Testing Generalized Rasch Models

Abstract

Item response theory (IRT) models provide a useful and well-founded framework for measurement in the social sciences. The family of IRT models is still expanding (see, for instance, De Boeck & Wilson, 2004; Skrondal & Rabe-Hesketh, 2004), so characterization of the family of IRT models is not easy. But to provide some demarcation, IRT models can be defined as stochastic models for multiway data, usually two-way data consisting of responses of persons to items. An essential feature in this definition of IRT models is parameter separation, that is, the influences of the various factors, say items and persons, on the responses are modeled by distinct sets of parameters.