Association for Public Policy Analysis & Management

The Food and Nutrition Service (FNS) requires states to conduct face-to-face interviews with households at certification and recertification to determine SNAP eligibility. However, over the last decade, most states applied for and received waivers that allow for telephone interviews, and a few states currently operate waivers that eliminate recertification interviews for elderly and disabled individuals. Other states have applied unsuccessfully for waivers to eliminate SNAP interviews for a broader range of individuals. Although the interviews are often useful tools for state staff and clients, some states suggest the interviews can be inefficient, in light of new technology, and lead to additional client burden. To better understand the effects of eliminating the eligibility interviews on client and worker outcomes, FNS awarded grants to Oregon and Utah to conduct demonstrations in which the eligibility interviews at certification and recertification were eliminated. Oregon implemented the demonstration in five sites and selected comparison sites for evaluation. Utah implemented the demonstration statewide and randomly assigned their caseload to a demonstration or comparison group. For both of these demonstrations, Mathematica evaluated how key outcomes, such as program access, payment accuracy, and administrative costs vary with and without an interview. The study also assessed client and staff satisfaction. To calculate program effects, the study used a comparison group design in Oregon—using a difference-in-differences approach to compare changes in demonstration counties to changes in similar comparison counties over time—and a randomized controlled trial design in Utah. Even though Oregon and Utah used different evaluation designs, we found similar patterns in the results. In general, eliminating the interview had few significant effects on client outcomes for the overall populations, but some subgroups experienced larger effects. The effects on worker and program outcomes were mixed; demonstration cases often took longer to process but had lower error rates.