Noninvariant Measurement Versus Traditional Approaches for Studying Cultural Differences: A Case of Service Expectations

Abstract

Current cross-cultural comparisons in hospitality and tourism research are largely based on differences in simple average scores. This method assumes equal weights for the items and hence fails to address cross-cultural differences in the relative importance of each attribute in comprising the overall evaluation. In this study, a comparison between United States customers and Hong Kong customers in restaurant service expectations is used to demonstrate that latent variable scores are more appropriate than simple average scores for cross-cultural comparisons. The two samples demonstrate differences in relative importance for some items measuring restaurant service expectations. By calculating the latent variable score both with and without factorial invariance, researchers can decompose cross-cultural differences into a component due to item responses, and a second component due to unequal item-construct relationships. It is demonstrated that meaningful cross-cultural comparisons are possible only when noninvariant items are carefully examined.