摘要
非参数认知诊断方法只需Q矩阵,无须估计参数等优势,使其近年来备受关注,但其判准率如何,哪些因素会影响判准率,并未有相关研究。本文通过模拟研究,探讨属性个数、样本容量和属性层级结构对该方法判准率的影响,结果表明:1属性个数的增加会导致判准率的下降,但其稳健性较参数模型要好;2属性间逻辑关系的紧密度会对判准率产生影响,但其表现形式与参数模型不同;3样本容量对判准率影响很小,属性个数较少时100人已足够,随着属性个数的增加,500人的样本已是较佳样本。本研究为CDA走向小型测评及课堂评估提供了参考信息。
Cluster diagnosis method as a kind of nonparametric cognitive diagnosis method, which was only requires the Q-matrix made it more and more popular in recent years, but what kinds of factors will influence the classification accuracy of this method is not clear. This paper uses simulation studies to discuss the Influence Factors of this method’s classification accuracy, such as the number of attribute, sample size and attribute hierarchy. The result indicated that：①more attributes lead to the lower classification accuracy, but this method is more stable than the parameter model.②The tightness of the logical relationship between attributes will influence on the classification accuracy, but the form is different from parameter model. ③The sample size has little impact on the classification accuracy, the size of 100 people is enough and the size of 500 people can achieve a good level. The above results provide some reference information for the application of CDA in small assessment and classroom assessment.
出处
《中国考试》
2015年第2期25-32,共8页
journal of China Examinations
基金
浙江省高校重大人文社科项目攻关计划(2013QN048)资助项目
