论文部分内容阅读
近些年多层面Rasch模型(MFRM)和结构方程模型(SEM)在语言研究中的应用越来越广泛,但是却很少有研究将这两种分析方法结合起来研究语言考试的效度。本研究采用MFRM和SEM相结合的方法研究了一项口语考试分析性评分量表的构念效度。MFRM和SEM可以从不同的角度为评分量表的构念效度提供证据。MFRM的分析结果表明,评分员的评分信度和评分量表的质量均比较理想;SEM分析的结果表明,高阶能力因子模型与数据的拟合情况理想。本研究的相关结果为评分量表的构念效度提供了重要支撑,同时本研究所采用的MFRM和SEM相结合的方法也为表现型测试评分量表的效度研究提供了重要参考。
In recent years, multi-level Rasch model (MFRM) and structural equation model (SEM) are increasingly used in linguistic research, but few studies have combined the two methods to study the validity of language tests. In this study, we used the combination of MFRM and SEM to study the constructive validity of an oral test analysis scale. MFRM and SEM can provide evidence for the validity of the rating scale from different perspectives. The results of MFRM show that the score reliability of the scorers and the quality of the rating scale are ideal. The results of the SEM analysis show that the fit of the high-level capability factor model and the data is ideal. The results of this study provide important support for the constructive validity of the scoring scale. At the same time, the combination of MFRM and SEM used in this study provides an important reference for the validity of the phenotype test scoring scale.