Estimating effects of school quality using multiple proxies
利用ECLS-K数据和代理变量模型，研究发现将班级规模、教师学历等作为学校质量的噪声度量时，能检测到学校质量对学生成绩的显著正向影响，而传统方法低估了约50%的效果。
The recent literature on school quality has shown that the school a child attends has significant effects on achievement. However, the literature relating different school characteristics to student achievement has produced mixed results, particularly when using student-level data. Using data from the ECLS-K and a proxy variable model that addresses the problem of measuring school quality, we show that significant effects of teaching and resource quality can be detected from student-level data. We find a significant, positive relationship between school quality and student achievement if school characteristics such as class size and teachers' schooling are treated as noisy measures of school quality. However, this effect is not detected when using models which do not account for measurement error in school quality. Our results suggest that conventional approaches underestimate the effect of school quality by about 50%.