Comparing Voting Districts with Uncertain Data Envelopment Analysis

TL;DR

This paper introduces a novel method using uncertain data envelopment analysis to evaluate voting districts, allowing for flexible, district-specific fairness assessments without relying on a fixed metric.

Contribution

It develops a new paradigm for comparing voting maps that avoids predefined fairness metrics, enabling more nuanced and optimal district assessments.

Findings

Effective evaluation of voting maps demonstrated

Method accommodates multiple fairness metrics

Assessments highlight differences in district fairness

Abstract

Gerrymandering voting districts is one of the most salient concerns of contemporary American society, and the creation of new voting maps, along with their subsequent legal challenges, speaks for much of our modern political discourse. The legal, societal, and political debate over serviceable voting districts demands a concept of fairness, which is a loosely characterized, but amorphous, concept that has evaded precise definition. We advance a new paradigm to compare voting maps that avoids the pitfalls associated with an a priori metric being used to uniformly assess maps. Our evaluative method instead shows how to use uncertain data envelopment analysis to assess maps on a variety of metrics, a tactic that permits each district to be assessed separately and optimally. We test our methodology on a collection of proposed and publicly available maps to illustrate our assessment strategy.