This chapter focuses on the evaluation models. Models that are designed to increase the understanding of psychological phenomena needed to reflect the conditional and nonlinear relations that are certainly operating among psychological variables. Models serving the more pragmatic goal of efficiently predicting important criteria are always handicapped by the inclusion of complex relations. The development of simple models that rely on higher order variables automatically includes the complexity that exists. The most basic model for predicting one variable from another is the simple linear relation model. This model is an extreme simplification of nearly all psychological phenomena. Multivariate extension model is the mainstay of statistical prediction for most of the history of personality assessment. Specifically, this model is limited to a linear combination of the predictor variables. Psychological phenomena are almost certainly caused by a more complicated, nonlinear function of a set of variables. Interactive models have contributed to their power to predict human behavior. These models have not fared well in applied settings. There is little doubt that multivariate and interactive models are more faithful models of human behavior.
