Research On Web-based Computorized Adaptive Testing System

A traditional fixed-length exam presents the same number of questions to each test taker, without considering how well the person is performing on the exam. The score from this type of test usually depends on the number of questions answered correctly. For any given person, some questions on the test will be far too easy and some far too hard. The test taker's answering easy questions correctly doesn't tell us much about him or her. Most people answer the easy ones correctly. For a similar reason, the test taker's answering the difficult questions incorrectly also tells us very little. If a test were able to discover the level, on a scale of easy to difficult, at which the person taking it begins to find the questions challenging, a score could be derived for that level and the test would be more effective. A computerized adaptive test (CAT) does just that.CAT is based on Item Response Theory. A CAT tailors itself to the ability of the test taker. By taking into account how each person taking the same CAT answered previous questions, the test will adjust to give the low-ability examinee a different set of questions from what it gives a high-ability examinee. The CAT usually ends when the amount of measurement error around the ability estimate reaches an acceptable level. Both individuals may answer the same percentage of questions correctly, but because the high-ability person can answer more difficult questions correctly, he or she will get a higher score.Significantly less time is needed to administer CATs than fixed-item tests since fewer items are needed to achieve acceptable accuracy. Shorter testing times also reduce fatigue, a factor that can significantly affect an examinee's test results. CATs can provide accurate scores over a wide range of abilities while traditional tests are usually most accurate for average examinees. In general, computerized testing greatly increases the flexibility of test management.Item Response Theory (IRT), as the base of CAT, is a theory of measurement based on mathematical model. Item Response Theory is the study of the mathematical relationship between abilities and item responses and characterizes the relationship by a mathematic model. In our CAT system , we characterize the relationship between abilities and item response by Three-Parameter Logistic Model(3PL Model). In this article, we analyzes the CAT both in theory and practice.The first part of this article points out the disadvantage of traditional testing and the advantage of CAT, and shows the development of CAT.The second part presents an introduction of IRT, which is the theoretical basis of CAT. It focus on the effect of 3PL model and Information function in the CAT.The third part puts forward a framework of CAT system, and account the structure of CAT, the process of testing, including the module of our system, the structure of Database, estimate of parameter and so on.The forth part introduces Active Server Page and ActiveX Data Objects which we used in CAT system in brief.The fifth part demonstrate the program of our CAT system in detail.Finally, the article give an account of the testing of the software and summarizes the research work.