认知诊断测验的属性分类一致性和分类准确性指标  

An Evaluation of Attribute- level Classification Consistency and Accuracy Indices in Cognitive Diagnostic Assessment

作　　者：汪文义[1] 宋丽红[2] 陈平[3] 丁树良[1] 程艳[1] 

机构地区：[1]江西师范大学计算机信息工程学院,南昌330022 [2]江西师范大学初等教育学院,南昌330022 [3]北京师范大学中国基础教育质量监测协同创新中心,北京100875

出　　处：《心理学探新》2016年第3期264-269,共6页Psychological Exploration

基　　金：全国教育科学规划教育部重点课题(DHA150285);国家自然科学基金(31500909;30860084;31160203;31360237;31300862;61262080);教育部人文社会科学研究青年基金项目(13YJC880060);江西省社会科学研究"十二五"(2012年)规划项目(12JY07);江西省教育科学2013年度一般课题(13YB032);江西省教育厅科技计划项目(GJJ13207);东北师范大学应用统计教育部重点实验室开放课题(KLAS130028614);国家社会科学基金(12BYY055);国家教育科学规划项目(CCA110109);江西师范大学青年成长基金和博士启动基金

摘　　要：分类一致性和分类准确性是衡量考试信效度的两个重要评价指标。基于项目反应理论下分类一致性和分类准确性指标,提出认知诊断测验的属性(模式)分类一致性和分类准确性指标,讨论分类一致性指标、分类准确性指标与属性估计误差之间的关系,并由属性掌握概率的估计标准误推导出属性分类准确性的上限。结果显示:属性(模式)分类一致性可准确估计重测一致性;分类准确性指标计算简单,可准确估计认知诊断测验的判准率。Classification consistency and accuracy indices can be used as important indicators of the reliability and validity of classifica- tion results. Cognitive diagnostic assessment focuses on make attribute classification - based decisions while reliability of classification is often not reported in diagnostic score reporting. Classification consistency and accuracy indices（ Cui, et al. ,2012）were used to evaluate the consistency and accuracy of classification results at the whole - pattern level, but it needs to develop attribute - level classification consistency and accuracy indices for CDA（ Cui ,et al. ,2012）. The study investigated attribute -level and pattern -level classification consistency and accuracy indices based on the Rudner method or the Guo method （Guo ,2006）in item response theory. Procedures were developed for the computation of classification consistency and accuracy indices specifically designed for cognitive diagnostic assess- ments. We also used simulated data from the deterministic input noisy ＂and＂ gate model to evgluate their performances.. According to the simulation results, the classification accuracy indices performed well with simulated diagnostic tests in that their values matched closely with the simulated correct classification rates across different simulation conditions. In addition,classification consistency indices matched with the results of the test - retest consistency indices.

关 键 词：分类一致性 分类准确性 属性经验信度 属性α信度 确定性输入噪声与门模型 

分 类 号：B841.2[哲学宗教—基础心理学]