Examining the association between shelter status and homelessness duration: A case study of Halifax, Canada

Several studies have investigated factors associated with urban homelessness and its duration. Most focus on social and health factors and overlook shelter status, potentially due to reliance on datasets from sheltered homeless individuals. Shelter status, defined as whether an individual was staying in formal temporary accommodation or in an unsheltered location not intended for habitation, represents a critical distinction in subsequent housing pathways. Using a near-complete count and survey of homeless individuals in Halifax, Canada, this study examines the association between shelter status and homelessness duration. The study also examines whether shelter status moderates the association between social and health-related factors and the likelihood of exiting homelessness. The analysis consists of three steps. First, factors related to homelessness duration are identified using Cohen's d and the t -test. Second, a baseline proportional hazards model is estimated, with homelessness duration as the dependent variable, shelter status as the explanatory variable, and social and health factors as control variables. Third, a model including interaction terms between shelter status and social and health factors is developed. The results show that unsheltered status is associated with prolonged homelessness, along with significant interactions with immigrant status, first homelessness after age 50, victimization, and substance use history. These patterns are examined using survival curves to understand differences in exit probabilities over time. Findings indicate that shelter status is associated with homelessness duration and moderates the association between individual vulnerabilities and duration.