Analysis of ACS Food Stamp Program Participation Underestimate

• Activity Number: 197
• Type: Contributed
• Date/Time: Tuesday, August 10, 2004 : 8:30 AM to 10:20 AM
• Sponsor: Section on Government Statistics
• Author(s): Dean Resnick*+ and Susan Love and Cynthia Taeuber and Jane M. Stavely
• Companies: U.S. Census Bureau and U.S. Census Bureau and University of Baltimore/Jacob France Center and University of Baltimore/Jacob France Center
• Keywords: American Community Survey ; food stamps ; survey methodology ; ACS

Abstract

The U.S. Census Bureau's American Community Survey (ACS) has a question that asks households whether they have received food stamps in the previous year. Based on responses to this question, the Census Bureau estimates the number of households who are participating in this program. However, when these estimates are compared to family participation counts generated from food stamp administrative records, they are found to be quite a good deal lower. To understand the limitations of Census surveys for estimating program participation, we investigated the sources of this discrepancy. We matched ACS records to Maryland food stamp records at the individual level, and then used these individual level match results to generate family match results. Analysis of these family match results suggests that the largest factor causing this discrepancy is families reporting that they do not receive food stamps when in fact they do. This misreporting seems to be largely of function of who in the household received the food stamps, for how long they were received, and the interval since they were last received. Family definition and survey coverage also contributed to the underestimation.