Who are "the homeless"? Reconsidering the stability and composition of the homeless population.

OBJECTIVES: This study assessed the extent to which the use of point-prevalence samples biases conclusions drawn about homeless people. METHODS: Three data sets and published research were used to examine the degree to which conditions leading to point-prevalence bias (turnover in the homeless population, variability in the persistence of homelessness, and associations between personal characteristics and persistence) characterize the homeless population. Results were compared from point-prevalence studies concerning persistence of homelessness and characteristics of homeless people with those from a study of formerly homeless people. RESULTS: Conditions leading to point-prevalence bias strongly characterize the homeless population. Moreover, profiles of homeless people differed dramatically between point-prevalence studies and the study of formerly homeless people. In the former, average duration of homelessness was longer, and samples included higher proportions of men, minorities, non-high school graduates, and people with histories of psychiatric hospitalization, incarceration, and detoxification. CONCLUSIONS: Reliance on point-prevalence samples, when such samples are generalized beyond the currently homeless population, leads to overestimations of the persistence of homelessness, the demographic distinctiveness of the homeless population, and the prevalence of personal disabilities and deviant lifestyles among homeless people.