Association for Public Policy Analysis & Management

This paper draws on parent and child survey data from a 20-month follow-up wave of data collection in a 12 site experimental study of housing interventions for homeless families for 1,126 children age 8 to 17 to examine whether heterogeneity exists in well-being for children in homeless families and factors associated with profiles of risk or resiliency. The housing interventions studied include a permanent housing subsidy (primarily Section 8 vouchers), a temporary housing subsidy (community-based rapid rehousing), project-based transitional housing, and a usual care condition of continuing to work with shelter staff to obtain housing. First, we use latent class analysis to identify whether discrete profiles of risk or resilience in child well-being outcomes exist across multiple outcome domains, including education, physical health, mental health, social relationships, and self-efficacy. Second, we examine predictors of classes of well-being. These analyses will include both the direct experimental intent-to-treat effects of housing interventions and a mediation model where the intent-to-treat effects are mediated by their influence on housing stability and measures of parenting and home environment, controlling for parental challenges, child stressful life events, and child gender.

Results from the latent class analysis indicates heterogeneity in child functioning and that a two-class model best fit the data for both children age 8 to 12 and 13 to 17. Two-thirds of children in both age groups were identified as displaying similar profiles of resilience across all domains. One-third of children in each age group displayed elevated risk, particularly for mental health challenges, peer relationships, and below-average self-efficacy. Children age 13 to 17 in the higher risk group also displayed greatly elevated educational risks and challenges with substance use and police involvement.

Preliminary non-experimental analysis suggests that housing stability was not associated with children belonging to the resilient class. However, stronger family routines and a vigilant parenting style were associated with increased probability of children being in the resilient class. For children age 8 to 12, more chaotic home environments and more challenging parenting environment were associated with decreased probability of a child belonging to the resilient class. Initial findings suggest that many children in families that have experienced homelessness are resilient. Age-based differences in risk profiles and influence of parenting and home environment on resilience indicate the need for developmentally sensitive interventions. Final results will report the extent to which resilience was associated with assignment to experimental interventions and to what extent the influence of the housing and service interventions was exerted through changes in housing stability relative to influence on family routines, parenting style, and home environment.