1. When and how to use set-exploratory structural equation modelling to test structural models: A tutorial using the R package lavaan.
- Author
-
Marsh H and Alamer A
- Subjects
- Humans, Factor Analysis, Statistical, Models, Statistical, Data Interpretation, Statistical, Models, Structural, Latent Class Analysis, Software, Psychometrics methods, Computer Simulation
- Abstract
Exploratory structural equation modelling (ESEM) is an alternative to the well-known method of confirmatory factor analysis (CFA). ESEM is mainly used to assess the quality of measurement models of common factors but can be efficiently extended to test structural models. However, ESEM may not be the best option in some model specifications, especially when structural models are involved, because the full flexibility of ESEM could result in technical difficulties in model estimation. Thus, set-ESEM was developed to accommodate the balance between full-ESEM and CFA. In the present paper, we show examples where set-ESEM should be used rather than full-ESEM. Rather than relying on a simulation study, we provide two applied examples using real data that are included in the OSF repository. Additionally, we provide the code needed to run set-ESEM in the free R package lavaan to make the paper practical. Set-ESEM structural models outperform their CFA-based counterparts in terms of goodness of fit and realistic factor correlation, and hence path coefficients in the two empirical examples. In several instances, effects that were non-significant (i.e., attenuated) in the CFA-based structural model become larger and significant in the set-ESEM structural model, suggesting that set-ESEM models may generate more accurate model parameters and, hence, lower Type II error rate., (© 2024 The Authors. British Journal of Mathematical and Statistical Psychology published by John Wiley & Sons Ltd on behalf of British Psychological Society.)
- Published
- 2024
- Full Text
- View/download PDF