认知诊断测验编制的原则
首发时间:2009-06-01
摘要:Tatsuoka给出的实例表明,不同知识状态可能对应同一理想反应模式,即对知识状态产生误判。如果不是对测验进行事后分析,找出属性及层级,而是采用Leighton等人所倡导的方法,在认知诊断测验编制之前确定所测属性及其层次,导出可达阵,这时可以证明只要将可达阵作为认知诊断测验蓝图的一部分,则可避免这一问题。这一原则不仅对认知诊断测验蓝图的设计有指导作用,而且对制定有认知诊断功能的计算机化自适应测验的选题策略有着重要的参考作用。
关键词: 测验蓝图 认知诊断 可达矩阵 理想反应模式 计算机化自适应测验
For information in English, please click here
A principle of Construction of Cognitive Diagnostic Test
Abstract:Some researchers ,such as Tatsuoka and Leighton,Gierl ,named a few, all consider the Qr matrix can be interpreted as a cognitive blueprint or cognitive test specification. But when there are so many attributes involved in the diagnostic test and their dependence is weaker, then the number of the columns in Qr matrix is too large to be included in the test blueprint.Tatsuoka pointed out that it is clear that a set of attribute mastery patterns(knowledge states) and a set of ideal item-response patterns do not have an one-to-one correspondence. Several attribute-mastery patterns correspond to a single ideal item-response pattern. If several attribute-mastery patterns correspond to a single ideal item-response pattern, it is called mismatch of attribute mastery pattern, and it is harmful to classification accuracy. In this paper a cognitive diagnostic test blueprint that could not bring mismatch of attribute mastery pattern is proposed when the logical order of the development of cognitive diagnostic test is according to Leighton and Gierl and if the reachability matrix is included in a blueprint, the blueprint for a cognitive diagnostic test is a test without mismatch of attribute mastery pattern .To include the reachability matrix in the test blueprint is a sufficient condition to avoid mismatch of attribute mastery pattern .This is a fact that is proved and is benefit to construct test for cognitive diagnosis purpose. Some possible applications of this principle are considered in this paper.
Keywords: test blueprint cognitive diagnosis reachability matrix attribute mastery pattern computerized adaptive testing
基金:
论文图表:
引用
No.3272747448512438****
同行评议
共计0人参与
勘误表
认知诊断测验编制的原则
评论
全部评论0/1000