Test Objective

• In January through March of 2006, the American Community Survey (ACS) conducted the first test of new and modified content since the ACS reached full implementation levels of data collection. The results of that testing will determine the content for the 2008 ACS.
• An analysis of the food stamp benefit data for the state of Maryland using the national 2001 supplemental survey (SS001) found the survey state-level survey estimates were a good deal lower than the count of households that participated in the Maryland food stamp program (2004,Taeuber, Resnick, Love, et al). This report concluded that the observed discrepancy in recipiency was due to respondents misreporting that no one in the household received food stamp benefits during the survey’s reference period when, in fact, Maryland issued benefits to someone in the household.
• In order to decrease item non-response and get at comparable estimates of food stamp receipt as those available from the mentioned benchmarks, the wording of the food stamp question was changed to more closely reflect the wording used in the Current Population Survey, Annual Social and Economic Supplement (CPS ASEC). In addition, since there is no legislative mandate that requires collecting the dollar amount of food stamp benefits received, this part of the question was dropped.

Methodology

• The content test compared two versions of the food stamp question set. The control version replicated the current ACS question. The test group changed the wording of the food stamp question to more closely reflect the wording asked in the CPS ASEC. The test question was changed to include the additional phrase at the end, “or a food stamp benefit card” and the question on the dollar value of food stamps for the prior 12 months was removed.

Research Questions and Results

• The following research questions were evaluated from the collected data on the control and test versions of the food stamp question. 1). Does adding the term “food stamps benefit card” and not asking for the total value of the food stamps received in the past 12 months reduce the under reporting of the receipt of food stamps? Comparing the control and treatment groups, we observe that changing the wording of the food stamp question, as well as not asking about the value of the amount of food stamps received, significantly increased the proportion of households that reported receiving food stamps, nationally and in the high and low response areas (see Table 2). Note that the size of the increase was consistent across the response strata. iv 2). Do the changes to the food stamp question affect the item nonresponse rate? Comparing the control and treatment groups, we observe that changing the wording of the food stamp question, as well as not asking about the total value of the food stamps received in the past 12 months, resulted in a reduced or equal item nonresponse (see Table 3). Nationally and for the high response areas, the test version reduced the item nonresponse rate. For low response areas, the test version maintained the item nonresponse rate.
• Based on the empirical findings from the study, we conclude that the test version provides a better measure of food stamp recipiency.