在香农熵选题策略的基础上,采用4种不同的测验蓝图考察运用可达阵对诊断精度的影响.Monte Carlo模拟试验表明:选择可达阵所有列对应的项目对改进计算机化自适应诊断测验(CD-CAT)的诊断精度有着重要的作用,而且对于CD-CAT的选题策略有重要的参考价值.在CD-CAT过程中选择较多可达阵的列对应的项目能够明显提高模式判准率,项目质量对结果的影响不大:而未包含所有可达阵的列对应的项目时,项目质量对于模式判准率有较大的影响.
On the basis of Shannon's Entropy item-selection strategy, in this paper four different test blueprints are adopted to examine effect on diagnostic accuracy in the process of applying the reachability matrix(RM). The results of Monte Carlo simulation show that choosing all the corresponding items of RM plays an important role to improve the diagnostic accuracy in computerized adaptive testing with cognitive diagnosis(CD-CAT), and it is also important reference to item-selection strategy in CD-CAT. In the process of CD-CAT choosing more corresponding items of RM can obviously increase the pattern match rate(PMR), items quality has little effect on the results, and when item pool does not contain all corresponding items of RM, item quality has great influence on PMR. Using RM is benefit to construct test and item-selection strategy for cognitive diagnosis.