Bringing content into the equation: Using a wordscores method to compare the effect of newspaper and television on vote choice in referendums

Citizens often lack policy-specific knowledge to cast a well-informed vote in a direct democratic ballot. To fill this information gap, citizens rely on the media to find policy arguments. With that in mind, this article compares the effect of newspapers and television on vote choice in a referendum. It is the first attempt to use the same method, i. e., a Wordscores content analysis, to position newspaper outlets and television format on a No–Yes scale within the same direct democratic campaign. Indeed, by linking this content analysis with a three-wave online panel survey (N = 686) this method considers the divergent influence of newspapers and television on vote choice. A panel model identifies a positive effect of newspapers on the vote choice of those who consume newspapers intensively. Alternatively, the model detects a positive effect of television content on vote choice, but only for citizens with low political awareness. In the end, this paper extends the literature on the importance of systematic reasoning in direct democratic votes.