ABSTRACT
The extension of the Rasch model for simple dichotomous data into the rating scale model for polytomous data (Andersen, 1977; Andrich, 1978) has had an enormous impact on Rasch modeling. Although Rasch’s original work had conceptualized all the subsequently developed members of the Rasch family of models, thinking up the model and developing the estimation procedures are two related but different tasks. The rating scale model is now used routinely for the analysis of Likert-type data (Chapter 6), although Andrich had originally intended the RSM for another purpose: the evaluation of written essays. It would assist the examination process by producing measures from values applied to qualitative essay rating scales by examiners. However, it also paved the way for all the Rasch procedures in common use that involve data with more than just two values (0, 1).
