Rüdiger Schmitt-Beck, Alexander Staudt
Media Bias and Voter Preferences at the 2009 to 2017 German Federal Elections

Invited talk at Koç University, Istanbul, 18. Dezember 2018

As a preliminary interim report from an ongoing research project our paper examines persuasive effects of two forms of news media bias on German voters’ electoral preferences –coverage bias, that is, the amount of news devoted to the various parties and candidates, and statement bias, that is, the evaluative tone in which these actors are addressed in news reporting. The expectation is that voters’ views of the parties and their lead candidates are more favorable, the larger the amount of coverage devoted to them, and the more positive the tone of this coverage. Combining the ‘attentiveness’ and ‘linkage’ approaches to studying media effects our analyses draw on merged data from rolling cross-section surveys and media content analyses (of all major TV news and national newspapers) from the first two rounds of the German Longitudinal Election Study (GLES), conducted at the 2009 and 2013 German federal elections. As dependent variables the paper looks at vote intentions, chancellor preferences as well as thermometer scales indicating voters’ evaluations of the parties and their lead candidates. The analyses simultaneously model direct effects of the media actually used by voters (where the link between media content and survey data is established by respondents' individual media use and the date of interview with a lag of one day for media content) and indirect ‘climate’ or ‘environmental’ effects of overall media coverage (where content data aggregated across all media are linked to the survey data via the date of interviewing, again with a lag of one day for media content). Our findings suggest that party and candidate evaluations are more sensitive to media effects than vote intentions and especially chancellor preferences. These effects are rather clear-cut and in line with expectations for statement bias, but partly implausible for coverage bias.