Theorists of affective polarization often argue that it motivates political participation, but erodes support for democratic norms. Cross-sectional findings tend to be consistent with these concerns, but experimental findings do not. What is the reason for this disjuncture? In this paper, we argue that most observational work on affective polarization, even longitudinal work, uses methods that do not cleanly separate effects derived from between-person differences (which could be spurious) from those derived from within-person changes (which are more credible). In this paper, we use multiple panel datasets from the United States, United Kingdom, New Zealand, and Chile to separate within-person and between-person effects of affective polarization on political engagement and democratic attitudes.
