摘要
为验证并提高PETS五级听力构建题评卷信度,本研究随机抽取考生答卷进行试评,修订标准答案,采集答案修订前后考生答题数据进行分析,并以问卷形式等就如何提高构建题的评卷信度进行研究。研究结果表明,有必要在大规模评卷之前增加一次试评,根据试评中出现的具体情况充实、完善标准答案。为了提高构建题评分信度减少评卷人的主观判断,须加强评卷人培训,严格按照修订后的评卷要求和标准答案评卷并坚持复评制度。
In order to verify and improve marking reliability of constructed-response questions in comprehension of PETS 5,a few hundred randomly selected answer sheets were pre-marked and the listening marking scheme was modified according to the pre-marking. And then the selected answer sheets were marked again with the modified scheme. The results of the pre-marking and remarking were analyzed, and the markers took part in the discussion after pre-marking and filled in questionnaires. marking before actual marking improve marking reliability and The study shows that it is necessary to have a preand have the marking scheme modified according to the pre-marking. In order to reduce markers' subjectivity and flexibility, markers must be trained always mark in accordance with the requirements and the modified marking scheme. Re-marking important approach to maintain marking reliability.
出处
《中国考试》
2011年第8期23-28,共6页
journal of China Examinations
关键词
构建题
评卷信度
试评
Constructed-response Questions
Marking Reliability
Pre-marking and should is another