Candidate Differentiation in Primary Elections

In partisan primary elections, candidates for Congress cannot rely on party labels to convey information to voters. In this dissertation, I evaluate the extent to which candidates respond strategically in both their issue positioning and how they communicate tangible information about their candidacy to voters during primary elections. Further, I also assess the implications of this behavior on voters' perceptions of candidates. I start by addressing whether or not incumbents' issue positioning responds to the positioning of primary challengers. While theoretical models of electoral competition suggest incumbents should adopt more extreme (moderate) positions, prior empirical work testing this hypothesis does not find support for this hypothesis. To overcome existing data limitations, I introduce Website EmBedding (WEB) Strategic Positioning Scores. Consistent with theoretical expectations, I show incumbent candidates become more extreme (moderate) in their issue positioning during the campaign in response to an extreme (moderate) primary challenger whereas previous measures do not. I then turn to assessing how partisan primary elections consistently nominate candidates with certain characteristics (e.g., ideologically extreme candidates, experienced candidates) when voters cannot rely on party labels and put little effort into obtaining detailed information about candidates. Using an original data set of campaign logos from candidates for the U.S. House of Representatives, I show candidates use different logo colors to communicate electorally valuable traits that differ from co-partisan candidates who do not have these traits. Specifically, Democratic candidates who are more liberal, a woman, a racial minority, or have prior elected experience and Republican candidates who are more conservative, male, or have prior elected experience. Finally, I assess whether or not these branding decisions actually influence citizen perceptions' of candidates' ideological positioning. Using a conjoint experiment, I find that colors in a campaign logo affect perceptions of candidates' ideological positioning among Democrats but not Republicans. I do not find support that these effects are concentrated among politically attentive respondents. Finally, I find that these results are consistent when respondents have policy information about the candidate versus when they do not.