Computerized adaptive testing in computer science

Abstract

research on Intelligent Tutoring Systems (ITSS) has not presented a substantial student model which can be generalized to all ITSS. The purpose of this study was to design, tes~and implement a quantitative student model (S yang, 1992) called the Angel Model to measure programming abilities of undergraduate students who have taken the first Computer Science course. This model (Angel Model) met the criteria for using an Item Response Theory (IRT) model; therefore, the data can be used for developing an IRT based Computerized Adaptive Testing (CAT) system to measure students' programming abilities at the completion of a CS 1 course. An important implication of this study is that when students' understanding of a domain (e.g., Computer Science, Chemistry, and Physics) can be built quantitatively, a computerized adaptive test can then be developed effectively to measure students' abilities in that domain. Hence, the quantitative student model can provide an Intelligent Tutoring System with knowledge about students' abilities in that domain.