属性关系与诊断测验分类影响的探讨

发布时间：2018-04-17 01:13

  本文选题：认知诊断 + DINA模型　； 参考：《统计与决策》2017年18期

【摘要】：认知诊断模型-DINA模型假定属性之间是相互独立、非补偿的关系,但在实际应用中,属性之间有可能存在补偿关系。文章探讨了在不同的被试总体下,DINA模型在属性之间独立和存在补偿关系时作答数据上的表现,结果表明,当测验中有部分属性间存在补偿关系时,会给DINA模型诊断分布带来较大的影响,在固定其他条件(被试人数、随机因子)条件下,存在和不存在补偿作用所导致的诊断分类率之间有显著差异,且在不同大小补偿作用下,诊断分类率之间的差异显著;随着随机因子的增加,不同大小的补偿因子所导致的诊断分类率之间的差异会变小,补偿因子和随机因子之间存在着显著的交互作用。
[Abstract]:The cognitive diagnosis model DINA assumes that the attributes are independent and non-compensatory, but in practice there may be a compensatory relationship between the attributes.This paper discusses the performance of Dina model in different subjects in terms of the independence of attributes and the existence of compensation relationship. The results show that when there is a compensation relationship among some attributes in the test,Under the condition of fixed other conditions (number of subjects, random factors), there is a significant difference between the diagnostic classification rate caused by the existence of compensation and the absence of compensation.With the increase of random factors, the difference between diagnostic classification rates caused by compensation factors of different sizes will become smaller, and the difference between diagnostic classification rates will become smaller with the increase of random factors.There is a significant interaction between the compensation factor and the random factor.
【作者单位】： 延安大学数学与计算机科学学院;淮北师范大学管理学院;
【基金】：国家自然科学基金资助项目(11471007)
【分类号】：O213
，

本文编号：1761403