Extended Alignment Method for Multigroup Measurement Invariance: Eight PISA (Programme for International Student Assessment) Constructs Across 30 Organisation for Economic Co-operation and Development Countries

• In Event: Modeling Latent Variable Interactions and Multiple-Group Models

Abstract

We demonstrate the flexibility of the alignment method for multigroup data, applied to the eight PISA2006 scales (30 OECD countries, N = 249,840). Traditional multiple-group CFA scalar models of item intercept invariance were not supported, but the alignment model fit the data well. We introduce a new alignment-within-CFA (AwC) approach enabling analyses not otherwise possible with the alignment method: testing invariance of uniquenesses, factor variances/covariances; multiple-group MIMIC models; contrasts on latent means; gender differences over 30 countries based on a 30-group MIMIC AwC (i.e., a SEM with gender as a covariate) and a 60-group AwC CFA (i.e., 30 countries x 2 gender); and structural equation models more generally. The AwC approach provides new flexibility to address substantively important issues.

Authors

• Herbert W. Marsh, Australian Catholic University

• Jiesi Guo, Australian Catholic University

• Tihomir Asparouhov, Mplus

• Bengt Muthen, University of California - Los Angeles