ABSTRACT
This chapter argues in favor of using indices of model fit at the subject level. The concern is with providing further evidence that each residual can particularly beneficial for ascertaining model misspecifications. In particular, these residuals used as more indices of fit in exemplifying that lack of latent relationship in a structural equation model does not imply lack of causal relation between its unobservable variables. The chapter aim is to prove utility of latent individual residuals as informative adjuncts to routinely used goodness of fit measures, hoping to stimulate thereby further interest, research into, and application of subject level residuals in social research employing the popular SEM methodology. The aim of this chapter was to present further evidence that subject level residuals, and in particular latent individual residuals (LIR), can very useful indicators of misspecifications in structural equation models. This chapter capitalized on an important analogy in the general context of examination for model misspecifications.
