Quantifying Polarization at Sub-State Levels Using New Geospatial Methods

• In Event: Party Polarization in Canada and the United States

Abstract

On the elite level, there are well-established methods for quantifying polarization. For instance, at the level of Congress and the presidency, D-W nominate scores are extensively used (Pool & Rosenthal, 1985). Those scores are based on congressional votes, and are thus formulated after actual political behavior. Yet, such elite-level measures may not be reflective of the public and apart from self-descriptions, we do not really have comparable behavioral measures for the public. Which means that we are short on methodological solutions for a host of questions related to polarization among the public, the diffusion of polarization from elites to the public and more. Furthermore, existing measures are insufficient to study the effects of gerrymandering and redistricting. Instead of using opinion polls, which suffer a host of biases related to statistical inference, non-responses and more, we focus on election returns. We use a host of geospatial methods from GeoPandas, Shapley and geoplot to reduce polarization scores to the level of the electorate based on actual election returns. We measure polarization on various sub-state levels, including counties, metropolitan areas and Nielsen Designated media markets (DMAs). This score will be based on actual political behavior (voting) and will not be attached to congressional districts as the geographical units. This novel methodological suit will provide us with the tools to study a host of questions such as the extent to which gerrymandering indeed generated polarization; correspondence between levels of polarization in Congress and in the public; the extent to which the public (as opposed to decisionmakers) at various sub-state geographical units is polarizing; the interaction between polarization and turnout, political mobilization, and compliance with COVID-19 regulations; and the effect of campaign finance on polarization at various levels and geographical units.

Coauthor: Mr. Or Rappel-Kroyzer

Authors

• Udi Sommer, Tel Aviv University

• Or Rappel-Kroyzer, Tel Aviv University