摘要
根据同一被试在锚题上的得分概率向量的模式应该十分相似的直观想法，该文提出了一种新的项目反应理论框架下的等值准则——余弦准则．我们采用GPC模型，利用Monte Carlo模拟方法将余弦准则与Haebara等值准则（Hcrit）glStocking-Lord等值准则（SLcrit）进行比较．研究结果表明，当项目参数估计精度较高或中等而A的取值范围在0．9-2．0之间，新方法往往比Hcrit和SLcrit方法估计的误差小且有显著性差异．
The probability mass vectors of an examinee who responses to the ancher items in two test forms are intuitively quite approximate. Based on this intuition, a new test equating criterion, known as COSine criterion（COSerit）, is proposed in this paper The COScrit is compared with Haebara equating criterion（Hcrit）and Stocking-Lord equating criterion （SLerit）through Monte Carlo simulation using Generalized Partial Credit Model. The study shows that the results of the new approach sometimes are better than Hcrit and SLerit in terms of the recovery of estimating the equating coefficients and have significant differences when the value of the equating coefficient A is about 0.9 to 2.0.
出处
《江西师范大学学报（自然科学版）》
CAS
北大核心
2008年第2期224-227,245,共5页
Journal of Jiangxi Normal University(Natural Science Edition)
基金
国家自然科学基金(60263005)
江西省教育厅科技项目(GJJ08154)
江西省高校人文社科项目(JY06201)
全国教育考试科研规划项目(2006JKS3063)
卫生部课题(JM20060070)资助
关键词
项目反应理论
等值准则
广义分部评分模型
等值系数
item response theory
equating criterion
generalized partial credit model （GPCM）
equating coefficients
