A Spatiotemporal Analysis of Evictions & Homelessness in the USA

Abstract

This bivariate spatiotemporal study primarily investigates clusters of homelessness and evictions on the county level in the contiguous United States between 2007 and 2018. Further investigation using a negative binomial regression model explores factors commonly associated with homelessness and evictions, as well as the relationship, if any, between the two. Using software for spatial statistics (SaTScan (Formula presented.)), mapping (ArcGIS Pro), and statistical analysis (SAS), the following are identified: geographic variations in clusters, numerous significant covariates, and no strong evidence suggesting an association between clusters of homelessness and evictions in the contiguous United States. This study uses an epidemiological spatiotemporal approach to pinpoint and address regions that exhibit abnormally high counts or rates of evictions and of homelessness. These clusters of counties could be of interest for future investigations into underlying factors contributing to these disproportionate counts or rates.