变长CD-CAT中的曝光控制与终止规则

发布时间：2018-01-17 22:15

本文关键词：变长CD-CAT中的曝光控制与终止规则　出处：《心理学报》2015年01期 　论文类型：期刊论文

【摘要】：本研究借鉴传统计算机化自适应测验的思想,并结合认知诊断的特点,在认知诊断框架下提出了4种变长CD-CAT的终止规则,分别是属性标准误法(SEA)、邻近后验概率之差法(DAPP)、二等分法(HA)以及混合法(HM)。在未控制曝光和采用不同曝光控制条件下,与HSU法及KL法进行了比较。研究结果表明:(1)终止条件越严格,平均测验长度越长,按测验长度最大值终止的测验百分比越大,模式判准率越高。(2)当未加入曝光控制时,4种新的终止规则均有较好表现,与HSU法十分接近。随着最大后验概率预设值的增加或?的减小,模式判准率呈上升趋势,平均测验长度逐渐增加,但在题库使用率方面均较差。(3)当加入项目曝光控制时,6种变长终止规则下的题库使用率有了极大的提升,仍能保持较高的模式判准率,并且不同的曝光控制方法对终止规则的影响是不同的。其中,相对标准终止规则极易受到曝光控制方法的影响。(4)综合来看,SEA、HM以及HA法在各项指标上的表现与HSU法基本一致,其次为KL法和DAPP法。
[Abstract]:Based on the idea of traditional computerized adaptive test and the characteristics of cognitive diagnosis, this study proposes four termination rules of variable length CD-CAT under the framework of cognitive diagnosis. They are the attribute standard error method (SEAA), the adjacent posterior probability difference method (DAPP), the dichotomy method (HAA) and the mixed method (HMX) under the condition of uncontrolled exposure and different exposure control conditions. Compared with HSU method and KL method, the results show that the more strict the termination conditions are, the longer the average test length is, and the larger the percentage of tests terminated according to the maximum test length. When the exposure control is not added, the four new termination rules have better performance and are very close to the HSU method. The average test length was gradually increased, but the utilization rate of item bank was poor. 3) when the item exposure control was added. The utilization rate of question bank under six kinds of variable length termination rules has been greatly improved, and still can maintain a high pattern accuracy rate, and different exposure control methods have different effects on the termination rules. The relative standard termination rule is easy to be affected by exposure control method. (4) taken together, the performance of SEAHM and HA method is basically consistent with that of HSU method. The second is KL method and DAPP method.
【作者单位】：西南大学心理学部;北京师范大学认知神经科学与学习国家重点实验室;伊利诺伊大学香槟分校教育心理学系;中国基础教育质量评价与提升协同创新中心;
【基金】：高等学校博士学科点专项科研基金资助课题(20120003110002)的资助
【分类号】：B841
【正文快照】： (3伊利诺伊大学香槟分校教育心理学系,香槟,伊利诺伊州61820美国)(4中国基础教育质量评价与提升协同创新中心,北京100875)1引言近些年来,国内外对认知诊断计算机化自适应测验(Cognitive Diagnostic Computerized AdaptiveTesting,CD-CAT)的研究越来越多。CD-CAT结合了认知诊断

【参考文献】