ABSTRACT
Although many psychometricians (e.g., Bartke, 1966; Burt, 1936; Ebel, 1951; Haggard, 1958; Hoyt, 1941; Jackson & Ferguson, 1941; Lindquist, 1953) had paved the way to the Generalizability Theory of Measurement, it was formally introduced by Cronbach, Gleser, Nanda, and Rajaratnam (1972). The generalizability theory is different from the classical theory in a number of important respects. Perhaps the most important difference is that of perspective. Specifically, within generalizability theory, reliability is considered within the context of a testing situation. Recall that in classical theory, measurement error is described through a rather vague concept called random error without any specific context. Within generalizability theory, however, measurements are viewed within a specific context of measurement situation.
