Strategic Heterogeneity in Policy-Level Positioning: Evidence from Congressional Campaigns

Do candidates maintain consistent left-right issue positions in congressional elections? How do district conditions influence variations in this consistency? While these questions are fundamental to the study of U.S. elections, surprisingly little is known about issue positioning in campaigns. We combine campaign platforms with machine learning methods to estimate the left-right orientations of U.S. House candidates (n = 4,505) across six salient issue areas in multiple election cycles (2018-2022). Our validated measures reveal meaningful within-candidate variation in issue positioning that systematically reflects district conditions. Candidates exhibit greater positional flexibility in districts that are not safe for their party and where constituent policy preferences are heterogeneous. We further demonstrate that this positional variation directly reflects district opinion, as issue-specific positioning aligns with constituency policy preferences. These findings reveal how candidate messaging varies systematically across issue domains and electoral contexts, fundamentally shaping the information environment that structures democratic choice and representation.