close× Call Us
Title Voter Distrust and the Resonance of Party Outsiders in Primary Elections
Post date 02/05/2019
C1 Background and Explanation of Rationale

The bounded rationality of the American voter is well documented. Heuristics like differentiation between the parties represent one solution for how citizens make complex decisions without full information, but in primary elections, when all candidates are running under the same party label, no such heuristic exists. Which heuristics, then, do voters utilize when making decisions in primary elections? This survey experiment will utilize a conjoint analysis design to estimate the marginal effects of five factors which the literature suggests can influence voting decisions in primaries: gender, strategic general election considerations, ideology, party endorsements, and previous political experience.

Prior to presenting respondents with profiles of two candidates running in a hypothetical open-seat U.S. House election, we will ask questions designed to measure the respondent’s attitudes toward both government and party institutions. We hypothesize that respondents who are less trusting of those institutions will be less likely to choose the more “establishment”-style candidate. While some scholars have pointed to voters’ risk tolerance to explain support for outsider candidates (Kam and Simas 2012), this study will build on the preliminary work of Dyck, Pearson-Merkowitz, and Coates (2018) to assess the effect of distrust toward government and party institutions on support for party outsiders.

C2 What are the hypotheses to be tested?

H1: Voters who indicate less trust toward government institutions and/or their own parties in part one of the survey will be more likely to support candidates who have the traits of an outsider in part two.

H2: Democratic respondents will exhibit more trust in institutions and be more supportive of establishment-style candidates than Republicans.

Alternative hypotheses: We will test four alternative theories for voters’ decisions in primary elections. First, some scholars argue that gender shapes perceptions of ideology and candidate quality. On this view, we might expect Democrats to be more supportive of women regardless of their trust in government, while Republicans would be less supportive of women regardless of trust. Second, we will randomize whether or not the House district is likely to be competitive in the general election. Perhaps voters value quality and electability indicators like endorsements, ideology, and political experience to a greater extent if they want to maximize the chance of winning the seat in the general election. Third, perhaps voters who turn out to vote in primary elections (especially open seat races) seek to elect the most liberal or the most conservative candidate possible, regardless of their feelings towards institutions. Fourth scholars of political parties have argued that in low-information elections, endorsements from party elites can serve as a quality signal for voters. On this view, voters should be more likely to support a candidate if they see that their party’s congressional committee or a high-ranking elected official has endorsed that candidate.

C3 How will these hypotheses be tested? *

After collecting demographic information, we will gauge subjects’ trust in government and party institutions by asking how often they trust the following to represent the public’s best interests: Congress, the Supreme Court, the federal government, the president, Republican Party leaders, Democratic Party leaders, etc.

Subjects will then indicate their preferences for candidates in hypothetical U.S. House primary elections for the party matching their reported party identification. Following a conjoint analysis design, candidate attributes will vary randomly across age, gender, race, district competitiveness, occupation, previous political experience, ideology, and endorsements. Subjects will each review five candidate pairings. We will estimate the average marginal component effect of each of these causal quantities of interest in order to test our hypotheses. We will also perform subgroup analyses among high- and low-trust respondents and among Democratic and Republican respondents.

C4 Country United States
C5 Scale (# of Units) 1,200
C6 Was a power analysis conducted prior to data collection? No
C7 Has this research received Insitutional Review Board (IRB) or ethics committee approval? Yes
C8 IRB Number 1901002307
C9 Date of IRB Approval January 18, 2019
C10 Will the intervention be implemented by the researcher or a third party? Researchers
C11 Did any of the research team receive remuneration from the implementing agency for taking part in this research? No
C12 If relevant, is there an advance agreement with the implementation group that all results can be published? not provided by authors
C13 JEL Classification(s) not provided by authors