ABSTRACT
Structural equation models are currently widely used in the behavioral, social, and educational sciences. They permit the modeling of complex multivariate phenomena, whereby measurement errors are accounted for both in the dependent and explanatory variables. Application of these models for testing behavioral and social theories has been on the increase for a considerable period of time. A large body of literature on evaluation of model fit has accumulated over the past two decades (e.g., Bollen, 1989; Browne & Cudeck, 1993). However, essentially all of it has been concerned with indexes of overall goodness-of-fit. Thereby, little attention has been paid to the assessment of model fit at the individual subject level.
