Unhoused and Mismeasured: The Accuracy of Surveys of Those Experiencing Homelessness

We examine the accuracy of self-reported information from homeless shelter residents in the American Community Survey by comparing survey responses to administrative records. While basic demographic information (age, gender, and citizenship) is reported with reasonable accuracy, we find substantial misreporting of economic outcomes, with error rates that tend to be moderately higher than those of housed populations. Notably, homeless respondents have high false negative rates for formal employment and SNAP receipt, leading to significant underreporting of these important sources of income. These findings suggest that survey-based research on homelessness may reliably capture demographics but substantially misrepresent economic outcomes.