FULL-INFORMATION FACTOR-ANALYSIS FOR POLYTOMOUS ITEM RESPONSES

A full-information item factor analysis model for multidimensional pol ytomously scored item response data is developed as an extension of pr evious work by several authors. The model is expressed both in factor- analytic and item response theory parameters. Reckase's multidimension al parameters for the model also are discussed as well as the related geometry. An EM algorithm for estimation of the model parameters is pr esented and results of the analysis of item response data by a compute r program incorporating this algorithm are presented.