A Blueprint for Measuring Complex Concepts in a Cross-National Framework: Stigma of Mental Illness as a Case

A central problem in cross-national survey research is to develop measures of complex concepts that are comparable across a wide-range of cultural contexts. It is widely agreed upon that meaningful and valid cross-national comparison depends on the equivalence of measures across the countries under investigation. The questions that remain, however, regard the type and degree of equivalence that is necessary, under what conditions it is necessary, and the most appropriate method for identifying and evaluating said equivalence. In this paper I use data from the Stigma in Global Context Mental Health Study (SGC-MHS) – a theoretically and methodologically coordinated empirical examination of the stigma of mental illness in 17 countries – to address these questions. Specifically, I use 25 items developed to tap six variants of stigma as a framework to discuss types (e.g., conceptual, functional, measurement), levels (e.g. configural, metric, scalar), and degrees of equivalence (e.g. partial or full); under what conditions each is necessary (e.g., is the goal to compare mean levels across groups or the pattern of correlations across groups?); methods of identifying and evaluating equivalence (e.g. item response theory, multiple group confirmatory factor analysis), and what to do when tests of equivalence fail (e.g. delete problematic items, utilize partial invariance, interpret invariance as cross-national data in their own right). I end by presenting a blueprint of the necessary steps that should be taken before meaningful and valid cross-national comparisons can be made.