ABSTRACT
Multidimensional item response theory (MIRT; Reckase, 2009), and in particular, full-information item factor analysis (Bock et al., 1988) has become increasingly important in educational and psychological measurement research (see a recent survey by Wirth and Edwards, 2007). In contrast to unidimensional item response theory (IRT) modeling, an item can potentially load on one or more latent dimensions (common factors) in an item factor analysis model. These common factors are also potentially correlated, in the tradition of Thurston's (1947) multiple factor analysis.
