Learning about policy positions through the news: how media content shapes voters’ knowledge of parties’ policy positions.

There is a consensus that a minimal knowledge of political events is necessary for citizens to remain politically involved (Delli Carpini & Keeter, 1996). Knowledge of party positions is particularly important for citizens in order to cast a substantively informed vote (Bartels, 1986; Rabinowitz & Macdonald, 1989). One period when they may learn about party positions is during election campaigns (Jenssen, 2009). The aim of this paper is therefore to investigate whether citizens indeed learn about the party system during campaigns and whether this effect is contingent on voter characteristics. Furthermore, I do not only focus on citizens but also investigate the supply side by examining whether campaign learning is more apparent when parties devote more attention to an issue during the campaign. Specifically, this study will contribute to existing research on campaign learning in three ways. First, rather than using the predominant operationalization of asking citizens three to five knowledge questions on institutions and actors, I will focus on policy position knowledge, the ability of respondents to position parties on a large number of policy proposals. Second, I am able to disentangle effects of different type of media channels because I possess detailed information on what type of media respondents used during the campaign. Finally, by means of a content analysis of campaign news, I will be able to also link the actual information provided during the campaign with voter data. Two datasets will be used. First, I use a two-wave electoral survey in the Dutch-speaking part of Belgium. In both waves, respondents (n=1681) had to indicate the positions of seven main parties on 18 different policy proposals. The question format allows us to create a unique and accurate measure of voters’ party position knowledge. In addition, we asked respondents for several media channels how often they used them during the campaign to follow political news. Second, I will code the content of both newspapers and social media (facebook) with regard to the policy statements included in the surveys. This allows us to follow parties’ communication on these different issues. These policy statements were selected by journalists right before the start of the electoral campaign and allowed us to grasp the core issues of the electoral campaign. For the analysis, I will use auto-regressive models. Concretely, this means that I use the ability to accurately position parties on the different proposals in wave 1 to predict the accuracy in the second wave. Moreover, given that I stack the data – as each respondent ranks seven parties on 18 policy proposals and is thus included multiple times - I use cross-classified multilevel analysis.