Estimating the causal effects of academisation of English schools with the data from the National Pupil Database

Abstract

Background Over the last 30 years there have been two key policy initiatives in schooling in England. First, there was a move to raise academic standards with a market-oriented model of schooling and the introduction of Academies with higher levels of autonomy. Second, there has been a lesser though nevertheless important trend to providing for more pupils with special educational needs (SEN) in ordinary rather than special schools, what has been called the inclusive education movement. Objectives We explore causal effects of converting English schools into academies on educational inclusiveness in England. Specifically we investigate the implications of a school’s academisation event in converting to a Sponsored Academy or a Converter Academy on the inclusion of pupils with SEN and the impacts on the education trajectory of these pupils. Methods (including data) Using administrative data from the National Pupil Database and combining it with publicly available data on school institutional history, we are able to construct extensive longitudinal samples of pupil cohorts and academisation school cohorts from 2003 to 2015. Utilising both from aggregate school level and individual level features, we study the various aspects of academisation on educational inclusiveness by modelling the admissions, re-classifications and potential exclusions on pupils with SEN from the contemporaneous and follow-up effects of academisation events. In addition, we examine the adequacy and effectiveness in causality modelling in a educational context under various regimes of modelling frameworks. Findings and conclusions The analysis to date suggests that academisation events have small negative impacts in educational inclusiveness. Specifically, we find that pupils with SEN are more likely to register with non-SEN in academised schools in the treatment group. In addition, these negative effects are found to remain statistically significant several periods after occurrence of the event.