带区分度约束的选题策略研究

The Item Selection Strategies with Discrimination Constraint

最大优先级指标(MPI)选题策略可以较好地满足非统计性约束,按a分层的选题策略可以有效提高低区分度项目的利用率,结合两者的优势,构造了附加区分度约束的两阶段MPI选题策略.Monte Carlo模拟研究表明:新选题策略在题库的未使用率方面有明显改进,在测量精度和约束条件控制等评价指标上较现有方法差异不大.

MPI can well meet the statistical constraints,and a-stratified method can effectively improve the utilization rate of low discrimination item. Combining the advantages of MPI and a-stratified method,a two-phase MPI item selection strategy with additional distinction constraint is constructed. The simulation study of Monte Carlo shows that the new item selection strategy has improved a lot in the inavailability of item bank,which is about the same as the existing approach in measurement accuracy,constraint management and other evaluation in dices.