The Necessity of Testing Measurement Invariance in Cross-Cultural Research: Potential Bias in Cross-Cultural Comparisons With Individualism– Collectivism Self-Report Scales

Abstract

Individualism and collectivism are some of the most widely applied concepts in cultural and cross-cultural research. They are commonly applied by scholars who use arithmetic means or sum indexes of items on a scale to examine the potential similarities and differences in samples from various countries. For many reasons, cross-cultural research implicates numerous methodological and statistical pitfalls. The aim of this article is to summarize some of those pitfalls, particularly the problem of measurement non-invariance, which stems from the different understandings of questionnaire items or even different character of constructs between countries. This potential bias is reduced by latent mean comparisons performed with Multigroup Confirmatory Factor Analysis and the Measurement Invariance procedure within a Structural Equation Modeling framework. These procedures have been neglected by many researchers in the field of cross-cultural psychology, however. In this article, we compare ‘traditional’ (comparison of arithmetic means) and ‘invariant’ (latent mean comparison) approaches and provide necessary R source codes for replications of measurement invariance and latent mean comparisons within other scales. Both approaches are demonstrated with real data gathered on an Independent and Interdependent Self-Scale from 1386 participants across six countries (Slovenia, Croatia, Bosnia and Herzegovina, Serbia, Macedonia and Albania). Our results revealed considerable differences between the ‘invariant’ and ‘traditional’ approaches, especially in post-hoc analyses. Since ‘invariant’ results can be considered less biased, this finding suggests that the currently prevalent method of comparing the arithmetic means of cross-cultural scales of individualism and collectivism can potentially cause biased results.