What Makes a Gerrymander? An Inductive Approach to Understanding Judicial Action Using GIS-based Gerrymandering Indicators and Statistical Models.

Abstract

Analysis of political districting for possible gerrymandering often proceeds deductively, working from precedent, common law, and principles of good governance to derive standards and metrics for identifying gerrymanders. Jurists have identified both national requirements for districting such as “one person, one vote” and traditional redistricting principles which may be considered by states (such as compactness, respecting communities of interest, accordance with municipal boundaries and prior districts, and protecting incumbents). Academics have also contributed, including by proposing metrics for evaluating compactness and political bias (e.g., the efficiency gap). However, jurists and academics have given little consideration to inductive approaches. Unlike deductive approaches that begin with ideas of how districts ought to be drawn, inductive approaches start from the districts that courts have struck down as gerrymandered and work backwards, identifying demographic, political, and geometric characteristics that have in practice been associated with gerrymandered districts.

Such an inductive approach opens consideration of several questions not addressable solely via traditional, deductive analyses:

• Do the data suggest that courts adhere to the standards and metrics cited in their decisions when adjudicating the lawfulness of districting?

• Are there other correlates not identified by courts that are associated with being adjudicated as gerrymandered?

These questions were addressed through correlation and logistic regression analyses conducted during an NSF REU. Results may assist courts in revising their approaches to adjudicating gerrymandering, either to ensure that gerrymandered districts are identified based on established principles or to revise these principles to account for the additional correlates identified here.