混合模型在英语听力诊断测评中的应用——基于Mixed-CDMs与G-DINA模型的对比分析  

The Application of Mixed-CDMs in English Listening Diagnostic Assessment--Based on the Comparison Analysis between Mixed-CDMs and G-DINA Model

作　　者：董艳云[1] 马晓梅[1] 孟亚茹[1] DONG Yan-yun;MA Xiao-mei;MENG Ya-ru(School of Foreign Studies,Xi’an Jiaotong University,Xi’an,Shaanxi,China 710049)

机构地区：[1]西安交通大学外国语学院,陕西西安710049

出　　处：《现代教育技术》2020年第3期52-58,共7页Modern Educational Technology

基　　金：国家社科基金项目“英语认知诊断测评模式构建及有效性论证”(项目编号:17BYY015)的阶段性研究成果。

摘　　要：文章使用GDINA R程序包,借助Wald检验为英语听力诊断试题中的多属性题目选出最优简约模型,组成混合模型(Mixed-CDMs)。基于Mixed-CDMs与G-DINA模型的对比分析,文章发现:在样本量不够大(N=726)的情况下,Mixed-CDMs满足模型-数据绝对拟合的较高要求,相对拟合性、人员拟合性、属性分类的可靠性以及参数估计的准确性都有所提高,且属性之间的关系更加直观易读。由此,文章验证了混合模型对于英语听力诊断测评具有适用性并有一定的应用优势,这为混合模型在英语听力测试中的应用提供了实证依据,有助于加深对英语听力认知属性关系的了解,并可为其它语言测试使用混合模型提供借鉴。Using the GDINA(Generalized Deterministic Inputs,Noisy"and"Gate)R program package and the Wald test,this paper selected the optimal reduced models for the multi-attribute items in the English listening diagnostic test,and formed the Mixed-CDMs(Cognitive Diagnosis Models).Based on the comparison analysis between Mixed-CDMs and G-DINA model,it was found that in the case of a small sample size(N=726),Mixed-CDMs met the high requirements of absolute fitting of model-data,and relative fitting,person fitting,reliability of attributes classification and accuracy of parameter estimations were improved.Meanwhile,the inter-attribute relationships were more intuitive and readable.Accordingly,it was verified that Mixed-CDMs had applicability and certain application advantages for English listening diagnostic assessment,which was expected to provide empirical basis for the application of Mix-CDMs in the English listening test,help deepen the understanding of cognitive attribute relationship of English listening,and provide reference for other language test using Mixed-CDMs.

关 键 词：认知诊断 英语听力 混合模型 G-DINA模型 

分 类 号：G40-057[文化科学—教育学原理]