The paper focuses on techniques for detecting unit segments in component-based SEM byresponse-based techniques in the case of unknown (latent) discrete moderating factors,i.e. when both the number and the structure of the classes are not a priori known. The first part of the work briefly introduces both PLS Path Modeling and Generalized Structured Component Analysis (GSCA). Further, the paper presents ways to handle unobserved heterogeneity in component-based approaches to SEM. In this framework, the methods allowing response-based clustering in PLS Path Modeling framework are discussed, and then Fuzzy Clusterwise Generalized Structured Component Analysis of Hwang et al. (2007) is investigated as a technique for response-based clustering in GSCA.
ESPOSITO VINZI, V. and TRINCHERA, L. (2008). Latent Class Detection in Component-Based Structural Equation Modeling. In: XLIV Scientific Meeting of the Italian Statistical Society (SIS). Proceedings of the XLIV SIS Scientific Meeting. CLEUP, pp. 147-154.
Mots clés : #Unobserved-Heterogeneity