Abstract
Measurement theory and practice have changed considerably in the last 25 years. For many assessment specialists to-day, item response theory (IRT) has replaced classical measurement theory as a framework for test development, scale construction, score reporting, and test evaluation. The most popular of the item response models for multiple-choice tests are the one-parameter (i.e., the Rasch) and three-parameter models. Some researchers have been quite adamant about using only the one-parameter model and have been rather critical of applications of multi-parameter models such as the three-parameter model. In this paper, nine arguments are offered for continuing research and applying multi-parameter IRT models. Also, the position is taken that both single and multi-parameter IRT models (and many others) have potentially important roles to play in the advancement of measurement practice and judgments about which models to use in particular situations should depend on model fit to the test data.