教育增值评价中嵌套数据增长百分位估计方法探析:多水平线性分位数回归模型的应用

Exploration of Growth Percentile Estimation Method for Nested Structural Data in Educational Value Added Evaluation: Application of Multilevel Linear Quantile Model

本研究重点探讨教育增值评价中,嵌套数据下学生及群体层面增长百分位的估计方法——多水平线性分位数回归模型,并基于实测数据,说明该估计方法在教育增值评价中的应用。多水平线性分位数回归估计方法可以充分考虑群体间差异,对学生能力进行合理预测,进而更准确地估计群体增长百分位,使得对学校、教师等教育效能的评价更加准确;还可以在对其他背景影响因素(包括个体层面因素和群体层面因素)进行合理分析的基础上,得到学生层面和群体层面(如学校、班级、教师)的增长百分位估计。

This study focuses on the educational value-added assessment of percentile growth at the student and teacher levels under nested structural data—multilevel linear quantile regression model, and illustrates the application of the estimation method in the value-added assessment based on the measured data. This model can fully consider the differences between groups and predict reasonably the students ability, making the estimation of group growth percentile more accurate. Thus, the educational effectiveness evaluation of schools and teachers will be more precise. This model can also estimate the growth percentile in the level of students and various groups, such as school, class and teacher based on the reasonable analysis of other background factors, including individual and group factors.