Novel methods to construct a representative sample for surveying California's unhoused population: the California...

Existing literature on people experiencing homelessness (PEH) draws on nonrepresentative samples from service providers, populations with comorbidities, or areas with disproportionately high amounts of sheltered homelessness, leading to bias. Nearly one-third of PEH in the United States and more than half of unsheltered PEH live in California. We designed a rigorous state-representative survey of PEH to investigate the antecedents of homelessness, understand health, and inform policy solutions. The multistage design randomized at 3 levels: county, venue, and individual. Stratifying the state into 8 regions, we sampled 1 county per region to reflect statewide demographics. Within counties, sampled venues matched the expected proportion of sheltered and unsheltered residents. Within venues, interviewers randomly sampled individuals. We adjusted for nonresponse and incorporated poststratification to benchmarks. In parallel, respondent-driven sampling reached subpopulations through social networks who may otherwise have been undersampled. Our community-engaged study yielded 3200 quantitative surveys. We purposively sampled 365 participants for qualitative interviews. Demographic estimates match those found in the PIT with the added strength of statistical inference. To our knowledge, this is the first large representative study of PEH, beyond a single county, to draw inference on a large population that did not depend on service utilization. Our methods may inform future efforts to understand homelessness.