Computer Adaptive Testing of Student Knowledge.docVIP

Acta Polytechnica Hungarica Vol. 7, No. 4, 2010 Computer Adaptive Testing of Student Knowledge Sanja Maravi? ?isar , Dragica Radosav , Branko Markoski2, 1 2 Robert Pinter , Petar ?isar3 1 1 2 3 Subotica Tech, Department of Informatics Marka Ore?kovi?a 16, 24000 Subotica, Serbia sanjam@vts.su.ac.rs, probi@vts.su.ac.rs Technical Faculty “Mihajlo Pupin”, Department of Informatics ?ure ?akovi?a bb, 23000 Zrenjanin, Serbia radosav@zpupin.tf.zr.ac.rs, markoni@uns.ac.rs Telekom Srbija, Subotica Prvomajska 2-4, 24000 Subotica, Serbia petarc@telekom.rs Abstract: Technological progress, responsible for the declining costs of computers, coupled with the advancement of computer adaptive software have promoted computer adaptive testing (CAT) in higher education, offering alternatives to the conventional paper and pencil examinations. The CAT testing process, statistically conducted through Item Response Theory, is able to react to the individual examinee, keeping examinees on target with test items of an appropriate level of difficulty. The basic goal of adaptive computer tests is to ensure the examinee is supplied questions that are challenging enough for them but not too difficult, which would lead to frustration and confusion. The paper presents a CAT system realized in MATLAB along with its development steps. The application can run from a Matlab command window, or it is possible to make a stand-alone application that does not require the installation of Matlab. The questions are written in a .txt file. This allows the examiner to easily modify and extend the question database, without specific knowledge of the syntax of any programming language. The only requirement is for the examiner (but it is only required) to follow a pre-determined format of question writing. The program enables the testing of student knowledge in C++. Keywords: computer adaptive testing; Item Response Theory; e-