Publication details

Cross-Cultural Scale Comparability Using Partial-By-Item&Country Invariance Analysis

Authors

CÍGLER Hynek STANCEL-PIĄTAK Agnes CHEN Minge

Type Conference abstract
MU Faculty or unit

Faculty of Social Studies

Citation
Description To conduct cross-cultural comparisons it is crucial to establish scales that are equivalent among cultures. The construction of respective measurement instruments has proved to be challenging in the context of Large Scale Assessment (LSA), in which the number of participating countries is large. This study uses Multiple Group Confirmatory Factor Analysis (MGCFA) to compare the results from the traditional measurement invariance analysis with two partial measurement invariance approaches. The data used in this study comes from the TALIS Starting Strong Survey, which is a unique international dataset of staff and leader characteristics in Early Childhood Education (ECEC). Traditional measurement invariance approach requires the parameters of all items for all countries the same. Due to the system diversity, it might be challenging to achieve cross-cultural comparability of latent traits in the ECEC context. The traditional partial invariance (partial by item) method allows some items to vary across all countries, whereas other items are still invariant. This study adopted a more flexible strategy by allowing some items to be freely estimated for some countries to create comparable constructs, an alternative approach we named as partial-by-item&country invariance. We expect that the partial-by-item&country measurement invariance will outperform the traditional invariance analysis as well as the partial-by-item invariance analysis with respect to the model fit as well as to the number of scalar invariant scales. This approach allows enhancing the cross-country comparability in future LSA without violating statistical assumptions while simultaneously considering cultural differences.