Replacement Effects and the Slow Cycle of Ideological Polarization in the U.S. House

Abstract

Ideological polarization and partisan enmity is arguably the driving force of the past several decades of U.S. electoral history. Today there is a vast ideological gulf between elected Democrats and elected Republicans in Congress and elsewhere. But this pattern is not unique in U.S. political history. McCarty, Poole and Rosenthal (2006), using the first dimension of DW-NOMINATE scores for the U.S. House, show compelling evidence for a cyclic pattern of ideological convergence and divergence over the period 1856-2006 that we call the accordion effect. Our interest is in the electoral mechanisms that must be in place to generate observed contemporary dynamics rather than in causality per se. First, in a model of one-dimensional ideological competition, for a fixed distribution of constituency medians, we model the gap between the mean Democratic and the mean Republican position in Congress as the product of two (potentially interrelated) factors: (1) the mean difference in roll call voting scores (first DW-NOMINATE dimension) when a Republican in a district is replaced by a Democrat (or conversely) and (2) the likelihood that districts of a given ideological stripe will elect Democrats (Republicans). We then formally model possible dynamics involving partisan replacement, leading to either increasing or reduced polarization. We suggest that one such dynamic, where each party “chases the tail” of the other party that is closest to its own position, has been the driving force in enhancing polarization since 1980. We suggest however, that a quite different replacement dynamic was found in much of the first half of the 20th century, one leading to considerable overlap in party ideologies by the 1950s that lasted through the 1970s.