Measurement Invariance in cross-cultural and comparative research: Controversies and new procedures
In the last years more and more researchers have tested the cross-national equivalence of measures (metric and scalar invariance) in different large international data-sets like the ESS, EVS, ISSP, PISA, PIACC and WVS using cross-sections or repeated cross-sections. They were applied mostly to individual data in those surveys measuring individual attitudes or values(Davidov et al. 2014, Ciechuch/Davidov/Schmidt 2016). Now discussions have started whether one really needs these testing procedures also, when using aggregated data sets and multilevel models(Welzel/Ingelhart 2015). In the first part of this workshop the arguments in this controversy are discussed and compared. Furthermore preliminary conclusions are inferred. In the second part I describe some new techniques like alignment(Aspourov/Muthen 2016, Marsh et al. 2017, Munck et al. 2017 and Bayes estimation(Aspourov/Muthen 2014, van der Schoot et al. 2015, Zercher et al. 2015) for testing measurement invariance, which are more flexible than the classical procedure and more suited in the case of many groups. These new procedures are applied by using different cross-national data sets dealing with values, attitudes toward immigration, nationalism and patriotism.
Jan Cieciuch, Eldad Davidov, Peter Schmidt & Ren Algesheimer. 2016. “Assessment of Cross-Cultural Comparability” In: Christof Wolf, Dominique Joye, Tom W Smith & Yang-chih Fu: The SAGE Handbook of Survey Methodology.