Balancing Act: How an Unbalanced Media Affects the Electorate

Abstract

Previous studies of media bias generally do not consider intra-campaign shifts in the composition of the coverage, leaving unanswered the question of whether coverage is consistently balanced or merely appears so when all the ups and downs of the election are tallied up. Even this aggregate assessment of balance frequently reveals imbalance in coverage, but there are few studies that test for effects of media imbalance. A lack of comprehensive content analysis data makes substantive work in this area challenging and frequently dependent upon single-election studies or those that examine a single variable over multiple years. This project takes advantage of a highly detailed, project-generated content analysis of all NBC Nightly News broadcasts and New York Times articles during the general election period (from just prior to the national party conventions through Election Day) for four election years (1996, 2000, 2004, and 2008) to address the question of whether aggregate and intra-campaign imbalance exists along three axes: volume of coverage (total coverage of the campaign and candidates), share of coverage (percentage of coverage received by each candidate on a given day, not conflated with volume), and tone of coverage (average daily valence score of coverage, positive or negative). Share of coverage is not stable at any point of equilibrium, but swings back and forth, favoring one candidate then the other throughout the general election. Analysis of tone of coverage reveals a surprising correlation between the positive/negative coverage of the candidates, with tone of coverage of the candidates correlating to a statistically significant degree. With this data it is possible to test whether shifts in the composition of coverage cause shifts in electoral support as measured by trial heat polls. Analysis of the data shows a significant relationship between changes in share and tone of coverage and changes in levels of electoral support. The corollary question, "what attracts media attention?" shows that media outlets are attracted by conflict as determined by a qualitative and quantitative analysis of headlines and share of coverage, and a further test shows a significant correlation between media coverage of an issue and both candidates' mention of that issue in their remarks, though not between media coverage and a single-candidate mention of the issue. The end result is a more-comprehensive picture of political reporting, its effect on the electorate, and the ways in which candidates may attract attention than is currently available in the literature. The combination of a multi-election measure with a high degree of intra-campaign diagnostic sensitivity provides a useful benchmark for the further study of media coverage of campaigns, as well as more than sufficient justification for its substantive significance as an area of political study, by virtue of the observed relationship between coverage and levels of electoral support.