Using Administrative Records to Assess Accuracy of Reporting in the Survey of Income and Program Participation: The Case of SNAP in Texas

Jeongsoo Kim, U.S. Census Bureau
Daniel Perez-Lopez, U.S. Census Bureau

Many studies have found that benefit receipt tends to be under-reported in population surveys. This paper examines the accuracy of reported Supplemental Nutrition Assistance Program (SNAP, formerly Food Stamps) benefits and amounts in the Census Bureau’s Survey of Income and Program Participation (SIPP). We link Texas SNAP administrative records to SIPP 2008 Panel data, Wave 1, covering the period of May through November 2008. Among SIPP respondents who can be matched to Texas administrative records, we find that about 54 percent of persons of the linked data under-reported (false negative report) SNAP receipt in SIPP. Logistic regression shows that Hispanic, never-married, high school graduate, and female head of householder populations were more likely to under-report than their counterparts. For example, Hispanics are more likely to under-report their SNAP receipt than non-Hispanics by 8 percentage points, holding other factors constant.