Association for Public Policy Analysis & Management

This paper investigates the framing effects of SNAP receipt on the responses to the food security module (FSM) questions. While previous research has documented the effect of the cyclical effects of SNAP receipt on diet outcomes — calories, diet quality, and eating occasions, among others — this paper looks at how the receipt of SNAP cognitively frames responses to 12-month FSM questions in the National Health and Nutrition Examination Survey (NHANES). Some recent research has shown that respondents are more likely to respond affirmatively to the less severe 30-day FSM questions at the end and the beginning of the SNAP month; these households are more likely to be classified as food insecure during this part of the SNAP month. We show that SNAP households are more likely to respond affirmatively to the more severe 12-month FSM questions during the first and last week of the SNAP month; households are more likely to be classified as having very low food security during this time. We show that this anomaly produces bias in the estimated bounds of the effect of SNAP on very low food security, and show that it also likely implies that estimates of very low food insecurity prevalence that do not take this response pattern into account are lower bounds. In particular, the downward bias is about 5 percentage points for the SNAP sample (28 percent of prevalence) and .9 percent for the entire population (about 11 percent of prevalence). In short, we show that, absent some accounting for cyclical response behavior in FSM questions, we are underestimating both very low food security and the effect of SNAP on very low food security.