国际教育成效评价协会儿童认知发展状况测验项目功能差异分析  

Validity Equivalence between the Chinese and English Versions of the IEA Child Cognitive Developmental Status Test

作　　者：王蕾[1] 黄晓婷[2] 

机构地区：[1]美国加州大学伯克利分校教育研究院 [2]美国加州大学伯克利分校

出　　处：《考试研究》2006年第4期94-107,共14页Examinations Research

基　　金：国家留学基金委资助

摘　　要：本研究旨在从一维和多维的角度检测国际教育成效评价协会(IEA)儿童认知发展状况测验中中译英考题的项目功能差异(DIF)。我们分析的数据由871名中国儿童和557名美国儿童的测试数据组成。结果显示,有一半以上的题目存在实质的DIF,意味着这个测验对于中美儿童而言,并没有功能等值。使用者应谨慎使用该跨语言翻译的比较测试结果来比较中美两国考生的认知能力水平。所幸约有半数的DIF题目偏向中国,半数偏向美国,因此利用测验总分所建立的量尺,应该不至于有太大的偏误。此外,题目拟合度统计量并不能足够地检测到存在DIF的题目,还是应该进行特定的DIF分析。我们探讨了三种可能导致DIF的原因,尚需更多学科专业知识和实验来真正解释DIF的形成。In this study,we examine the validity equivalence between the Chinese and English versions of the IEA (International Association for the Evaluation of Educational Achievement) Child Cognitive Developmental Status Test, through unidimensional and multidimensional analysis of differential item functioning (DIF).The test data from the subjects of 871 Chinese children and 557 American children are analyzed.It was found that half of the items exhibited substantial DIF,suggesting that this test did not show functional equivalence between the two versions so that the users should be very cautious in using the test to compare group difference between Chinese and American children.On the other hand,approximately half of the DIF items favored American while the other half favored Chinese,suggesting the scale obtained from the raw scores was not seriously biased.Furthermore,the item infit statistics were not sensible to detecting DIF items.Three plausible causes of DIF were speculated.More content knowledge and experiment are needed to explore the causes of DIF.

关 键 词：项目功能差异(DIF) Rasch测量模式 考卷翻译 多维Rasch测量模式 项目反应理论 认知发展 

分 类 号：B844.1[哲学宗教—发展与教育心理学]