基于不同CDM视角的CD-CAT题库建设

在CD-CAT题库建设过程中,选择合适的认知诊断模型(CDM)标定题库参数是一个关键的步骤.若一个题库均采用一个CDM来标定题库所有参数,则可能会造成模型资料失拟.提出了一种基于混合模型的题库建设思路,通过Monte Carlo 模拟研究验证了基于混合模型建立题库的效果,并与传统的基于单一模型的题库建设进行了比较,为CD-CAT在选用合适的模型方面提供了理论借鉴.

The Study on the CD-CAT Item Bank Based on the Perspective of Different CDMs

The item bank is a very critical part of the cognitive diagnostic computerized adaptive testing(CD-CAT).In regular CD-CAT,an item bank uses the same cognitive diagnostic model to calibrate the item parameters,however,all the items in item bank can’t be fully fitted with the same CDM.In the study the advantage of the method proposed by de la Torre to select an appropriate CDM for each item of item bank within a G-DINA framework has been taken,for convenience,this method is referred to as a Mixed CDMs,and use a complex simulation study to compare the performance of Mixed CDMs and five special models(G-DINA,DINA,DINO,A-CDM and RRUM)in CD-CAT.The results show that across all conditions,the Mixed CDMs provided the ideal classification accuracy and utilization rate of the item bank.