论文部分内容阅读
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。
Based on generalization theory and multi-level Rasch model, this study analyzes the test results of a spoken English examination model for pre-service middle school teachers with GENOVA and FACETS to explore the reliability and validity of the test, To improve the test design to provide the basis. The analysis results of the two methods provide us with complementary information: The results of GENOVA analysis show that reducing test tasks properly will not have a great impact on the reliability of the test, however, reducing the number of scorers will have a greater impact on the reliability The result of FACETS analysis shows that the individual task does not fit with the overall model and the scoring consistency of a scorer is not good. These results provide very important reference information for modifying the exam content and improving the scoring process.