Measuring school value added with administrative data: The problem of missing variables

Abstract

The UK Department for Education (DfE) calculates contextual value added (CVA) measures of school performance using administrative data that contain only a limited set of explanatory variables. Differences in schools' intakes regarding characteristics such as mother's education are not accounted for due to the lack of background information in the data. In this paper, we use linked survey and administrative data to assess the potential biases that missing control variables cause in the calculation of CVA measures of school performance. We find that ignoring the effect of mother's education leads the DfE to erroneously overpenalise low-achieving schools that have a greater proportion of mothers with low qualifications and to over-reward high-achieving schools that have a greater proportion of mothers with higher qualifications. This suggests that collecting a rich set of controls in administrative records is necessary for producing reliable CVA measures of school performance. © 2011 The Authors Fiscal Studies, © 2011 Institute for Fiscal Studies.