An Examination of Measurement Invariance Using a Multilevel Explanatory Rasch Model

Abstract

In this study, we proposed a multilevel explanatory Rasch model for examining measurement invariance of international food-insecurity measures on the Food Insecurity Experience (FIE) Scale from different countries. The United Nations Food and Agriculture Organization developed the FIE Scale to quantitatively measure the severity of food insecurity. We discussed the advantages of using a multilevel explanatory Rasch model to detect DIF when data are nested. Our results indicated the presence of differential item functioning for gender on the pooled (global) FIE Scale. Several person-level explanatory variables (educational attainment and socioeconomic status) also explained a significant amount of the variation in food-insecurity measures. Furthermore, separate analyses were conducted for six geographic regions of the world, showing gender-related DIF, as well as the impact of person-level explanatory variables across the geographic regions.