Evaluating the Impact of the Shelter Diversion Program on Homelessness Prevention and Long-term Stability

The shelter diversion program we study is operated by Strategies to End Homelessness (STEH), which operates a centralized emergency shelter helpline and coordinates homelessness prevention and other services in Hamilton County, Ohio. Delivered by trained case managers, the Shelter Diversion program comprises three core components: rental assistance, housing identification support, and comprehensive case management services. The program is designed to provide consistent and tailored support to individuals and families at risk of homelessness.

The intervention begins with active assistance in identifying stable housing options. Once housing is secured, the program provides rental assistance (including security deposit, last month’s rent, and landlord incentive bonus, as needed). It also provides ongoing case management to address barriers to housing stability and long-term financial security. This program focuses on immediate housing needs and the underlying factors contributing to homelessness, such as financial instability and lack of access to supportive services. The intervention is strategically designed to stabilize participants’ living situations before they enter the emergency shelter system, thereby reducing the strain on shelters and improving participants’ long-term outcomes. Case management includes personalized action plans, regular check-ins, and connections to additional community resources as needed.

The intervention’s goal is to provide each participant with the necessary tools and support to secure and maintain stable housing, thereby preventing homelessness and fostering long-term stability. Participants in the control group will continue to receive standard housing referrals, while all other individuals not in the study will follow the usual intake process for housing assistance.

2024-11-13

2026-11-12

Emergency shelter use, street outreach engagement, and CAP Helpline calls.

We will analyze the following outcomes:

-- Emergency shelter stays: We will consider both indicators for whether the participant entered shelter, as well as the total number of emergency shelter days within the follow-up periods of 60 days, 90 days, 1 year, and 2 years after randomization. This outcome will measure the effectiveness of the Shelter Diversion Program in preventing participants from entering emergency shelters and reducing time spent in shelters. We will also measure number of days in shelter (where days are set to zero for those without a shelter stay).
-- Street outreach engagement: We will consider an indicator for whether a participant engages in street outreach services, during the same time windows after randomization. This outcome will assess whether the program reduces the need for on-the-street homelessness services.
-- Additional CAP Helpline calls: We will consider both an indicator for whether a participant newly calls the CAP helpline within 60 days, 90 days, 1 year, and 2 years after randomization. This outcome will evaluate the program’s impact on reducing continued housing instability that leads to repeated requests for assistance.

All of these outcomes will be tracked across multiple follow-up periods to capture both short-term and long-term impacts of the intervention.

Involvement in Other Homeless Services and Programs
Criminal Justice Involvement
Housing Stability and Neighborhood Quality
Educational Outcomes
Employment Outcomes
Access to Social Support Programs
Credit Outcomes

We plan to track involvement in other Homeless Services and Programs as tracked by STEH.

We also plan to track interactions with the criminal justice system.

Housing stability will be measured by eviction records as documented by Strategies to End Homelessness, as well as frequency of moves measured using Infutor address data. Alternative measurements will include address moves to lower socioeconomic status neighborhoods and neighborhood characteristics of most recent address.

Education outcomes will be evaluated using data from Project Connect, which provides services to children in Cincinnati Public Schools who are experiencing homelessness. This data includes, school enrollment, changes in school attendance, absences, and participation in Project Connect services.

Criminal justice involvement will be assessed through records from Strategies to End Homelessness, focusing on any interactions with the criminal justice system.

Economic and employment outcomes will be measured through employment status, earned income from the Ohio Department of Jobs and Family Services (ODJFS), and credit outcomes, including credit scores and balances, sourced from Experian.

Access to Social Support Programs will be monitored by tracking enrollment in SNAP, TANF, Ohio Works First, and other social support programs, as recorded by ODJFS and STEH. We will also track usage of food patnries and transportation assistance as recorded by STEH.

Using a randomized controlled trial design, LEO and STEH aim to answer the research question: What is the impact of offering support in identifying stable housing, 6 -months of rental assistance, and robust case management services on individuals’ housing stability? Study enrollment will occur starting in November 2024 and continue through November 2026, with the possibility of extending for another year if feasible.

During the initial intake conversation, CAP Helpline staff will screen new clients for eligibility for the Shelter Diversion Program. Clients will be eligible to participate in the study if they meet the program’s criteria, including having no income or an income at or below 15% of the area median income (AMI) and having stayed or slept in two or more places in the past 30 days. Once the staff member determines that a client is eligible, they will inform them about the study and give them the opportunity to consent to participate. All individuals who are eligible for the program will be randomized each week, but only those who consent will be included in the study. The randomization odds will be adjusted weekly based on the number of open spots available in the program. Randomization will take place in Stata, ensuring that the assignment to either the treatment or control group is done equitably and consistently. STEH will send a list of all randomized clients and baseline data to the LEO research team each week. The research team will then track participants’ outcomes using administrative data from relevant state agencies, analyzing the impact of the intervention on both short- and long-term housing stability and other key outcomes.

The Shelter Diversion Program’s services, including rental assistance and case management, are part of the standard care offered by Strategies to End Homelessness. These services have been designed to prevent individuals and families from entering emergency shelters by providing them with the necessary support to secure stable housing. While these services are already part of the program, the current study aims to rigorously evaluate their effectiveness through a randomized controlled trial. The purpose of this project is to determine the impact of providing these established services on outcomes such as emergency shelter stays, housing stability, and long-term financial security for clients experiencing housing crises. The results of this study will help STEH refine their service delivery and better allocate resources to maximize the impact of their interventions. If the study demonstrates that these services significantly improve outcomes for clients, it could support the case for scaling up these efforts, potentially benefiting a larger population. Additionally, the findings from this research will contribute valuable insights to the broader field of homelessness prevention, informing best practices for similar programs nationwide.

Not available

Computer-based randomization

Unit of randomization: households

No

1,760 households (including both families with dependents and single adults)

1,760 households

Control: ~1,547 households Treatment: ~213 households

The study will conduct enrollment over a period of two years, with the option to extend to a third year if feasible. During this period, we will track interim outcomes while continuing to enroll participants. The primary comparison will be between individuals who receive the full suite of Shelter Diversion Program services, including rental assistance and case management, and those who do not receive these services. For two years of enrollment, we anticipate enrolling approximately 1,760 individuals, with ~213 individuals in the treatment group and ~1,547 in the control group. According to STEH data, 14.9% of callers who are eligible for Shelter Diversion but do not receive those services entered shelter within 12 months. With an expected take-up rate of 75-90% in the treatment group, we are powered to detect an 8.3-9.2 percentage point decrease in the likelihood of emergency shelter use from this baseline, which corresponds to a 55.6-61.6% decline. Additionally, we will measure additional calls to the CAP Helpline. According to STEH data, about 54.7 percent of callers eligible for Shelter Diversion but who do not receive services call back the CAP Helpline within 12 months. . With the anticipated sample size and take-up rate, we will be able to detect an 11.6-12.9 percentage point decrease in the likelihood of repeated CAP Helpline calls from this baseline, representing a 21.2-23.6% decline. If enrollment is extended to three years, increasing the sample size to approximately 2,640 individuals, the minimum detectable effect sizes will be smaller. For emergency shelter use, the detectable effect size will decrease to 6.8-7.5 percentage points, representing a 45.5-50.2% decline. For repeated CAP Helpline calls, the detectable effect size will decrease to 9.5-10.5 percentage points, representing a 16.4-19.2% decline. Combining two to three years of enrollment, and depending on program take-up and study duration, we are powered to detect a 45-62% decline in shelter use and a 16-24% decline in calling CAP again within the 12 months after randomization.