A Logical Model Approach to Detect Racially Polarized Voting

• In Event: Issues in Redistricting Reform (Virtual)

Abstract

Detecting whether minority and White voters systemically support different candidates in elections or racially polarized voting has been crucial in the research of American political behavior and jurisprudence of race, redistricting, and voting rights. I propose a novel approach to detect racially polarized voting based on a recently introduced logical model of minority representation, which only requires information about district racial composition and the vote shares of the most vote-earning minority and white candidates. Unlike previous methods focusing directly on individual voting behavior, the proposed method quantifies the deviation of actual election results from expected results in the presence of the strongest racially polarized voting. Applying this approach to district-level election data, I demonstrate that racially polarized voting still prevails both in Southern and non-Southern states and has not waned over time. Finally, I analyze the variation in the deviation and discuss several implications for Section 5 of the Voting Rights Act.

Author

• Yuki Atsusaka, Rice University