Growing Strong Study of Unconditional Cash Transfers Plus Peer Support for Families With Babies in Homeless Shelters

In the Growing Strong study, 200 families residing in the Women in Need (Win) homeless shelter system who have at least one child under two years of age living with them in shelter and who have been found "eligible" for shelter by New York City will participate. Families will be assigned to one of two cash gift conditions. One hundred families will be assigned to the active intervention group and will receive $1,500 per month via debit card for 24 months ($18,000 annually) to use as they wish along with access to free, voluntary peer support services. One hundred families will be assigned to the active comparison group and receive $50 per month via debit card for 24 months ($600 annually) to use as they wish but will not have access to the peer support services. A passive comparison group will be constructed from the administrative data of propensity-matched families within in the New York City homeless shelter system who would have met the criteria to participate in the study had they resided in one of the Win target shelters. This study is funded by a grant to Win from a private foundation, thus all target shelters are operated by Win.

Design

The Growing Strong study design is quasi-experimental with intentional assignment at the shelter level rather than experimental with random assignment at the family or shelter level for several reasons. Participants were not randomized at the family level in order to eliminate the possibility that comparably situated families living in the same shelter could enroll in the study and yet receive very different cash amounts. Participants were not randomized at the shelter level because there were too few shelters in the final design for randomization to balance the characteristics of families across conditions in expectation. The overall sample size for families was limited by the substantial cost of the intervention (between $600 - $18,000 plus costs of peer support annually per family).

Families can be assigned to treatment conditions based on their shelter placements because New York City makes placement decisions in a quasi-random fashion based on the availability and size of shelter rooms/apartments. Initially the investigator(s) selected three shelters within the Win shelter system that are as closely matched as possible on room sizes and other characteristics that affect placements. When recruitment went more slowly than anticipated, additional shelters in the Win system were included. Assignment to treatment groups is dependent solely on the shelter where the family is assigned by the New York City Department of Social Services, which depends on availability and room size. We have sought to balance shelters assigned to treatment and comparison groups by the sizes of rooms. The underlying, quasi-random City placement process should equate the two groups in expectation. The investigator(s) will check whether families in the two groups are in fact well-matched and control for any differences between them during analysis if needed.

Procedures

As the shelter operator, Win has information about families who are deemed eligible for shelter, including information to determine whether a family meets the Growing Strong eligibility criteria. Win administrative staff will notify the Win staff member who is the liaison to the study when a family meets the criteria to participate. This staff member will explain the study to families. Families who meet the criteria and would like to learn more will be referred for potential enrollment.

Families who choose to enroll will take part in a baseline survey and be invited to participate in follow-up surveys 6, 18 and 30 months later. The baseline survey collects data on demographic characteristics, family composition, work experience, income, finances, adult health and well-being (including use of alcohol and drugs), food security, social support, parenting, family routines, and child development. Measures refer to the adult participant (typically the mother), the family as a whole, or the young child who makes the family eligible for the study (the target child). Follow-up surveys mostly consist of the same measures taken at baseline plus additional information about housing and neighborhoods. The 30-month follow-up survey, which will take place 6 months after the family has stopped receiving cash, will also examine how the family is faring without the cash support.

The research team will conduct qualitative interviews with 15 families in each of the active study groups. These 30 families will be representative of the overall sample with respect to demographic characteristics. The qualitative interviews will ask about respondents' experiences in the study, including the cash (e.g., what they used it for, how it changed things for them) and the peer support services (e.g., did services help them to get out of shelter and, if so, how). In addition to probing the topics in the quantitative surveys, qualitative interviews will ask about respondents' goals and family processes. Qualitative interviews will take place 6, 18, and 30 months after enrollment.

Families who choose to enroll in the study will provide consent for us to collect administrative data for members of their household from various state and local agencies. Administrative data will include length of stay in shelters, returns to shelter, involvement with child protective services, income from employment, and emergency and routine health care, among other things. Administrative data will be used to compare outcomes for the families enrolled in Growing Strong to those of similar families outside of the study who receive usual care in the New York City homeless shelter system. Administrative data will be collected by New York City's Center for Innovation through Data Intelligence (CIDI).

Safeguards

Growing Strong has received a waiver from New York State that exempts cash gifts from the study in income calculations for benefits provided or mediated by the state (e.g., TANF) for as long as the participant is enrolled. Though families are not expected to withdraw from the study, they may do so at any time, effectively ending the protection of the waiver if they choose to continue to receive the cash gifts (which are unconditional) after withdrawal.

In an effort to hold families harmless, the investigator(s) have developed procedures that protect families' benefits under the waiver to the greatest extent possible. More specifically, the research team will wait to collect all administrative data except shelter exit time until after families have received their cash gifts in full (e.g., after 24 months of participation). Thus, any families who wish to withdraw from the study can time the withdrawal to occur after receiving the final cash gift but before administrative data are pulled for their household. A dedicated Win staff member will help families who choose to withdraw with the timing of withdrawal in order to protect/preserve their benefits under the waiver. Though the study has this safeguard in place, families are not expected to choose to withdraw from the study.

The study has also been granted a Certificate of Confidentiality from the National Institutes of Health. This means that the study cannot release or use information that identifies families in any action or suit without the agreement of the family. This protection includes civil, criminal, administrative, and/or legislative proceedings at the federal, state, and/or local level.

Outcomes

The primary outcome is length of time between families' being found eligible for shelter and shelter exit assessed with survival analysis (Cox proportional hazards).

The investigator(s) will test for differences between the active intervention and active comparison groups on additional outcomes at each follow-up point controlling for race and language of interview as covariates. If the majority of participants are not women, as anticipated, analyses will also include controls for gender. If overall differences between groups at baseline exceed chance estimates, analyses will control for baseline measures that differ beyond p < .1. If follow-up rates fall below 90%, analyses will also control for propensity to respond to follow-up surveys in order to account for any differential attrition.

Following the practices of other studies of interventions to assist families experiencing homelessness, analyses will use a .10 level for statistical significance and not adjust for multiple testing. To avoid preferential selection of outcomes ("cherry-picking"), the investigator(s) have designated a single primary outcome and one or at most two measures per outcome domain as "key secondary outcomes" to feature in all reports.

Following the 30-month/final follow-up interview, the investigator(s) will use hierarchical linear models with time points (max = 4) nested within respondents to examine trends over time. Increases in most measures of well-being are expected as families in both groups stabilize in housing as well as treatment-by-time interactions showing the superiority of outcomes for the active intervention group.

Secondary Outcome Notation

Secondary outcomes have been pre-specified and grouped into the following nine domains: Housing, Target Child Development, Parenting, Family Separation, Adult Well-Being, Domestic Violence, Financial Well-Being, Employment, and Health Care. Secondary outcomes are labeled as follows in subsequent sections of the pre-registration: Domain names are prepended to outcome names and an asterisk is appended to outcomes designated as "key" secondary outcomes. For example, the key secondary outcome "Family Use of Emergency Medical Care or Hospitalization" is labeled "Health Care: Family Use of Emergency Medical Care or Hospitalization*."