摘要
受假设检验方法的启发,该文引出了一种基于项目反应理论的新等值方法--平方根等值准则.它具有一些特点:定义式中答对、答错概率同时出现而不能互相替代;极易从0-1评分模式的版本转换到多级评分版本;它可以看成是Haebara等值准则的加权形式.以等值系数估计值的误差大小为衡量标准,以Wilcoxon符号秩检验为依据,大量的Monte Carlo模拟结果显示了一种有趣的现象,即等值方法的运用范围既与项目参数估计精度有关,又与等值系数A的范围有关,但与另一个等值系数B的范围无关.当项目参数估计精度较高或中等而A取值在0.9～1.3之间,新方法往往比Stocking_Lord方法和Haebara方法的估计误差小且有显著性差异,当项目参数估计精度较低时,而A从1.0～2.0时新方法都有优越性.
Motivated by Freeman-Tukey statistical test, a new equating criterion under item response theory, known as square root criterion （SQRcrit）, is proposed. The new method has some features, such as the probabilities for correct response and error response appear simultaneously in the formula and they could not be replaced each other. It is easy to generalize from dichotomous version to polytomous version. It could be regarded as one of weighted versions of Haebara equating criterion. In terms of the recovery of estimating the equating coefficients and based on Wilcoxon sign-rank test, the findings of Monte Carlo study show an interesting phenomenon that if the equating approaches are proper relates to the accuracy of the estimation of the item parameters （AEIP） and to the domain of the equating coefficient A, but not to the domain of B. When the AEIP is higher, the new equating criterion is proper for 0.9≤A≤1.3. And when the AEIP is lower, the new equating criterion is proper for 1≤A≤2.
出处
《心理学报》
CSSCI
CSCD
北大核心
2005年第5期674-680,共7页
Acta Psychologica Sinica
基金
国家自然科学基金(60263005)
江西省自然科学基金(0411021)
江西省教育厅科技项目
教育科学规划重点课题(DBB010501)
江西省分布计算工程技术研究中心开放课题基金资助。
关键词
项目反应理论
等值方法
等值系数
item response theory, test equating criterion, equating coefficient.
