Examining the Validity of Risk Assessment in Domestic Violence Shelters: Informing Clinical Practice and Fostering Research-Practice Collaboration

Background and Purpose: Domestic violence (DV) is a major cause of housing instability. In New York City (NYC), emergency housing for DV survivors is available through the Human Resources Administration’s No Violence Again (NoVA) program. Applicants who indicate during shelter intake that they experienced DV are immediately referred to the on-site NoVA program, where a social worker completes a multi-part clinical structured assessment of current risks for DV re-occurrence and intimate partner homicide (IPH). Respondents are provided emergency housing based on the risk level identified. As this program-developed assessment has been in use for over 10 years, program leadership wanted to ensure that it reflects the current evidence base. Therefore, we evaluated the face and content validity of the NoVA assessment by comparing its content to the DV risk factors identified in research and literature and to validated DV assessment tools. We also compared the population of shelter applicants who were administered NoVA assessments to the populations on whom other validated DV assessment tools were tested.

Methods: We conducted literature searches to identify key risk factors for IPH and DV re-occurrence and validated DV risk assessments. We focused on risk factors with the highest-quality evidence (i.e., identified in more than two studies that had high methodological rigor and/or a systematic review or meta-analysis). Key risk factors were organized into perpetrator, relationship, and household characteristics, then mapped against the NoVA assessment items to determine areas of overlap and potential gaps. We then compared the item content and scoring of validated risk assessments to that of the NoVA assessment. Finally, we compared sociodemographic characteristics (sex, race/ethnicity, marital status, and education) of clients referred to receive NoVA assessment to characteristics of the test populations of validated risk assessments, as available in the literature. Sociodemographics of NoVA-eligible clients were collected via administrative data from the NYC Department of Homeless Services.

Results: The NoVA assessment showed evidence of face and content validity, capturing the majority of key risk factors for DV re-occurrence and IPH identified in the research literature. These include escalation of abuse, the perpetrator having access to a gun, and the perpetrator experiencing substance misuse. We also identified seven validated assessments of DV re-occurrence and/or IPH for DV survivors. While assessment content varied, overall, their content and scoring were similar to the NoVA assessment. Finally, validated DV assessments were tested in populations roughly similar to those referred for NoVA assessment in terms of gender, age, race/ethnicity, and education. Though no validated assessments were developed specifically for those experiencing housing instability, most were tested in urban areas, and many samples included shelter clients.

Conclusions and Implications: The NYC NoVA DV risk assessment, long utilized in clinical practice, showed evidence of validity. More broadly, the research fostered partnership-building between clinical staff and researchers, which can be valuable in supporting programmatic enhancement. Findings are now being utilized to inform a second study phase analyzing item-level assessment responses, which will guide decisions about streamlining the assessment.