CML based estimation of extended Rasch models with the eRm package in R

Abstract

This paper presents an open source tool for computing extended Rasch models. It is realized in R <br/>(R Development Core Team, 2006) and available as package eRm. In addition to ordinary Rasch models <br/>extended models such as linear logistic test models, (linear) rating scale models and (linear) partial <br/>credit models can be estimated. A striking feature of this package is the implementation of conditional <br/>maximum likelihood estimation techniques which relate directly to Rasch's original concept of specific <br/>objectivity. The mathematical and epistemological benefits of this estimation method are discussed. <br/>Moreover, the capabilities of the eRm routine with respect to structural item response designs are demonstrated.