Measuring Policy Positioning in U.S. Congressional Elections

Measures for the policy positions of political actors are essential to testing foundational questions about political representation, partisan polarization, and electoral competition. Thousands of candidates run for the U.S. Congress each election season, yet we lack systematic information on the issue-level positions of the vast majority of these electoral contenders. To that end, we pair an original collection of campaign platforms with tools for machine learning to produce novel issue-level measures of congressional candidates’ policy positioning. Our measures span multiple salient issue areas (e.g., guns, immigration, and abortion) for candidates who ran for the U.S. House of Representatives between 2018 and 2022. Through a series of validation tests, we demonstrate that our Candidate Positioning Indexes (CPIs) reliably capture latent issue-specific positions from text. We go on to demonstrate that unidimensional “ideal points” for candidate policy positioning mask important variation in issue polarization at the party level and multidimensionality in left-right positioning at the candidate level.