Compare Spatial and Multilevel Regression Models for Binary Outcome in Neighborhood Study

Hongwei Xu, University of Michigan

The widely used multilevel regressions in neighborhood research typically ignore potential between-neighborhood correlation due to underlying spatial process, and hence produce inappropriate inferences about neighborhood effects. In contrast, spatial models make estimation and prediction over space by explicitly modeling the spatial correlations among observations in different locations. A better understanding of the strength and limitation of spatial models as compared to multilevel models is needed to improve the research on neighborhood and spatial effects. This research systematically compares model estimation and prediction for binary outcomes between spatial and multilevel models in presence of both within- and between-neighborhood correlations through simulations. Preliminary results show that multilevel and spatial models produce similar estimates of fixed effects, but different estimates of random effects. Both the multilevel and pure spatial models tend to overestimate the corresponding random effects, compared to a full spatial model when both non-spatial within neighborhood and spatial between-neighborhood effects exist.