A Comparison of Classification Methods for Use in Predicting School-Based Outcomes

There is growing evidence that current classification methods are not consistent with current empirical knowledge of childhood psychopathology and the optimal way to classify school-age children remains controversial. The current study investigated three classification methods (categorical, dimensional, person-oriented) for use in predicting school-based outcomes. Children (grades 1-5; N=558) were administered the Behavior Assessment System for Children – Teacher Rating Scale and results were used to form three classification systems. Educational outcome variables were collected seven months later and the predictive validity of the three classification systems was compared using regression analyses. Findings indicated that all three methods for predicting educational outcomes were modest and were best able to predict later grade point averages. Results indicate the relative superiority of person-oriented and dimensional methods of classification; however these classification methods warrant further investigation.