ABSTRACT
Generalizability theory is an alternative measurement theory to classical test theory that does not treat all measurement differences across time, rater, or situation as “error” but rather as a phenomenon of interest. G-theory examines the extent to which scores are consistent across a specific set of conditions. G-theory can be a useful way of estimating multiple aspects of reliability and validity of measures in the same model, which classical test theory cannot achieve. In G-theory, the true score is conceived of as a person's universe score—the mean of all observations for a person over all conditions in the universe—this allows us to estimate and recognize the magnitude of multiple influences on test performance (e.g., times, raters, items, groups, instruments).
The G-Theory equivalent of the classical test theory reliability coefficient of a measure is the generalizability coefficient or dependability coefficient. The generalizability and dependability coefficients evaluate the extent to which variance is due to person variance rather than variance as a function of items, occasion, raters, methods, or other factors.
