Files
Abstract
Food insecurity is a condition characterized by the lack of access to the food necessary for a healthy and active lifestyle. The U.S. Department of Agriculture (USDA) measures food insecurity in the United States annually with the Household Food Security Survey Module (HFSSM), an 18-item scale that references food hardships among adults and children in a household. The HFSSM is administered as a supplement to the Current Population Survey (CPS; Coleman-Jensen, Rabbitt, Gregory, & Singh, 2019). In this research, measurement is viewed by the process by which households and items are located on a line representing the latent construct of interest, household food insecurity. The five requirements for invariant measurement are: (1) item-invariant measurement of persons; (2) non-crossing person response functions; (3) person-invariant calibration of HFSSM items; (4) non-crossing item response functions; and (5) unidimensionality of the HFSSM (Engelhard, 2013). The Rasch model is an ideal-type item response theory (IRT) model that meets these requirements. It expresses the probability of endorsing an item as the function of a household’s latent food insecurity and the difficulty of the item (Bond & Fox, 2015). The Rasch model was also used to calibrate the HFSSM, and the scale has been maintained over time with this model (Coleman-Jensen et al., 2019; Engelhard, Engelhard, & Rabbitt, 2017). The purpose of this dissertation is to the use properties of invariant measurement to evaluate the psychometric properties of the HFSSM. The research consists of three studies that focus on household measurement, item calibration, and dimensionality.