二项式模型在标准参照性语言测验长度研究中的应用  

The Use of Binominal Probability Model in the Decision of Criterion- Referenced Test Length

作　　者：柴省三[1] 

机构地区：[1]北京语言大学汉语水平考试(HSK)中心

出　　处：《考试研究》2013年第4期51-59,共9页Examinations Research

基　　金：北京市共建项目"北京语言文化建设(2011)"研究的阶段性成果之一

摘　　要：随着国内外教育测量理念的转变,传统的常模参照测验所提供的相对性评价信息已无法满足考试用户和考生的需求,标准参照测验(CriterionReferenced Test,CRT)的社会价值越来越受到重视。在对被试掌握程度进行分类决策的CRT测验中,如何确定恰当的测验长度和合格分数是影响测验分类误差的重要因素。本文在对CRT测验研究的现状、原理和用途进行考察的基础上,专门介绍了二项式概率模型在CRT测验长度决策研究中的理论和过程,并以误差控制为原则,对二项式模型在综合性标准参照语言测验长度和合格分数决策中的应用过程进行了研究。With the development of cognitive psychology and cognitive linguistics, the psychological and educational fields need more and more criterion-referenced tests in order to elicit more detailed diagnostic information about test - takers. But the decision of test length and qualified test score of the test is one of the most important factors influencing the classification consistency and accuracy of the criterion-referenced test. This paper, based on the discussion of the state quo of criterion - referenced tests, specifically addresses itself to the use of Binominal Probability Model in the decision of test length and qualified test score, and shows how to use the Model to determine test length in general criterion-referenced language tests.

关 键 词：二项式模型 测验长度 标准参照测验 分类一致性 

分 类 号：G424.74[文化科学—课程与教学论]