Using linked education administrative data to evaluate and extend traditional studies of school effects on student outcomes

学校效能的研究人员估计个别学校的影响，作为第一步，以确定学校的政策和做法，这可能解释为什么有些学校比其他学校更有效。教育部估计个别学校的影响，以追究学校的责任（进展8），通知学校检查（英国教育标准局），通知家长选择学校（compare-school-performance.service.gov.uk），方法标准方法是使用NPD将学校的“增值”固定或随机效应线性回归模型拟合到学生的GCSE考试成绩，其中学生的KS 2考试成绩和社会人口统计学（年龄，性别，种族，语言，FSM，SEN，IDACI）作为协变量输入，试图调整由此产生的预测学校效应，以反映入学时学生群体的学校差异，否则会使学校比较产生偏差。第一波GUIE数据（2014/15年队列）中可用的社区特征，以扩展标准方法，对入学时学生群体的学校差异进行更丰富的调整。我们将研究这些额外调整的重要性及其对预测的学校效果的影响，并在此过程中评估标准方法是否足以解决学生非随机分配到学校的问题。主题2我们将使用学生的就业和收入1，3，毕业后5年和10年，LEO可扩展标准方法，以探索学校对学生长期劳动力市场结果的差异。具体来说，我们将改变标准方法中的因变量，以适应不同的劳动力市场结果，我们将探索有和没有GCSE分数作为协变量的模型。我们将研究由此产生的预测学校的影响，以及它们如何在不同的结果，以突出在学校系统的不平等变化。

CONTEXTSchool effectiveness researchers estimate individual school effects as a first step to identifying school policies and practices which might explain why some schools are more effective than others. The DfE estimates individual school effects to hold schools to account (Progress 8), to inform school inspections (Ofsted), to inform parents choosing schools (compare-school-performance.service.gov.uk), and to provide schools with information for self-improvement.APPROACHThe standard approach is to use the NPD to fit school 'value-added' fixed or random effect linear regression models to student GCSE examination scores where student KS2 test scores and sociodemographics (age, gender, ethnicity, language, FSM, SEN, IDACI) are entered as covariates to attempt to adjust the resulting predicted school effects for school differences in the student body at intake which would otherwise bias the school comparisons.TOPIC 1We will use the Census 2011 student, household, and neighbourhood characteristics available in the Wave 1 GUIE data (2014/15 cohort) to extend the standard approach to make far richer adjustments for school differences in the student body at intake. We will study the importance of these additional adjustments and their impact on the predicted school effects and in doing so we will assess the adequacy or not of the standard approach to address the non-random assignment of students to schools.TOPIC 2We will use students' employment and earnings 1, 3, 5 and 10 years after graduation available in LEO to extend the standard approach to explore school differences on students' longer-term labour market outcomes. Specifically, we will change the dependent variable in the standard approach to different labour market outcomes and we will explore models with and without GCSE score entered as a covariate. We will study the resulting predicted school effects and how they change across the different outcomes to highlight the inequalities in the school system.