System Design And Realization Of Computerized Adaptive Testing

A test is one of the main measures of the educational assessment. The traditional test, especially the large scale of traditional paper-and-pencil (P&P) test is time-consuming and energy-exhausting. Meanwhile there are some difficulties with the procedures of organization and management when the traditional tests are administered. Along with the rapid development of computer technology, the computer plays a major role in the standard test. But most computerized testing are computer-based tests (CBT), they are the electronic form of P&P tests with no adaptive function.For the last two decades, the computerized adaptive testing (CAT), based on the Item Response Theory (IRT), is rapidly developed and applied in examination field. The objective of CAT is to construct an optimal test for each examinee, Unlike P&P tests; different examinees can receive different tests of differing length and contents. CAT has several advantages over P&P tests: Shorter tests, higher measurement precision, test on demand, and immediate test scoring as well as reporting all make CAT more attractive. However, the initial costs of CAT are high, Considerable financial and human resources are needed to staff and organize a CAT program.In this thesis the system design and realization of variable- length CAT is discussed and explored. This system consists of two. modules: user administration module and adaptive testing module. In the user administration module the functions of increasing, deleting, modifying data and revising the information of user are included. In the adaptive testing module, the following functions are included, they are estimating the examinee's ability, selecting the suitable item administered sequentially, terminating the test when the predetermined target is reached, querying the results of test, managing the item pool, and so on.The strategy of selecting the item is maximum information method proposed by Lord. And the method of estimating the examinee's ability is the conditional maximum likelihood estimation. The rule of termination of the test is based on the test Fisher information.To examine the behaviors of the CAT system, Monte Carlo study is employed. The results of the Monte Carlo study show that the behaviors of the system go well.The CAT system is designed in Visual Basic and using Access Database.In this thesis some of the future possibilities of CAT are discussed.