Redesigning a Risk Assessment for Homelessness Prevention Services through Research-Practice Partnerships

Background and Purpose: As housing costs continue to increase nationally, there is the potential for further growth in the already-large population experiencing homelessness. Focused prevention programs seek to avoid housing loss among people at risk for homelessness. The New York City Department of Social Services/Human Resources Administration’s (NYC DSS/HRA) Homebase program is one of the largest and most comprehensive homelessness prevention programs in the country. To ensure that more in-depth services are targeted to individuals most at risk of homelessness, Homebase utilizes a structured assessment, the Risk Assessment Questionnaire (RAQ). The assessment was developed in partnership with academic researchers, who analyzed self-reported items among Homebase applicants that were found to be associated with risk of homeless shelter entry over subsequent years. Separate RAQs were created for family with children (FWC) and single adult (SA) clients, as each have different risk profiles. Both initial RAQs were developed over a decade ago (the FWC RAQ using data on 2004-2008 applicants, the SA RAQ using data on 2004-2010 applicants), and several contextual factors have changed in NYC since then, including Homebase practices, housing costs, and City housing policies. Recently published work led by the NYC DSS Office of Evaluation and Research reevaluated and revised the content of FWC RAQ using more recent data. The current analysis does the same for the SA RAQ.

Methods: Utilizing Homebase program intake data and NYC Department of Homeless Services administrative data on shelter entry, we evaluated the SA RAQ’s performance in a recent cohort of single adults who applied for Homebase services from 2021–2022. We tested the predictive power of the current assessment to predict homeless shelter entry over the two years after Homebase application, using Cox proportional hazards survival models. We also used survival models to test the power of existing and potential new individual items that have been found in the literature to increase homelessness risk among single adults. The sensitivity and specificity of a revised measure was also examined, and findings were validated by examining trends by year of Homebase application and select client subgroups.

Results: We will share findings on the associations between current and potential SA RAQ items and the likelihood of shelter entry over two years. Discussion of proposed modifications to the RAQ will be accompanied by results from simulations of the modified RAQ to show changes in predictive power.

Conclusions and Implications: Given the ever-changing nature of populations at risk for homelessness, housing policy, and housing availability, it is essential to reassess the predictive accuracy of homelessness risk assessments over time. In collaboration with social work researchers, NYC’s Homebase program applied well-established scientific methods to develop and improve their risk assessment. Homelessness prevention services in other jurisdictions could similarly utilize research-practice partnerships to develop their own risk assessments or adapt the Homebase program’s tool. This could improve the efficiency of their services and ensure that they are helping those who are most at risk of homelessness.