Individual Submission Summary
Share...

Direct link:

The Effect of Primaries on Intra-Campaign Positions in Congressional Elections

Fri, October 1, 4:00 to 5:30pm PDT (4:00 to 5:30pm PDT), TBA

Abstract

Whether congressional primaries are a source of polarization in the U.S. Congress remains contested. Central to claims that primaries polarize is the strategic positioning dilemma, which contends that candidates must satisfy an extreme selectorate to earn the nomination before facing a comparatively moderate general electorate. We test whether candidates adapt positions across the primary and general election cycle using a unique dataset of positions over time. We apply an original scaling method based on candidates’ language on Twitter to estimate positions of over 800 primary candidates in the 2020 U.S. congressional election. We show the validity of these positions by correlating them with NOMINATE (r > 0.88) where possible. Democratic candidates who did not become the nominee moderated significantly following the primary, indicating strategic positioning for perceived electoral benefit, and where the presence of a primary caused artificial partisanship. For Republicans, we observe no centrifugal effect of primaries. Moderation among nominees after the primary was conditioned by district partisanship, with greater moderation in less favored districts for a candidate’s party. We consider explanations and implications of these findings in the context of communication strategy in sequential elections.

Congressional primaries have been attributed with increasing polarization in Congress by fostering the nomination of candidates who are more extreme than general electorates or would otherwise be have selected by party organizations. Candidates must first appeal to an extreme selectorate to earn the nomination before attempting to garner wider support among a general electorate who prefer moderate candidates (Ansolabehere et al. 2001). Accordingly, candidates are presented with a strategic positioning dilemma (Brady et al. 2007). We test candidates' responses to this dilemma during the 2020 election cycle using a new scaling technique based on language on Twitter, asking the question: did candidates take more extreme positions during the primary and then moderate for the general election? Our method enables us to continue scaling losing candidates after the primary, allowing us to determine the ‘genuine’ positions of candidates absent electoral incentives. We first test whether losing candidates took artificial positions during the primary for perceived electoral reward. In this paper we test the adaptive rather than the selective effect of congressional primaries; our interest is in the change in behavior rather than outcome effects such as more extreme candidates defeating comparative moderates. For nominees we assess positions after the primary, hypothesizing that moderation is conditional on district partisanship.

Though some evidence suggests that candidates adopt more extreme positions in primaries (Brady et al. 2007, Burden 2001), a systematic study of candidate positions across a primary and general election cycle remains lacking, in part due to the limited availability of time-series data of positions of elected officials and losing candidates. Traditional ideal point estimates are only available for elected members of Congress (McCarty et al. 2006) or across the entire election cycle (Bonica 2014). To fill this gap, we measure changes in primary candidate positions across the primary and general election cycle using a unique data set of dynamic social media-based positions on the left-right axis. Based on saliency theory (Budge 2001), we use multidimensional scaling to identify the liberal-conservative dimension on Twitter, based on more than 1,700,000 tweets by 930 candidates. We validate the measurement using NOMINATE scores of candidates who were also members of the 116th Congress, with correlations of over 88 percent.

We find that losing candidates in Democratic primaries move to the center after their primary defeat, indicating that candidates adopted artificially or strategically extreme positions to appeal to a perceived extreme primary electorate. This finding runs counter to research indicating that candidates are more extreme than both primary and general electorates (Bafumi and Herron 2010). Among Republican losers, we find no shift in behavior after primary elections, in line with extensive empirical literature indicating that primary electorates are not a source of candidate extremism (Abramowitz 2008, Hill 2015, Hirano et al. 2010, McGhee et al. 2014, Sides et al. 2016).

For primary winners we find that moderation among is conditioned by district partisanship. We use the established method of district partisanship, Cook's partisan vote index (PVI). Candidates in districts less favored their party moderated more following the primary election. Our finding is in line with other literature on this topic, indicating that general election candidates attempt to align themselves with their district.

Authors