Following residential segregation by race spatiotemporally: A search for causality.

Objective: The study tried to test three existing explanations on residential segregation through spatiotemporal modeling: (1) spatial assimilation, (2) racial prejudice and discrimination, and (3) residential preferences. Since residential segregation is fundamentally a spatial and longitudinal phenomenon, the analysis is expected to provide a fresh perspective on residential segregation. Method: The study used a spatiotemporal modeling through integrated nested Laplace approximation, which is a Bayesian approach to very complex regression modeling, and the data came from the American Community Survey data from 2005 to 2019 at two different spatial levels—Public Use Microdata Area and census tracts. Results: The findings of this study showed that: (1) socioeconomic variables, including income, unemployment, and poverty, have meaningful impacts; (2) housing‐related variables do exert a significant impact on black household concentration; (3) the ratio of Asian and Hispanic populations turns out to be important predictors in a negative direction; and (4) even with the predictors, the non‐trivial temporal and spatiotemporal effects are discovered. Conclusion: The study reveals that a reductionist approach to residential segregation is unlikely to offer meaningful insights into the problem and that proper understanding is possible only when careful consideration of the spatiotemporal context is taken, including the spatial level. [ABSTRACT FROM AUTHOR]