A New Method Of Cheating Detection For Computerized Adaptive Testing

Computerized Adaptive Testing (CAT) is a fully integrated product of computer technology and psychological measurement theory and contemporary pedagogy. Currently, computer technology changed so fast, CAT gradually stand out from the rapid development of large-scale computerized tests, research and application of CAT attract more and more researchers attention. Along with the development of the internet and intelligent terminal equipment, instantly measurable will be possible, but face great challenge of exam ethical issues (cheating) supervision as well. The discussion on cheating supervision is more currently, but theoretical research about that is weak. Under the item response theory (IRT), some researchers put forward some diagnostic methods for cheating from the perspective of project response time. However, because of difficulties of measure response time limit its practicality, and due to the complexity of diagnosis model which affect the operability of the model. Based on IRT, explore the mathematical model of ability estimate sequence in the CAT, and reveal the mathematical characteristics of the cheating in the CAT, then proposed A New method of Cheating Detection for Computerized Adaptive Testing. Using Fisher and Kullback Leibler selection strategy, the new method is verified by simulation. Simulation results show that this method can obtain good diagnosis result of cheating under the premise of without considering the instead of examinee directly, diagnostic rate at about90%.In this paper, the new method of cheating detection for computerized adaptive testing lays the groundwork for further research of cheating detection and development of test research in the new situation, and provide theory support for scientifically oriented education.