Computerized Adaptive Testing Using MIRT
Computerized adaptive testing (CAT) is a methodology for constructing a test, administering it to an examinee, and scoring the test using interactive computer technology. This methodology has a history that is as long as that of interactive computing. An early summary of CAT methods is given in Weiss (1974). A detailed description of the development of an operational application for the Armed Services Vocational Aptitude Battery is given in (Sands 1997). There are also several books available that describe the basic components of CAT procedures (Wainer, Dorans, Flaugher, Green, Mislevy, Steinberg and Thissen 1990; Parshall, Spray and Davey 2002; van der Linden and Glas 2000) so the basic details of the methodology are not presented here. A review of that literature will show that most of the current CAT methodology is based on the assumption that a unidimensional IRT model accurately represents the interaction between persons and test items. In this chapter, the generalization of the CAT methodology to the multidimensional case is considered. To provide a framework for this material, a brief summary of the conceptual basis for CAT is provided.
KeywordsTest Item Item Parameter Fisher Information Matrix Item Bank Item Pool
- Lehmann EL, Casella G (1998) Theory of point estimation (2nd edition). Springer-Verlag, New YorkGoogle Scholar
- van der Linden WJ, Glas CAW (eds) (2000) Computerized adaptive testing: theory and practice. Kluwer Academic, Dordrecht, The NetherlandsGoogle Scholar
- Weiss DJ (1974) Strategies of adaptive ability measurement (Research Report 74-5). Psychometric Methods Program, University of Minnesota, MinneapolisGoogle Scholar