|| Checking for direct PDF access through Ovid
Generalizability Theory (GT) provides a flexible, practical framework for examining the dependability of behavioral measurements. GT extends classical theory by (a) estimating the magnitude of multiple sources of measurement error, (b) modeling the use of a measurement for both norm-referenced and domain-referenced decisions, (c) providing reliability (generalizability) coefficients tailored to the proposed uses of the measurement, and (d) isolating major sources of error so that a costefficient measurement design can be built. Unfortunately, GT has not been readily accessible to psychological researchers, perhaps because its development and presentation have been largely technical. G theory's inaccessibility may explain why classical theory remains the preferred method for estimating reliability. The purpose of this article is to present GT and its wide applicability to a broad audience. Our intent is to demystify GT and provide a useful tool to psychological researchers and test developers.