The Electoral Consequences of Candidate Ideology: Evidence from a Within-Precinct Analysis of U.S. Elections

Most research on the electoral penalty of candidate ideology relies on between-district or longitudinal comparisons, potentially confounded by turnout and ballot composition effects. We adopt a within-precinct design, leveraging granular precinct-level election data from the MIT Election Data and Science Lab (MEDSL) for 2016-2022 alongside extensive data on candidate ideology. Our design exploits within-precinct variation in two-party vote shares for races appearing simultaneously on the same ballot, thus isolating persuasion effects as voter turnout is held constant. This allows for precise estimation of the electoral penalty and comparisons across electoral contexts. Our findings reveal that a standard deviation change in the midpoint between candidates is associated with an average vote share penalty of 0.6-1.0 percentage points, depending on the measure of ideology used. This effect varies substantially by office type, with gubernatorial races exhibiting the largest effects, compared to more modest effects for U.S. House contests. This suggests voters may be more attuned to ideological differences in certain electoral contexts. Our design additionally enables empirical tests of how information availability, incumbency status, and partisan asymmetries influence the electoral penalty.