Judging Gerrymandering: Improving Methods for Measuring Partisan Distortion and Its Component Parts

Abstract

This paper improves upon the existing mathematical methods for measuring partisan distortion and evaluating partisan gerrymandering in plurality-won, single-member district electoral systems. The measures and models presented in the paper are evaluated using data from Congressional, State House, and State Senate elections in the state of Florida. How can partisan distortion be measured in a manner deemed acceptable by the U.S. Supreme Court? How are metrics of distortion affected by their implicit assumption of uniform or uncorrelated district turnouts? Can the component of distortion due to non-uniform district sizes be extracted from these metrics in attempt to isolate the pure gerrymandering component of distortion? Does the creation of a minority opportunity district create partisan distortion? First, I present a conceptually improved extension to the method of uniform partisan swing, called "diagonalized partisan swing," which accounts for non-uniform, correlated district turnouts. Second, I successfully extend Ralph Brookes' method for isolating the components of partisan bias to the efficiency gap metric. I find that, for both the efficiency gap and partisan bias metrics, there are non-negligible differences between the isolated gerrymandering component of the metric and the metric itself. Third, I use algebraic and statistical models to argue that the creation of a minority opportunity district creates partisan distortion in a manner that disfavors the minority group's preferred party. The novel models presented in this paper lay the foundation for further investigation into the sources of distortion and their proper measurement. These advancements are important as the judiciability of partisan gerrymandering will remain in question until a robust method for its measurement can be developed.