Risk Models for Returns to Housing Instability Among Families Experiencing Homelessness

Zachary Glendening
Marybeth Shinn
Vanderbilt University

This study developed risk models for returns to housing instability (that is, homelessness and unstable doubling-up situations) among families exiting emergency shelter. Participants included 446 families randomly assigned to receive priority offers of longterm housing subsidies and 578 families randomly assigned to usual care in the Family Options Study, a multisite experiment designed to test the impact of various housing and service interventions for homeless families. Relationships between family features recorded at shelter entry and returns to housing instability 20 months later were examined empirically. Correlation, hierarchical logistic regression, and receiver operating characteristic curves were used to combine family features into predictive risk models. Results indicated that few observable family features beyond previous housing instability offered predictive utility. Access to long-term housing subsidies appears to reduce housing instability. Further research should examine whether disability benefits, reliable employment, or effective substance dependence treatment reduce housing instability.