英语标准化考试评价中IRT与CTT的比较研究  

The Comparative Study of IRT and CTT in the Assessment of English Standardized Examination

作　　者：陈谨[1] 何静[2] 佟仁城[1] 许健[1] 

机构地区：[1]中国科学院研究生院管理学院,北京100049 [2]北京中医药大学管理学院,北京100029

出　　处：《数学的实践与认识》2011年第20期1-9,共9页Mathematics in Practice and Theory

摘　　要：考试质量的评价对教学管理决策至关重要.随着大规模标准化考试的不断涌现,对公正考试和科学评判的要求越来越高,阐述了IRT理论的原理及估算方法,并通过实证研究将IRT与经典测试理论(CTT)进行了理论比较,并基于中科院研究生院英语标准化考试的数据,从试题评价参数估计和考生能力估计等方面进行了实证对比.结果表明:IRT对项目参数的估计不依赖样本变化而变化,比CTT具有明显的稳定性;同时,IRT记分充分考虑了各个项目的性能差异,被试的得分更少雷同,更为精确.The assessment of examination quality is of great importance to policy making in teaching administration. With the emergence of large-scale standardized examinations, more importance and stringent requirements have been placed on fair examination and scientific critical evaluation of examinations. The present paper addresses the basic principle and computational methods of IRT theory, makes an empirical comparison between the IRT theory and CTT and contrasts between the test items parameter estimation and the students＇ ability estimation on the basis of the data collected in the standardized examination at GUCAS. It is found that the test item parameter estimation of IRT is independent of the quality changes of the sample, showing better stability over CTT; at the same time, the scoring system of the IRT theory takes into full consideration the functional difference among the test items, test scores being more accurate with fewer identical scores.

关 键 词：IRT CTT 英语标准化考试 比较 

分 类 号：H319[语言文字—英语]