Measuring Historical Bias

Abstract

This chapter briefly examines several existing measures of partisan bias and responsiveness: mean/median difference, partisan symmetry, and efficiency gap. It argues that while these measures tend to be well-adapted when applied to swing states under close electoral conditions, they are inappropriate when applied to noncompetitive states and wave elections. It introduces a new measure referred to as “historically weighted efficiency gap” (HWEG) that estimates bias under the expected range of electoral conditions for any given state. It shows how this measure can be adapted to visualize bias and responsiveness across all plausible election results. The chapter details the data and steps used in the calculation and shows three examples of its use in recently contested gerrymanders (in Pennsylvania, Maryland, and Arizona). The chapter also discusses the strengths and weaknesses of HWEG and describes which states stand out as extreme results under this measure.