基于多层面Rasch模型的多体裁自选任务写作测试评分效度研究  

On the scoring validity of self-chosen-task writing test by many-faceted Rasch model

作　　者：邵健[1] Shao Jian

机构地区：[1]浙江工商职业技术学院国际交流学院,浙江宁波315012

出　　处：《西安外国语大学学报》2021年第3期72-77,共6页Journal of Xi’an International Studies University

摘　　要：写作体裁是重要的文本特征之一,但体裁在写作测试中的研究仍显不足。本文基于一项多体裁自选任务写作测试的数据,运用多层面Rasch模型,从考生、评分员、任务和评分量表4个层面分析了该测试的评分效度,结果发现:不同测试任务之间不存在显著差异,彼此之间具有等效性,能够有效区分不同水平考生;评分员之间一致性较好,内部稳定性理想;评分员能够运用评分量表对不同体裁作文进行有效评价,对学生的能力区分稳定。总体而言,多体裁自选任务写作测试具有较为理想的效度。The genre is a key characteristic of the text,whose role in writing test is not well investigated in the testing field.The paper analyzes the scoring validity of a self-chosen-task writing test by means of MFRM,in which the facets of testees,raters,tasks and scale are probed into.The results reveal that the tasks are not significantly different and are equal in judging students with different language proficiency.The inter-rater or intra-rater consistency is ideal and the raters’judging procedure is consistent and stable in differentiating students.In a word,the MFRM outputs indices support the self-consistency in the very writing test and proves its validity.

关 键 词：体裁 自选任务写作测试 评分效度 多层面RASCH模型 

分 类 号：H319[语言文字—英语]