Assessing Differences in the Psychometric Characteristics of the Latin American and Caribbean Food Security Scale in Colombia and Mexico

Abstract

The Latin American Food Security Scale (ELCSA) was applied to nationally representative samples of Colombia and Mexico (n= 13,611 and 1,560, respectively). This study explores differences in psychometric characteristics of ELCSA when applied in two different national settings. Rasch Modeling was used to assess infit and relative item severities for all questions in the scale separately in the countries. Differences in estimated item severities between the Colombian and the Mexican samples were assessed using Differential Item Functioning (DIF) models. The scale presented adequate fitness and most item infit values were within adequate range (0.8–1.2). Most items in ELCSA presented no DIF larger than 0.5 logit units, and only one item showed a DIF larger than 1.0 logit units (“socially unacceptable copying strategies”; DIF=1.22). Besides this item, households in both countries presented similar relative item severities for all other items, following similar patterns of increasing relative item severities with each item in the questionnaire. This research demonstrates that the psychometric properties of ELCSA are similar when applied in two different Latin American countries. The results of this study support the use of the proposed scale as a regional tool, but suggest also the revision of some of the items used in ELCSA for its better adaptation to diverse settings.