Project Lead: Daryl DeFord, Assistant Professor of Data Analytics in the Department of Mathematics and Statistics at Washington State University
DSSG Fellows: Rowana Ahmed, Katherine Chang, Ryan Goehrung, Michael Souffrant
Data Science Leads: Bernease Herman, Vaughn Iverson
Gerrymandering, the manipulation of legislative district boundaries for personal or partisan gain, is a fundamental threat to democracy. While the history of American gerrymandering extends back 200 years, representing a wide variety of specific aims and harms, recent advances in computational methods and data analysis have exacerbated this problem. However, computational methods may also be used to evaluate proposed maps more efficiently and flag outlying maps as potential cases of gerrymandering before they are adopted.
This summer, we used modern computational methods and statistical techniques to evaluate potential tradeoffs between redistricting criteria and proposed methods for challenging newly drawn plans. Recent evidence shows that Markov chain Monte Carlo (MCMC) methods can establish baselines of neutrally-constructed maps, or ensembles, which can then be used to evaluate proposed districting plans along a variety of dimensions. We used the GerryChain software in Python to construct neutral ensembles tailored to individual state’s geographies and rules, as well as analyzed the properties of the resulting maps with respect to the relevant legislative and litigation history.
The main objectives of this project were to gain insights into GerryChain’s potential by applying it to unique redistricting problems in three states, Georgia, Colorado, and Texas, and use this experience to develop a comprehensive user’s guide to increase access to the GerryChain toolset. As the 2021 redistricting cycle is currently underway, this project has the potential to empower community members, activists, and non-partisan redistricting committees to use GerryChain to participate in their state’s redistricting process more effectively.