Given a set of items, a Multidimensional Computer Adaptive Test (MCAT) selects those items from the bank according to the estimated abilities of the student, resulting in an individualized test. MCATs seek to maximize the test's accuracy, based on multiple simultaneous examination abilities (unlike a Computer Adaptive Test - CAT - which evaluates a single ability) using the sequence of items previously answered. Although MCATs have been very well studied from a statistical point of view, there is no computational system that covers all the steps needed for its appropriated use such as: the use of a calibrated item bank, proposal of initial and stopping criteria for the test, criteria for estimating the ability of the examinee and criteria to select items. The purpose of this paper is twofold: (i) to present an innovative architecture of an MCAT for real users, as a Web application, and (ii) to discuss the theoretical and methodological development of such MCAT, through a new approach named here Computer-based Multidimensional Adaptive Testing (CBMAT). The proof of concept of CBMAT was an implementation called Multidimensional Adaptive Test System for Educational Purposes (MADEPT). In simulations, MADEPT proved to be a computer system suitable for applications with real users, secure, accurate and portable.
Tags: computer-based testing, multidimensional computer adaptive test, multidimensional item response theory, web architeture.
ACM: Jean Piton-Gonçalves and Sandra Maria Aluísio. 2012. An architecture for multidimensional computer adaptive test with educational purposes. In Proceedings of the 18th Brazilian symposium on Multimedia and the web (WebMedia '12). ACM, New York, NY, USA, 17-24. DOI=10.1145/2382636.2382644 http://doi.acm.org/10.1145/2382636.2382644