Selecting good redistricting plans from a large pool of available plans using the efficient frontier

Abstract

As part of a widespread frustration with partisan gerrymandering, many states have considered or implemented redistricting reforms – and others will eventually have to – that include a higher degree of citizen participation in proposing and evaluating redistricting plans. In some states without redistricting reform, public interest groups have created shadow commissions that encourage citizens to submit their own maps. For example, the new map for Pennsylvania Congressional districts, chosen by the state Supreme Court, was proposed by a citizens group.

As citizen participation grows, analytical methods for rating plans that recognize the different mapping criteria are needed to sort through multiple maps, both for highlighting good maps and for providing measures that allow courts to rule that a map is gerrymandered. Using a modified version of a model called data envelopment analysis (DEA), we present a nonpartisan approach that can score maps while not imposing any prior weights on the criteria. Our modification measures how close a plan is to the convex hull of the Pareto frontier when bigger is better for some criteria and smaller is better for others. Thus, we provide a novel and scalable way to filter out poor plans from large corpora of redistricting plans.