基于属性多级化的认知诊断计算机化自适应测验设计与实现

发布时间：2018-05-15 21:39

本文选题：pCD-CAT + 属性多级化　；参考：《心理学报》2015年11期

【摘要】：本研究在传统CD-CAT的基础上进行拓展,开发设计了可以处理属性多级化的CD-CAT(记为p CD-CAT),而且当测验所有属性的水平数Lk=2时则p CD-CAT可简化为CD-CAT,因此传统CD-CAT是本研究设计开发p CD-CAT的一个特例。Monte Carlo模拟实验结果表明:基于属性多级化框架下设计的p CD-CAT具有较好的诊断正确率、题库安全性和较高的测验效率,弥补了传统CD-CAT不足;当属性多级化时,若采用传统CD-CAT方法,则诊断正确率很不理想(属性模式判准不到30%),表明传统CD-CAT在属性多级化测验情景时不适宜,而本文设计的p CD-CAT是一种不错的选择(属性模式判准高达80%以上)。总之,本研究对于进一步拓展CD-CAT在实践中的应用提供了新方法和新技术支持。
[Abstract]:This study is based on the traditional CD-CAT. A CD-CAT (denoted as p CD-CAT) which can deal with multilevel attributes is developed, and p CD-CAT can be simplified to CD-CAT when the level number of all attributes is tested Lk = 2. Therefore, traditional CD-CAT is a special case of p CD-CAT in this study. The result shows that the design of p CD-CAT based on attribute multilevel framework has good diagnostic accuracy. The security of test bank and high test efficiency make up for the deficiency of traditional CD-CAT. When the attribute is multilevel, if the traditional CD-CAT method is used, The diagnostic accuracy is not satisfactory (attribute pattern is less than 30%), which indicates that traditional CD-CAT is not suitable for attribute multilevel test, and p CD-CAT designed in this paper is a good choice (attribute pattern is more than 80%). In a word, this study provides new methods and new technical support for further expanding the application of CD-CAT in practice.
【作者单位】：江西师范大学心理学院江西省心理与认知科学重点实验室;
【基金】：国家自然科学基金(31100756,31300876,31160203,31360237) 江西省社会科学规划项目重点项目(13JY01) 江西省教育科学规划项目(12YB088,13YB029) 高等院校博士点基金项目(20123604120001) 江西师范大学青年英才培育资助计划东北师范大学应用统计教育部重点实验室开放课题(130026509)资助
【分类号】：B842.1

【参考文献】