Examining Impacts of a Supportive Housing and Child Welfare Experiment

Background and Purpose: Child welfare and homeless research both acknowledge the intersection of homelessness/housing hardship and child welfare. Homelessness can precipitate child welfare involvement, threaten family preservation, and/or delay reunification. Similarly, entering the homeless shelter system can heighten the risk of family breakup. In seeking to address homelessness in child welfare, the U.S. Children’s Bureau of DHHS funded five-year demonstrations of supportive housing. This paper is based on one project serving homeless/unstably housed families with open child welfare cases in a small Midwestern city, addressing three questions:1) Did supportive housing result in stronger child permanency and safety outcomes than usual services? 2) Did supportive housing result in greater housing stability than usual services? 3) How did the project use research to better understand disparate outcomes and what are the implications for more just child welfare practice?

Methods: Seeking to build evidence, programs were strongly encouraged to use a randomized experimental design to randomly assign eligible families to an experimental (supportive housing) or a control group (standard child welfare services). The randomized experiment relied on child welfare administrative data for child welfare outcomes and family surveys conducted by a contracted research firm for housing outcomes. Additionally, the local evaluation used multiple sources of data for families in supportive housing, obtained through repeated assessments of children and families and detailed information on changes in housing and family circumstances over time. In total 100 experimental and 86 control families were studied.

Results: 1) There were no statistically significant differences between experimental and control groups in the likelihood of family preservation, reunification, and subsequent maltreatment reports. 2) The experimental group experienced significantly more positive housing outcomes, including greater likelihood of having a lease in their name (p<.001), lower likelihood of subsequent homeless spells (p<.01), fewer subsequent evictions (p<.01), fewer housing changes (p<.01) and spent fewer days in shelter (p<.05) than controls. Neither child welfare nor housing outcomes demonstrated racial/ethnic disparities. 3) Examining process data from families in supportive housing, court oversight of the child welfare case, relapse with a substance use or mental health disorder, and criminal justice involvement during the service period all reduced the likelihood of children remaining with or returning to their parents’ care by project’s end. In more than one-quarter of the cases (regardless of group assignment), courts ultimately terminated parental rights.

Conclusions and Implications: The experiment found no impact of supportive housing on child welfare outcomes of permanency and safety, although housing stability improved even for families that did not stay together or reunify. The project was highly successful in decreasing homelessness/housing instability in this vulnerable population but creating a more just child welfare system requires broader access to quality treatment services and importantly, stronger support from the courts and law enforcement systems. In light of research findings, changes in how supportive housing is being implemented since the conclusion of the demonstration will be discussed.