AL-SHIELD: Evaluating Human-Guided Review for Alabama Child Care Subsidy Integrity and Fraud Prevention - The University of Alabama and the Alabama Department of Human Resources (DHR), the state’s CCDF Lead Agency, propose a four-year evaluation of one program-integrity practice, AL-SHIELD. We will test whether this structured, evidence-guided review helps DHR find more substantiated intentional program violations for the same review hours, without harm to families and providers. Our partnership dates from 2019, and in 2024 ACF awarded it a CCDF Lead Agency Data and Research Capacity grant (Administration for Children and Families, Office of Planning, Research, and Evaluation, 2024). AL-SHIELD changes one step in DHR’s existing review. Before a reviewer acts on a case, the reviewer opens a one-page evidence card. It assembles the records in force on the service date, shows the evidence for and against a concern, and recommends a review level within the hours DHR has already allocated. The review hours, the rules of evidence, and the appeal rights are the same whether or not a reviewer uses the card. The reviewer decides each case, and only a completed DHR review can establish intent. A completed review ends in a) no issue, b) agency or system error, c) unintentional violation, or d) substantiated intentional program violation, and an incomplete review stays unresolved. In Year 1 we will verify the data and the workflow, test the outcome measure on blinded totals, and record a decision to go, pivot, or no-go. In Years 2 and 3 we will randomly assign Alabama’s CCDF providers, within strata, to the current workflow or to AL-SHIELD, the two arms of the test. Each arm’s review hours are allocated in advance. Before assignment we will draw a concealed panel of providers, and twelve months after assignment a masked follow-up will examine every panel provider the same way. We will test whether the practice also prevents new or repeat intentional violations, but only if the first-year check passes and the operational test succeeds first. In Year 4 we will finish the follow-up and hand the workflow and its monitoring to DHR. We will address three research questions: 1. Can DHR’s linked provider, child, attendance and payment records support a reliable measure of final substantiated intentional program violations, and is the AL-SHIELD workflow ready for a fair test? 2. Does the AL-SHIELD workflow produce more final substantiated intentional violations per 100 review hours allocated in advance than the current workflow (H2a)? If the first-year check permits, does it also reduce new or repeat intentional violations over the following twelve months (H2b)? And does it stay within limits set in advance on burden to cleared providers, staff time, delay, subsidy interruption, disputes and appeals, provider exit, and access in fragile local markets (H2c)? 3. What does the workflow cost in reviewer time and delay, what signal does it add beyond current practice, and can DHR run the workflow and its monitoring without us? No score alone triggers action against a provider or family, and DHR retains case authority. DHR signs each harm limit before any outcome is seen, for example on burden to cleared providers, and a breach pauses the study. We will report what we find, including inconclusive and unfavorable results, and DHR will use the results to decide whether to adopt, retune, narrow, pause, or stop the practice.
