Association for Public Policy Analysis & Management

Positive psychological traits like hope, resilience, and self-control that are intuitively associated with success are predictive of justice system recidivism and earnings, and may provide similar insight into the dynamics of shelter spells.  Homeless shelter providers and case managers currently lack tools to make a priori distinctions between short-term and long-term shelter stayers; predictive models including these traits may enable them to target resources to those most likely to remain in shelter the longest, reducing overall shelter utilization and costs. This study is the first to examine the positive psychological traits of homeless families, and uses that data to model subsequent shelter use.

The study focuses on families applying for shelter through New York City’s Department of Homeless Services (DHS), the nation’s largest municipal shelter provider.  The head of household of approximately 700 families completed a questionnaire consisting of employment, demographic, and housing information and three previously validated scales of psychological capital measuring hope, self-control, and resilience.  Scores on these scales are compared to other baseline characteristics and populations, and matched to administrative data to model subsequent shelter use.  Explanatory and predictive models using each of the three traits are tested using Tobit models.  The study also creates a latent variable composed of the three traits; Confirmatory Factor Analysis is used to assess the validity of the new scale and a Structural Equation model is used to test its relationship with shelter use.

Multivariate regressions modeling length of stay and likelihood of exiting shelter suggest that two of the three traits are predictive of subsequent shelter use.  On a 5-point scale of resilience, a one-point increase is associated with a reduction of 21 shelter days; a one-point increase on the 8-point hope scale has a similar impact.  A latent variable created from the three traits is valid based on a Confirmatory Factor Analysis, and Structural Equation Models suggest that a one point increase is associated with a 60 day reduction in shelter stay.

These findings have implications for perceptions of homelessness and suggest solutions to address it.  Baseline results suggest that homelessness is not caused by a deficiency of psychological attributes that should help a family remain housed, and may be helpful in re-framing the discourse on the precipitating factors of homelessness.  More significantly for shelter providers, case managers, and policymakers, the predictive validity of psychological characteristics may be a valuable and efficient tool in targeting scarce resources to long-term shelter stayers most in need of additional resources.  Findings also suggest that inexpensive interventions integrated into regular case management, like motivational interviewing, may be effectively reduce shelter utilization.