources, there is little evidence that selective exposure occurs on a sufficiently broad scale to affect deliberative democracy or contribute to mass polarization.

Keywords: selective exposure, news, media effects, media slant, audience measurement, homophily