ABSTRACT
This chapter proposes to examine the performance of measures of model fit for the latent change score model in small samples with missing data. Structural equation modeling is a flexible modeling framework that encompasses a broad set of statistical models. The empirical example demonstrated that the framework within which a model is fit can lead to opposing results even when comparing models using the likelihood ratio test. The missing data correction used should be of the form such that it, in effect, deactivates itself when data are complete. Otherwise, it should adequately characterize the quality of the observed information in the data. Such a small sample corrections would be of great use to researchers with small samples wishing to evaluate how well longitudinal models fit their data. The chapter concludes with a discussion of the results accompanied by some thoughts on how future research should proceed.
