ABSTRACT

According to the Standard 3.9 of the Standards for Educational and Psychological Testing (American Educational Research Association, American Psychological Association and National Council for Measurement in Education, 1999), evidence of model fit should be provided when an item response theory (IRT) model is used to make inferences from a test data set. Researchers, such as Yen (1981) and Wainer and Thissen (1987) have suggested that an incorrect choice of an IRT model can lead to biased estimates of examinee performance, unfair rankings of examinees, and wrongly equated scores. To avoid making an incorrect choice of an IRT model and to ensure the best possible model-data fit, it is essential to apply model-fit and model-comparison methods.