Improving the Use of Subscores on a Test Battery: Some Reliability and Validity Evidence from the Wechsler Intelligence Scale for Children- Fourth Edition

Abstract

It is widely believed that subscores can give us more information about an examinee. Thus they can be useful in planning instructional and remedial efforts, or making vocational or academic placement decisions. However, past research has shown that subscores are often not as useful as hoped either because they do not have high reliability or because they seem to add little information. This dissertation investigated if there is any evidence of reliability and validity of the subscores from the Wechsler Intelligence Scale for Children-Fourth Edition (WISC-IV). The dissertation is composed of three separate studies using three empirical data sets. In the first study, we investigated the reliability issue of subscore differences and concluded that difference scores can be reliable. In the second study, we proposed two graphical methods to help with latent variable interpretation and the construct validation of subscores that are based on factor models. One way to use subscores is to do profile analysis. In the third study, using subscores from a special population and a profile analysis technique, we found profile patterns are useful in differentiating cases in various diagnostic categories and profile patterns actually outperformed profile levels.