高校自主招生选拔中校本成绩校验的模型及成效比较

A Comparative Study of Statistical Models for Moderating School-based Exam Scores in College Admissions

作为一种过程性评价的结果，由学校或地区开展的校本考试成绩更能反映学生能力和学习情况，理应在学生评价中发挥更为重要的作用。然而，校本成绩的分散性增加了对其进行客观比较的难度，限制了其在大规模考试选拔中的广泛应用。本文采用了回归和多层线性模型，探讨了利用统一考试成绩来校验校本考试成绩的可行性，比较了两种统计分析模型的成效。结果表明，多层线性模型的分析结果更符合实际情况，效果相对较好，回归分析操作和解释更为简单，在大规模考试中应用具有一定的优势。

Scores on school-based exams，i． e． ，formative assessment developed by a school or district，can more effectively reflect students’ability and learning process than summative assessment results and therefore should receive more emphasis in student assessment． However，because school-based exams are conducted locally， exam scores may not be comparable across schools，limiting the application of these exams in large-scale exam-based selection． This study compares the effectiveness of multiple regression and hierarchical linear modeling in moderating school-based exam scores and finds hierarchical linear modeling is more practical and accurate while the multiple regression model is simpler and easier to conduct for large scale exams．