Abstract: Computerized adaptive testing (CAT) is a kind of test model, which is based on the measurement theory and computer techniques. It selects items according to each examinee’s answers of the previous items. Item selection method is an important component of a computerized adaptive testing and has significant effects on the measurement efficiency, test security, test reliability and validity. According to whether there are some non-statistical constraints, item selection methods are classified into different categories. Then the item selection methods developed for traditional CAT and cognitive diagnostic CAT are introduced and evaluated separately. At last, future studies should improve the synthesis performances of item selection methods and pay more attention to examine item selection methods for polytomous scoring items and cognitive diagnostic CAT.

Key words: computerized adaptive testing, item selection method, measurement accuracy, non-statistical constraint