Abstract:
|
Despite intense interest in homelessness among policymakers, local officials, and the public, fundamental questions about the size and characteristics of this population remain unresolved. In this paper, we compare two novel sources of data on homelessness – the 2010 Census and the American Community Survey (ACS) – to HUD's widely cited point-in-time (PIT) estimates to advance our understanding of the number of homeless individuals in the U.S. and their coverage in these sources. We first narrow the scope of possible reasons for differences by reconciling those we can most easily identify, such as definitional differences and bias arising from the ACS's weighting methodology. Next, we test hypotheses about the role of seasonality, shelter list completeness, and PIT methodology in explaining sub-national differences between the 2010 PIT and Census. Finally, we link administrative shelter-use records from Los Angeles and Houston to the Census to evaluate the coverage and accuracy of both data sources, focusing on the usefulness of Census microdata to study the homeless population. The results provide insight on the ability of the Decennial Census to include hard to find populations.
|