摘要
基于项目反应理论中的LOGISTIC双参数模型研究共同题非等组设计下,考生能力分布与被试量对等值的影响。等值方法采用分别校准下的项目特征曲线法、Stocking-Lord法、Haebara法。等值结果采用等值分数标准误、等值系数标准误、共同题参数稳定性三种方法进行评价。研究结果表明,考生能力分布越接近,被试量越大,等值误差越小;且Stocking-Lord法较Haebara法的等值结果更稳定。
Based on the LOGISTIC two-parameter model of Item Response Theory(IRT),this paper studied the influence of the ability distribution of examinees and the number of subjects on the score equating under the Anchor Test Design.The method adopts the Stocking-Lord method and the Haebara method under calibration.The results were evaluated by the standard errors of the equivalent score,the standard errors of the equivalent coefficient and the stability of the parameters of the common items.The results show that the closer the examinees’ability distribution is,the larger the number of subjects is and the smaller the equivalent error is.Moreover,the Stocking-Lord method is more stable than the Haebara method.
作者
韩晓杰
任杰
Han Xiaojie;Ren Jie(Beijing Language and Culture University,Beijing,100083)
出处
《考试研究》
2021年第1期58-64,共7页
Examinations Research
关键词
项目特征曲线法
等值
能力分布
被试量
Method of Item Characteristic Curve
Equating
Ability Distribution
Number of Examinees