Methods for Restricting Maximum Exposure Rate in Computerized Adaptative Testing

摘要: The Sympson-Hetter (1985) method provides a means of controlling maximum exposure rate items in Computerized Adaptive Testing. Through series simulations, control parameters are set that mark the probability administration an item on being selected. This presents two main problems: it requires long computation time for calculating and is slightly above fixed limit. Van der Linden (2003) presented alternatives which appear to solve both problems. impact these methods measurement accuracy has not been tested yet. We show how over-restrict some highly discriminating and, thus, decreased. It also shown that, when desired near minimum possible value, offer empirical clearly goal. A new method, based initial estimation selection with restricted (Revuelta & Ponsoda, 1998), this paper. can be used van Linden's methods. option, Sympson-Hetter, speeds convergence without decreasing accuracy. One objectives administering tests accurate as- sessment examinee's trait level. In order achieve adequate measurement, necessary responding correctly marked solely by their psychometric characteristics lev- el. case examinee receiving test prior knowledge he will have respond, would no longer hold there over-estimation his level reduce test's validity. risk especially present applied computerized adaptive (CAT). kind test, bank remain operative reasonably period time. future obtain part if receives information from already who remembers faced. higher overlap between examinees, understood as proportion shared, average,