Evaluating Racialized Economic Segregation in the Presence of Spatial Autocorrelation

Research on residential segregation has been active since the 1950s and originated in a desire to quantify the level of racial/ethnic segregation in the United States. The Index of Concentration at the Extremes (ICE), an operationalization of racialized economic segregation that simultaneously captures spatial, racial, and income polarization, has been a popular topic in public health research, with a particular focus on social epidemiology. However, the construction of the ICE metric usually ignores the spatial autocorrelation that may be present in the data, and it is usually presented without indicating its degree of statistical and spatial uncertainty. To address these issues, we propose reformulating the ICE metric using Bayesian modeling methodologies. We use a simulation study to evaluate the performance of each method by considering various segregation scenarios. The application is based on racialized economic segregation in Georgia, and the proposed modeling approach will help determine whether racialized economic segregation has changed over two non-overlapping time points.