A Randomized Controlled Trial on Permanent Supportive Housing for Individuals Exiting Prison

How to cite

Batistich, Mary Kate; Cassidy, Michael; Evans, William. 2023. "A Randomized Controlled Trial on Permanent Supportive Housing for Individuals Exiting Prison." AEA RCT Registry. AEARCTR-0011260. https://doi.org/10.1257/rct.11260-1.0

Download: BibTeX · RIS — for reference managers without a browser translator. Zotero and most others pick up this page's citation metadata directly.

Mary Kate Batistich* , Michael Cassidy , William Evans

* Primary investigator

Registry ID
AEARCTR-0011260
DOI
https://doi.org/10.1257/rct.11260-1.0
First registered
Last updated
Status
In development
Trial dates
2023-05-01 – 2030-11-01
Countries
Private
JEL codes
I38 (Government Policy/Provision and Effects of Welfare Programs), R28 (Household Analysis / Government Policy), H00 (Public Economics/General), J01 (Labor Economics/General)
Registry record
http://www.socialscienceregistry.org/trials/11260

Abstract

More than half a million prisoners are released from state and federal prisons each year. Formerly incarcerated people are nearly ten times more likely to experience homelessness than the general population, and the risk is highest in the first two years after leaving prison. Permanent supportive housing (PSH), a housing first model, provides affordable housing and support services to individuals facing homelessness, without time limits. PSH is a popular strategy for achieving stability and independence among vulnerable populations; there were over 375,000 year-round PSH beds in the United States in 2021. We will conduct and evaluate the first randomized controlled trial (RCT) of PSH for exiting prisoners at risk of homelessness. Through a partnership with the Corporation for Supportive Housing (CSH), a national leader in supportive housing, we will implement the study through two Ohio-based programs, Returning Home Ohio (RHO) and Community Transition Program (CTP). The programs offer the same services but differ in their eligibility criteria: RHO serves exiting prisoners with mental illness or HIV/AIDS, while CTP serves those with substance use disorders. We will measure the impact of PSH on criminal recidivism, housing stability, health, income, and employment. The study will enroll approximately 880 people, of whom 25 percent will be in the treatment group.

Experimental design

CSH accepts applications for placement in their PSH program on a rolling basis. While individuals are incarcerated, linkage workers, who are employed by Ohio’s Department of Mental Health and Addiction Services (OMHAS), will identify those who may be a good fit for one or both of the PSH programs. For the Community Transition Program (CTP), this means that the individuals will have participated in recovery services while incarcerated, and for Returning Home Ohio (RHO), this means that they will be diagnosed with a serious and persistent mental illness or be HIV positive. Linkage workers will conduct JD-VI-SPDAT assessments (which assess vulnerability and identify housing needs) and complete program applications for eligible individuals interested in participating in the program. Many individuals who are eligible for one program will also be eligible for the other.

The intake process will include a study consent form explaining the purpose of the research, the randomization process, and what personal information will be shared with the research team for those consenting to participate in the research study. Potential study participants will go through the consent process with a linkage worker, who will be able to answer questions. Consent to participate in the study will not affect the likelihood of being offered a spot. As there are many more exiting prisoners eligible for the programs than available slots, all slots will be offered via a lottery.

All individuals who are referred to either program, regardless of whether they consent to participate in the study, will be placed on CSH’s master waitlist. Eligible formerly incarcerated individuals released without having been screened for the programs can be referred to CSH by community partners within 120 days of release and added to the master waitlist.

Once per month, CSH meets with the local housing agencies who implement the programs. During these meetings, they identify the number of placements that are available in each location and any restrictions associated with those placements. For example, if an agency does not accept people with an arson or sex offense. After the specific placements have been identified, CSH will identify the individuals most recently added to the waitlist who qualify for each placement. They will attempt to contact these individuals one at a time to ascertain their continued interest in the program. Past experience shows that many people will not have working cell phone numbers or email accounts so they will be removed from the waitlist after three attempts to contact. For every available PSH slot, the CSH associate program manager will identify a set of four eligible participants that are still interested in the program. From this list, they will randomly select one person to be referred to each spot using a randomization tool that LEO has designed. Although we plan on a 3:1 ratio of controls to treatment, this will be a function of demand at the local level and some agencies may have lower levels of control to treatment ratios.

Individuals who are placed in the treatment group will be referred to the relevant housing agency, which will assist them in their housing search and attempt to place them in suitable housing. When their housing is secured, the individual will be able to use the PSH voucher the housing agency provides. Once placed in housing, the individual will be added to an intensive case management system (i.e.., the “supportive” component of PSH) and introduced to opportunities to access mental health treatment, workforce development, childcare, and healthcare. The housing agencies connect their clients to housing and supportive services. They do not directly house participants themselves. Participants who are placed in the control group will be given information on other services in the community that could be beneficial to them.

Design details

Randomization method
Computer random number generator
Randomization unit
Individual
Clusters
N/A
Observations
880
Arms
e anticipate 275 individuals to be randomized to Treatment and 825 individuals to Control. With an assumed 80% rate of consent to participation in research, our analysis sample would include 220 individuals in Treatment and 660 in Control. The exact number and ratios will depend upon the fraction of applicants who consent to participate in the research and the size of the waitlist relative to the number of open spots for each county-specific program each month.
Primary outcomes
The initial goal of the program is to reduce the incidence of homelessness of exiting prisoners. Our first primary outcome is then whether the program improved this outcome. We plan to track homelessness through Homeless Management Information Systems (HMIS) in relevant locations. This data will allow us to capture participants’ use of housing services. We also plan to track housing stability measured by frequency of moves using the consumer reference database Infutor. Homelessness is thought to be a precursor for future criminal involvement, so our other primary outcome is criminal recidivism (e.g., arrest, imprisonment, length of time incarcerated) within 1, 2, and 3 years after randomization. We will use the state’s Department of Corrections data to measure criminal justice outcomes.
Secondary outcomes
Secondary outcomes capture participant wellbeing through health, employment, and earnings. For health outcomes we plan to use state Medicaid data. The state where the experiment will occur is a Medicaid expansion state and most released prisoners will then be eligible for Medicaid. Outcomes here will include emergency department visits and inpatient hospital stays. Employment and income data will be obtained through state unemployment insurance data. These state records provide quarterly values of earnings in the sectors covered by unemployment insurance. Outcomes will include earnings from all jobs in the quarter and a dummy variable that measures whether a person worked in the quarter. This data does not have hours of work. Our ability to track these outcomes will be contingent upon securing data sharing agreements with relevant partners.

Cited by

No citation data available for this registration yet.

Keywords