Research On Item Response Theory-Based System Of Computerized Adaptive Testing

As the development of Computer-aided instruction software, system of Computerized adaptive Testing based on Item-bank has been paid more attention. The latest achievement of testing system theory research is Item Response Theory(IRT). The proposed application is just a Item Response Theory-based system of Computerized adaptive Testing(CAT).In order to improve the convenience and reliability of our system, especially to avoid the problem that aberrant response pattern will result in failing to find out maximum likelihood, the system proposed estimates the distribution parameters and maximum likelihood in two steps.The first problem that the system faces: as IRT theory model is imposed severer basic assumptions, so large groups of samples are needed to build the Item bank. To build the Item bank, our system utilizes teaching resources of universities by the auxiliary software. The approach avoids the most cost of construction on Item bank.The second problem that proposed system has been dealt with is the calculational errors when a large number of items are used. While estimating the maximum likelihood of examinees, we fix the length of calculation. The method not only has improved systematic operational efficiency but also has controlled the possibility of mistakes caused by aberrant response pattern.We characterize relationship between abilities (or other hypothesized traits) and item responses by three-parameter logistic model or two-parameter logistic model. The application shows that the proposed system is good at operation and facility. In this article, we analyze the CAT both in theory and practice.