摘要
和基于实测数据的DIF研究相比,基于模拟数据的DIF研究不仅可以自由操纵实验条件,而且可以给出检验力和I型错误指标。本文详细阐述了二级计分DIF模拟数据的产生原理,其产生过程包括四个阶段:选择DIF产生思路,选择项目反应理论模型,确定考生特征、题目特征和复本数,计算考生在题目上的正确作答概率并转化为二级计分数据。并且分别利用常用软件Excel和专业软件WinGen3展示了二级计分DIF模拟数据的产生过程。
Compared to DIF study based on real data, DIF study based on simulated data not only can manipulate the experimental conditions freely, but also provide indicators of power and Type I error. This paper describes the principle of generating dichotomous DIF simulated response, including four phases: choosing the way to introduce DIF, selecting the item response theory model, setting examinee characteristics, item characteristics and number of replications, calculating subject' s probability of giving a correct response and transforming it into dichotomous response. The process of generating dichotomous DIF simulated response is showed by using the common software Excel and specialized software WinGen3 separately.
出处
《考试研究》
2012年第6期80-87,19,共9页
Examinations Research
关键词
项目功能差异
模拟研究
Differential Item Functioning, Simulation Study