Teoría G: un futuro paradigma para el análisis de pruebas psicométricas

Abstract

Generalizability Theory (G Theory) allows to measure the reliability of a test by means of the quantification of the importance of each one of its sources of variability. G Theory redefines the error as conditions or facets of measurement, using the Generalizability coefficient as an indicator to estimate the reliability. It is important to point out that this approach does not contradict the fundamental assumptions of Classical Test Theory. On the contrary, it can be seen as an extension of the latter. It is concluded that, even though in many cases the psychometric instruments are used to make relative decisions (norm referenced model), being Classical Test Theory sufficient for such situations; other instances, like those which involve the use of educational tests, often require decisions based on absolute standards of performance, where G Theory constitutes a very useful tool, much more informative than the classical approach.