Randomized Item Testing ; its Framework and Infrastructure

Abstract

As a key branch of CBT (computer-based testing), RIT (randomized item testing) is defined and discussed. CBT is now familiar as an innovative method for measuring an amount of acquired knowledge of an examinee. Techniques for precise measurement of knowledge is getting more importance in the present society because of the increasing need for the acreditation to qualify intelligent professions.
We begin with noting that CBT contains various different aspects. An important one is measurement of an amount of knowledge. To distingush the most important aspect in CBT from others, we introduce RIT. It is noted that the two key factors of RIT are a well-prepared and well-maintained item pool and a set of items randomly sampled from the item pool. Various difficulties, which arise in traditional testing, can be mitigated by applying RIT. It is emphasized that such difficulties are associated with those in scientific epistemology. We identify the four major steps in the test construction process by RIT, and duscuss the relation of each step with the purpose of measurement.