Using HMIS and Homeless Prevention Services Data to Better Understand the Use and Outcomes of Permanent Housing Services
Partner(s): American Institutes of Research (AIR), US Department of Housing and Urban Development (HUD), and Montgomery County, MD.
Github Repo: https://www.github.com/dssg/hud_air/
Team: Erika Salomon, Iván Higuera-Mendieta, Adolfo De Unanue, Christina Sung, Rayid Ghani
Despite being one of the wealthiest counties in the United States, Montgomery County faces big inequalities. Near half of the renter population are “housing burdened”, and a better management and prioritization of services is needed to end homelessness in the County. As part of the County efforts to curtail this problem, DSaPP has teamed up with AIR and HUD to help the County to better design the frequent evaluation of their triaging tools, and their case management to minimize return to service and better outcomes to their target population.
Montgomery County is one of the wealthiest counties in the United States. With a 65% of home ownership rate, one of the highest in the country, and a house value 2.32 times higher than the national average; living as a renter in the County can pose an important financial burden. In 2015, 49% of the County renters reported spending more than one third of their incomes in rent, which can have a bearing on their homelessness risk. As part of the Montgomery County commitment to end homelessness, the local government has been testing new prioritization tools to improve their allocation of services, and to tailor their services to the target population needs.
As 2017, the County has been using the Calgary Homeless Foundation’s Acuity Scale to assess risks and needs upon entering permanent housing, track how those needs change over time, and tier case management interventions by providing more frequent case management to households with greater risk or need. After one year of implementation, the County wants to evaluate the predictive quality of the scale, by inquiring how well this scale identifies the return to homeless housing services.
The goal of this project was twofold. First, to help better understand the homeless service data, we wanted to have a detailed view of the County homeless services, and find different relationships between service features, (e.g. length and return to service, destination after service), and other individual features (e.g. disability status, and income). Also, in this process, we identified data recollection pitfalls, and suggest better ways of recollect and explore information. Second, and with the intent of improving the County’s allocation and management of homeless services, we investigated the predictive capabilities of the Calgary Homeless Foundation’s Acuity Scale. For this we propose a Study Design that provides a detailed roadmap to make a robust statistical assessment of the Acuity score ability to predict service management.
This project used two main data sources for homeless services in the County: Homeless Management Information System (HMIS) and the Enterprise Integrated Case Management System (EICM) at both Client and Service level; for analyzing the permanent housing services, the project focused uniquely in the latter. As a complement, we used the American Community Survey (ACS) for County level demographic comparisons.
Analysis and results
DSaPP cleaned and analyzed the data giving the county different data insights. First, we described the target population demographic composition and compared with the County historical distribution. We found the extensive use of homeless services by the County’s racial minority. Second, we thoroughly explored the HMIS service data finding a relevant relationship between retention in service and the likelihood of returning to another service, showing that a longer service attention can improve people’s outcomes.
Lastly, by exploring the Acuity Score, DSaPP recommends a Study Design that will help the county to evaluate the predictive ability of the score. This design proposal defines clear sample size variables, and times to have a robust assessment of the score ability to capture retention in permanent housing services, and its comparison with other standard available tools, like the VI-SPDAT.