To date, the research that examines food accessibility has tended to ignore ethnic food outlets. This void leaves us with a limited understanding of how such food stores may, or may not, impact food security. The study discussed herein addressed this by conducting a geospatial assessment of ethnic food outlet accessibility in two U.S. cities: Flint and Grand Rapids, Michigan. We used Geographic Information Systems (GIS) tools to create a revealed accessibility index for each food outlet, and used the index to determine access within active travel service areas. We utilized an ordinary least squares regression (OLS), and two local models: spatial autoregression (SAR) and geographically weighted regression (GWR) to enhance our understanding of global and localized relationships between outlet accessibility and type (while controlling for known covariates). The results show that the local models outperformed (R2 max = 0.938) the OLS model. The study found that there was reduced access to ethnic restaurants in all service areas of Grand Rapids. However, in Flint, we observed this association in the bicycling areas only. Also notable were the influences that demographic characteristics had on access in each city. Ultimately, the findings tell us that nuanced planning and policy approaches are needed in order to promote greater access to ethnic food outlets and reduce overall food insecurity.
Keywords: GIS; GWR; demographic characteristics; ethnic food; food access; local regression; service area analysis; space syntax; urban design.