The Case for Using the General Linear Model as a Unifying Conceptual Framework for Teaching Statistics and Psychometric Theory

Abstract

The present paper argues for teaching statistics and psychometric theory using the GLM as a unifying conceptual framework. This helps students understand what analyses have in common, and also provides a firm grounding for understanding that more general cases of the GLM (canonical correlation analysis and SEM) can be interpreted with the same rubric used throughout the GLM. And this approach also helps students better understand analyses that are not part of the GLM, such as predictive discriminant analysis (PDA). The approach helps students understand that all GLM analyses (a) are correlational, and thus are all susceptible to sampling error, (b) can yield r 2 -type effect sizes, and (c) use weights applied to measured variables to estimate the latent variables really of primary interest.