This project examines how communities of interest and spatial segregation contribute to the practice of gerrymandering. We design realistic synthetic city models featuring varied population and minority distributions to simulate real-world scenarios. By modifying these data models, we develop a framework aimed at detecting gerrymandering.
How do communities of interest and spatial segregation facilitate gerrymandering? We hypothesize that increasing spatial segregation and larger communities of interest reduce the number of favorable redistricting options available to minority groups.
Synthetic cities are modeled on grid-based structures that reflect typical population densities, highest at city centers and decreasing outward linearly or stepwise. Minority populations are distributed in varied patterns, and community sizes are adjusted to test the spatial segregation index. We analyze the resulting "state space" of possible redistricting maps.
We have implemented the Absolute Clustering Level index (Massey and Denton, 1988) and Moran's I to quantify how clustered minority populations are, allowing segregation to be treated as a measurable variable.