摘要
"选考自由"是我国高考改革史上的一项重大尝试,但目前的学业水平选考科目赋分方案导致了很多学生的等级分数被系统低估或高估,进而引发了物理等较难学科选考人数大幅度下降的后果,对基础教育和高校招生带来了比较大的负面影响。在考试行业常用的"标准设定"和"测验等值"方法使用条件不成熟的情况下,建议使用基于大数据代表性样本的等级赋分方案。本文所报告的是该方案的大数据模拟研究证据。
Providing the choice of selecting multiple subject tests instead of taking them all shows great progress in reforming the college entrance examination system in China. The current scaling method used for selective subject tests,however,has a strong negative impact on both compulsory education and college admission.This is because the rank scores of medium-competency students are systematically significantly underestimated when their competitors are extremely strong performers,which causes many students to give up subjects perceived as generally more difficult,such as physics,during high school. This scaling issue can be fixed by the methods of"standard setting"and"test equating"used in the testing industry. Unfortunately,this psychometric solution cannot be implemented in China due to concerns over test security and the fact that most tests in China have many openended items. Therefore,we propose a new scaling method that uses representative samples to determine rank scores for selective subject proficiency tests. The big data analysis results indicate that this new method can solve the essential problems in the current scaling method without big changes in the current college admission policy or much additional work,which saves cost in administration.
出处
《教育测量与评价》
2019年第1期3-10,共8页
Educational Measurement and Evaluation
关键词
学业水平考试
赋分方法
大数据分析
subject proficiency tests
scaling method
big data analysis
