Education and Political Polarization: A Level or Slope Effect?

Abstract

This paper asks whether education changes the rate at which partisan identity hardens or only the baseline propensity to switch. It develops a mean-field model of partisan sorting in which agents balance tribal conformity against private signals. In the model, education reduces susceptibility to tribal pull but does not alter the hardening rate itself. The prediction is a level effect, not a slope effect: education shifts baseline switching rates without changing the tenure profile of partisan lock-in.

The paper tests this implication using ANES panels from 1972–1976 and 2016–2024. In both eras, the education-by-tenure interaction is close to zero and statistically insignificant. College graduates are less likely to switch party identification at baseline, but the Converse curve is essentially parallel across education groups. The evidence is consistent with a separation between media susceptibility and identity formation. Education appears to affect the starting point of partisan stability more than the rate at which party identity hardens with tenure. The findings do not support models in which education accelerates convergence toward stable partisanship.

Keywords: education, partisan sorting, Converse curve, party identification, identity formation, ANES

Request paper