Using Latent Regression Rasch Model to Evaluate Surveys

• In Event: Advanced Psychometric Analyses in Rasch Modeling

Abstract

Explanatory item response models provide a framework for including both person characteristics and item design factors in the analyses of survey data. This study applies latent regression Rasch models (LRRM) to the household food security survey module (HFSSM) to explore the influence of person characteristics on item responses. Our results suggest that women are more prone to report experiencing food insecurity compared to men and that households receiving SNAP (Supplemental Nutrition Assistance Program) were more inclined to report being food insecure in comparison to households that did not receive SNAP. Our approach can be applied to other surveys with the goal of providing a better understanding of the relationships between latent traits, item responses, and external variables.

Authors

• Jing Li, University of Georgia

• George Engelhard, University of Georgia