Abstract: Outcomes from a Randomized Controlled Trial of Unconditional Cash Transfers on Child Welfare Involvement (Society for Social Work and Research 31st Annual Conference)

Schedule:

Sunday, January 24, 2027

Union Station 16 (Hilton San Francisco Union Square)

* noted as presenting author

Stacia West, PhD, Associate Professor, University of Tennessee, Knoxville, TN

Amy Castro, PhD, Associate Professor, University of Pennsylvania, PA

Abram Lyons, PhD, Lecturer, University of Delaware, Newark, DE

Frank Edwards, PhD, Associate Professor, Rutgers University, NJ

Zoe Bouchelle, MD, Chair of Inpatient Pediatrics, Denver Health Medical Center, CO

Allison Thompson, PhD, Executive Director, The Center for Guaranteed Income Research, University of Pennsylvania, Philadelphia, PA

Ryan Brown, PhD Candidate, Rutgers University

Ihsan Al-Zouabi, PhD Candidate, Rutgers University

The relationship between poverty and CPS involvement is well established. Anti-poverty policy interventions have been associated with improved child outcomes and reduced CPS involvement. Most studies have limited ability to infer causality. This study addressed the gap through a randomized controlled trial to test the impact of unconditional cash transfers (UCTs) of $500 per month over one year on CPS outcomes among families with a recent CPS referral across three counties in New York state. The primary aim was to determine if UCTs reduced the incidence of: 1) total child maltreatment reports; 2) confirmed child maltreatment reports; 3) enrollment in preventive services, and 4) foster care entries.

This parallel group randomized controlled trial included families with closed CPS cases under New York’s Office of Children and Family Services (OCFS) Family Assessment Response (FAR) program. Families with a FAR track case in Monroe, Onondaga, or Westchester county New York closed six months prior were eligible (N=800). 165 were allocated to treatment and 300 to control. Outcomes were measured from OCFS records: (1) 12 months prior to baseline, (2) 6 months into treatment (3) treatment cessation, and (4) a post-treatment period 6 months immediately following treatment close.

We used multilevel regression models with child-level random intercepts. Count outcomes were modeled using a Poisson likelihood and binary outcomes with Bernoulli likelihood. Models include fixed effects for time and county of residence. The sample average treatment effect (SATE) appears as an interaction between treatment status and study period.

We provide posterior medians and 95% highest density posterior credible intervals for the SATE in each study period. We provide the exponentiated SATE as incidence rate ratios (IRR) for count models and odds ratios (OR) for binary models. We conduct a two-tailed test adjusted for multiple comparisons with Benjamini-Hochberg adjustment calibrated for a 5% false discovery rate.

Maltreatment reporting was 21.4% lower in the treatment group (IRR 95% CI Period 2: [0.65, 1.20], p=0.57; Period 3: [0.41, 0.83], p<0.01); confirmed maltreatment was 22.9% lower in treatment than control (IRR Period 2: [0.64, 2.82], p=0.57; Period 3 [0.24, 1.11], p=0.13); preventive services were 14.6% lower in treatment (OR Period 2 [0.06, 0.72], p=0.02; Period 3 [0.53, 4.10], p=0.65); and no foster care entries occurred in treatment compared to 3.0 per 100 children in control (IRR Period 2 [0.00, 1.27], p=0.07; Period 3 [0.00, 0.61], p = 0.02).

Findings suggest UCTs of $500 per month for a period of one year for families previously involved in the child welfare system were causally associated with reduced child maltreatment reports, preventive services enrollment, and foster care entries. Confirmed child maltreatment reports were not impacted by the UCTs. These results represent the first causal findings of the impact of UCTs on child welfare involvement, and serve as a pivotal piece of evidence for future child welfare practice.