Association for Public Policy Analysis & Management

This paper describes and examines an alternative food security measure to that currently used in official statistics published by the Economic Research Service. A review of USDA’s food security measurement methods by the National Academies Committee on National Statistics found that there are some biases in the measure that should be addressed to make estimates of food insecurity between households with and without children more comparable. The motivation for developing this new measure was to address these potential biases with a straightforward approach. In presenting this new measure, Nord and Coleman Jensen (2016) argue that it is to be preferred on statistical and measurement-theoretical grounds. In this paper, we examine whether the approach is also to be preferred based on empirical comparisons with alternative measures of food adequacy: in particular, we focus on food insufficiency, needing to spend more to meet food needs, using a food pantry, dietary quality and health status.

The proposed experimental food security measure would address the weakness in the current measure by assessing food security among adults and children separately. This new cross-classification method (experimental measure) utilizes the same survey items, but instead of combining adult and child items to determine household food security status, the experimental measure determines the food security status of adults and of children separately.  If there is either food insecurity among adults or food insecurity among children, the household is be classified as food insecure. With this method, food insecurity among adults is statistically comparable between households with and without children. The experimental food security measure is simply an alternative way of summing responses to the food security scale for households with children, still relies on raw score, and does not require changes in data collection.

Data for the analysis come from two sources. The Current Population Survey Food Security Supplement (CPS-FSS) is the data source for USDA food security statistics and data from the years 2009-2013 are used in this analysis. We examine cross-tabulations of the current/experimental food security measures and alternative measures of food adequacy. We use these frequencies to examine whether the experimental or current food security measure is more consistent with alternative measures of food adequacy. The second data source is the National Health and Nutrition Examination Survey (NHANES). We use these data to examine the associations between the food security measures and dietary outcomes and self-reported health status. We conduct t-tests to determine whether differences are statistically significant. Preliminary findings suggest that although the experimental measure is to be preferred on statistical and theoretical grounds, the current method appears to be more closely associated with alternative indicators of food need and dietary quality. Additional multivariate models will examine how demographic characteristics relate to the probability of food insecurity with each measure and examine the characteristics of those households that differ on the two measures.