ABSTRACT
Generalized structured component analysis is a component-based approach to structural equation modeling. That is, it denes latent variables as weighted composites or components of indicators, thereby permitting the provision of unique latent variable scores. At the same time, however, this characteristic may lead to a potential issue that the quality and interpretability of latent variables depend directly on which indicators are used. For example, it may be difcult to interpret a latent variable clearly when it is constructed from a large number of indicators. In addition, possessing irrelevant or uninformative indicators is likely to obscure or distort the meaning of a latent variable. Furthermore, the use of too many indicators for a single latent variable tends to increase the likelihood of having redundant indicators, thus resulting in less stable parameter estimates, as discussed in Chapter 8. Accordingly, it is important to select an appropriate set of indicators for each latent variable in the measurement model of generalized structured component analysis.
