摘要
本文探究不同因素对确定性输入噪声"或"门模型(DINO)判准率的影响。模拟实验表明:DINO模型更适用于离散型属性层级结构;测验长度的合理增加有助于提高诊断的准确率;DINO模型对属性层级结构不敏感;认知属性个数的增加会降低DINO模型的诊断准确率,实际应用中建议认知属性个数控制在6个以下。
DINA is a cognitive diagnosing model in wide use. DINO is developed from the DINA aimed at psychological scales. Various factors, such as attribute hierarchy, test length, the numbers of attributes etc. will affect the accuracy of diagnosis to some extent. Based on DINO, affecting degree of different factors to accuracy index of classification is investigated in the paper. The results show different factors have varying degree of influence on the diagnostic accuracy ： DINO is more suitable to the discrete attribute structure; Test -length is not longer the better when ignore the test time ; DINO is not sensitive to attribute construction ; Increasing the number of attributes will reduce the diagnostic accuracy of DINO, so the number of cognitive attributes is suggested less than 6 in real application.
出处
《考试研究》
2013年第4期60-67,共8页
Examinations Research
