Project
New structural equation modeling methods to find and account for measurement non-invariances

Our research focuses on the challenges of comparing latent variables across many groups in social science. Accurate comparison requires measurement invariance (MI), ensuring that constructs are measured consistently across groups. To assess MI, researchers often use factor analysis. However, when dealing with many groups, MI often does not hold, and pairwise comparisons to assess the sources of non-invariance become less feasible due to their exponential increase in number. This prompts the use of mixture multigroup factor analysis (MMG-FA) to cluster groups based on measurement parameters. This study examines the performance of MMG-FA under conditions of ordinal data and non-normality, using simulations to explore model limitations and potential improvements. The findings aim to refine and improve MMG-FA for handling complex, real-world data.

Supervisors
Prof. Dr. Kim De Roover
Prof. Dr. Eva Ceulemans
Prof. Dr. Francis Tuerlinckx

Financed by
KU Leuven

Period
18 September 2023 – 18 September 2029