The Rise of Affective Polarization: Is It What We Think, or Who We Are?

Scholars and political observers express alarm at rising affective polarization in America. Many blame demographic sorting, claiming voters’ social characteristics ("who we are") align more consistently with partisanship now than in the past. Others attribute rising polarization to increased sorting of voters’ attitudes ("what we think") by party. Unfortunately, past research lacks comparable over-time measures of demographic and attitudinal sorting — and sometimes even conflates these two constructs. This frustrates efforts to assess how much each form of sorting has contributed to rising affective polarization. In this paper, I develop measures of both forms of sorting that are comparable with each other and over time. Using data from 1952-2020, I find demographic sorting has scarcely increased over time, and its relationship with affective polarization has remained stable. By contrast, attitudinal sorting — based on symbolic ideology, policy attitudes, and group sentiments — has risen substantially since 1972. Further calculations show that the likely contribution of demographic sorting to the rise in affective polarization since 1980 is no more than 2.5%, while attitudinal sorting can explain a meaningful share (53-63%) of its rise. Finally, original panel analyses and a meta-analysis of past experiments indicate that the relationship between attitudinal sorting and affective polarization is likely causal. Affective polarization’s rise appears partially explained by increased partisan sorting in what we think, but not in who we are.