概化理论和多层面Rasch模型在建立“职前中学英语教师口语考试模型”中的应用  

Application of Generalizability Theory and Many-facet Rasch Model in Developing a Speaking Test for Prospective Secondary School English Teachers

作　　者：孙海洋[1] 

机构地区：[1]中国科学院研究生院

出　　处：《外语与外语教学》2011年第5期57-62,共6页Foreign Languages and Their Teaching

摘　　要：本研究以概化理论和多层面Rasch模型为基础，运用GENOVA和FACETS分析了一项针对职前中学英语教师的口语考试模型的试测结果，旨在探索此项考试的信度和效度，为完善考试设计提供依据。两种方法的分析结果为我们提供了互为补充的信息：GENOVA分析结果表明，适当减少考试任务不会对考试的信度有很大影响，然而减少评分员人数会对信度有较大影响；FACETS分析结果显示，个别任务与整体模型不拟合，一位评分员打分一致性不好。这些结果为修改考试内容和完善评，分过程提供了非常重要的参考信息。Taking Generalizability Theory and Many-facet Rasch Model as its theoretical basis, this study investigates the reliability and validity of a trial test results of a speaking test for prospective secondary school English teachers by using GENOVA and FACETS. The results of the GENOVA analysis showed that the dependability of the test would not change much if fewer tasks were included in the test, whereas the reliability of the test would have a significant change if fewer raters were employed. The results of the FACETS analysis revealed that some tasks were misfitting with the model and one rater was not consistent in her marking. These results provide useful complementary information for the improvement of the test content and the rating process.

关 键 词：概化理论 多层面RASCH模型 口语考试 

分 类 号：H319[语言文字—英语]