Are Cross-Cultural Shyness Comparisons Valid? Testing Invariance With Multigroup CFA and the Alignment Method Across Eastern and Western Cultures

Public Significance Statement American defined constructs assessed through questionnaires are assumed to apply across all cultural groups. Constructs are rarely tested for equivalence in measurement structure across groups. This study found that the same underlying shyness construct was being measured across Canadian and Chinese participants, allowing us to look at meaningful and unbiased mean differences in shyness. Chinese young adults had significantly higher mean levels of shyness than Canadian young adults. Temperament/personality research has been largely based on an imposed-etic strategy: Self-report temperament/personality inventories created in and reflecting American culture were tacitly assumed to apply in other countries and cultures. Correspondingly, invariance tests on the underlying phenomena and their associated constructs have been rarely applied in cross-cultural studies and in those instances unsuccessfully. To explore this oversight, we investigated whether a western-based anxious shyness measure was equivalent across eastern-western cultures (Chinese, n = 816, 47.2% male; Canadian, n = 995, 30.8% male). In the first instance, we used the well-established multigroup confirmatory factor analysis (MGCFA) to show that there was measurement noninvariance in the one factor shyness scale across the two countries and the two countries by sex. We further examined the issue of noninvariance using the newer alignment method, an approach providing detailed information on noninvariance for each country model by parameter (across intercepts and loadings) as an alternative to the MGCFA restrictive assessment of whole scale construct validation. The findings suggested acceptable approximate invariance in the shyness scale to support an unbiased comparison of mean levels between the two countries and the two countries by sex. Chinese young adults had significantly higher mean levels of shyness than Canadian young adults. Despite some limited noninvariance, we were able to conclude that the underlying construct of shyness as measured in this study was equivalent across Chinese and Canadian cultures. Findings illustrated the difficulties and importance of first establishing fundamental measurement properties and equivalence in personality constructs before inferring cross-cultural universality in complex traits and characteristics.