A Tutorial on Multilevel Generalized Structured Component Analysis Using GSCA Pro
Hierarchically structured data are common in behavioral and social science research, yet ignoring nested structures can lead to misleading statistical inferences and obscure meaningful variation in measurement and structural relationships across groups. Multilevel generalized structured component analysis (GSCA) extends component-based structural equation modeling to hierarchical data by allowing measurement loadings and structural path coefficients to vary across higher-level units while retaining common component weights. This tutorial introduces two-level GSCA and demonstrates its practical implementation using GSCA Pro, free software with a graphical interface. Data from 5,014 customers across four corporate chains provide a worked example. We guide readers through model specification, estimation, fit assessment, and interpretation of average and group-specific measurement and structural parameters, highlighting between-group heterogeneity.
Authors
- Heungsun Hwang (ORCID: https://orcid.org/0000-0002-5057-7479)
- Gyeongcheol Cho (ORCID: https://orcid.org/0000-0002-9237-0388)
- José Luis Roldán (ORCID: https://orcid.org/0000-0003-4053-7526)
- Sunmee Kim (ORCID: https://orcid.org/0000-0003-3723-9974)
- Silvia Martelo-Landroguez (ORCID: https://orcid.org/0000-0002-4447-1088)
- Gabriel Cepeda‐Carrión (ORCID: https://orcid.org/0000-0003-3839-2006)
Publication Details
- Journal
- PsyArXiv (OSF Preprints)
- Published
- 2026-11-13
- DOI
- https://doi.org/10.31234/osf.io/fj3p8
- Primary Topic
- Psychometric Methodologies and Testing
- Type
- preprint