Journal of Education and Psychology

In modem measurement theory, the Rasch model and Bimbaum's two- or three- parameter models among others together form so called item resEconse theory. Many researchers treat the Rasch model as a special case of the two-parameter model, and not thing more. Therefore, the two - and three- parameter models are highly recommended. In this paper, I compare the Rasch model with the two- and three- parameter models on the following aspects: scales in the models, meaning of the parameters, dimensionality, guessing in multiple-choice items, and model development. I concludes that only the Rasch model is a genuine measurement model, and the two- and the three- parameter models are only statistical models without measurement per se. Consequently, the Rasch model is a necessary condition for measurement. If test data do not fit the Rasch model, should we still insist on the Rasch model and modify the test data or switch to multiparameter models? This Rasch debate is still lasting. In this paper, I show how the debate can be resolved through recognition of objective measurement. Furthermore, if item difficulties can be inferred directly from test construct theory instead of from real data analyses, general objectivity is achieved, as in the natural sciences.

_{[ Chinese version | Index | Journal of Education and Psychology | Other Journals | Faculty Home Page | Enquiry ]}

Mail any comments and suggestions to en@fed.cuhk.edu.hk.