From August 2nd to August 12th, 2024, we conducted a large-scale national survey of registered voters. We used a conjoint survey design which allowed us to see what policies voters preferred and how much they cared about specific issue areas relative to each other. We also asked voters traditional survey questions. 

The key takeaways are:

  1. Unsurprisingly, voters care a lot about party cues. Party affiliation – regardless of a candidate’s policy views – was the most important feature for voters when evaluating candidates. However, this attribute was far from determinative on its own.
  2. Voters overwhelmingly prefer candidates who support protecting abortion rights, and a candidate’s view on abortion rights was the second most important feature for voters when evaluating candidates.
  3. Voters strongly dislike candidates who support defunding the police, but the  importance of crime policy in general was not especially high when compared to other policies.

Conjoint results

The plot below shows the average level of support for a candidate with a specific attribute using survey weights. We weighted the survey to the national voter file on race, gender, age, education, household income, turnout score, partisanship score, and the interactions between race and education and race and household income using raking. If the point estimate is to the left of the dashed line at 50%, then that feature is relatively disfavored. If it is to the right, then that feature is relatively favored. These averages give us a sense for how a specific policy stance influences vote choice given  other policy stances. 

These results do not tell us what proportion of voters support a policy. They only tell us how a policy stance might affect a voter’s choice in comparison to the other policies we evaluated.

We want to emphasize that this conjoint analysis provides a useful heuristic for what policies voters reward on average, but the issue support and salience models we developed using this data will tell you which specific voters react positively to a given policy. Therefore, we caution from overinterpreting these high-level results. 

Our key findings are as follows:

  1. Party alone was the most salient candidate trait. All policy positions being equal, voters slightly preferred Republicans to Democrats, but this was not statistically significant.
  2. Voters overwhelmingly favored candidates who would support protecting access to abortion. At the same time, voters overwhelmingly disliked candidates who would seek to ban access to abortion.

While immigration ranked as the second most salient issue we tested, none of the immigration policy stances we evaluated moved voters in a significant way (in the aggregate, at least).

  1. Voters strongly disliked candidates who would support defunding local police departments. They favored candidates who say we should enforce current laws, and they might support candidates who focus on prevention and rehabilitation. There was no clear signal for or against candidates who would support hiring more police officers.
  2. Voters disliked candidates who would favor banning discussion of racism and sexuality in schools, and they favored candidates who would support allowing (but not forcing) schools to discuss racism and sexuality. Voters did not clearly favor or oppose candidates who would want to require schools to discuss racism and sexuality.
  3. The least salient issue area we evaluated concerned policies meant to protect American workers – ranging from strengthening labor unions to enacting harsh tariffs on Chinese imports. None of the policy stances we evaluated made a major difference in overall candidate support.