The Maximum Priority Index Method With Flexible Content Balance Constraints In Computerized Adaptive Testing

The maximum priority index (MPI) method is a new heuristic approach which is proposed by Hua-Hua Chang in recent years. The method is for severely constrained item selection in computerized adaptive testing. Compared with the weighted deviation modeling method which is also adopted frequently, it leads to fewer constraint violations and better exposure control. But the MPI method in its current form is limited to constraints that are in the form of upper bounds. For flexible content balancing constraints that involve lower bounds as well, the MPI method can not to be used.In this paper, based on the two-phase item selection and used of the weight of the upper bound and the lower bound, we modify the function of the original method. The problem presented above is solved. Our simulation study shows that the modified MPI method still leads to fewer constraint violations while maintaining the same level of measurement precision in compare to the weighted deviation modeling method.