Examining the cross-national applicability of multi-item, multi-dimensional measures using generalizability theory

Abstract

Establishing the applicability of multi-item measures is important for making valid inferences when testing theories cross-nationally. Typically, researchers have relied upon the tenets of classical measurement theory (CT) using confirmatory factor model invariance testing to conclude that a unidimensional measure is applicable across countries. However, two important issues remain unresolved via CT techniques: (1) if the measure is found not to be invariant, CT tells us little as to why the measure varies across countries; and (2) if the measure is multi-dimensional, what factors affect its cross-national applicability? Our research seeks to address these issues and the cross-national measurement applicability of multi-dimensional scales via generalizability theory (GT). In this paper, we use a cross-national data set and simulated data sets to demonstrate the usefulness of GT to cross-national multi-dimensional measurement.