Introduction to Latent Variable Modeling and Confirmatory Factor Analysis
In Chapter 1, we introduced the notion of a latent construct (latent variable) and discussed its relevance for behavioral, social, and educational research. The previous chapter used that concept in the context of factor analysis (FA), where latent constructs are typically referred to as ‘common factors’, ‘latent factors’, or simply ‘factors’. We also discussed there the traditional model of FA. As was made clear, the FA model represents a set of equations relating in a linear fashion observed measures with one or more common factors, in addition to several distributional assumptions. The discussion in Chapter 3 was focused on a model concerned with orthogonal factors or alternatively with correlated factors but with no particular explanatory relationships being postulated between them. The latter type of relationships can be examined using a more general modeling approach, called latent variable modeling (LVM) that will be of concern in this chapter. The general framework of LVM will be of special relevance in the remainder of this book.