题组反应理论及其在中学英语考试中的应用研究

题组作为众多测验中的一种常见题型,由于项目间存在一定程度的依赖性而违背了局部独立性假设,若用项目反应模型进行参数估计将会出现较大的偏差.题组反应理论将被试与题组的交互作用纳入到模型中,解决了项目间相依性的问题.笔者对题组反应理论的发展、基本原理及其相关研究进行了综述,并将其应用在中学英语考试中.与项目反应理论相对比,结果发现：（1）题组反应模型与项目反应模型在各参数估计值的相关系数较强,尤其是能力参数和难度参数;（2）在置信区间宽度的比较上,题组反应模型在各个参数上均窄于项目反应模型,即题组反应模型的估计精度优于项目反应模型.

A Review of Testlet Response Theory and Its Application in English Exam Data

One important assumption made in item response theory （IRT）is the assumption of local independence. Violation of this assumption occurs under some conditions such as testier or item - bundle in which items share the same stimulation. Lots of examinations contain testlet items or item - bundle. When item response theory models are applied in testlet items can lead to distortion of parameter estimation. This paper introduced related studies both in application and theory about testlet response theory（TRT） ,and applied TRT into middle school English exam data. Compared with IRT result,we found that（ 1 ）the correlation coefficients of every parameter beween TRT and IRT are strong, especially ability parameter and item difficulty;（ 2 ）In comparison of MPIW, the result of TRT is shorter than result of IRT, and that is to say TRT is more precise than IRT.