Causal effects of educational between-school selection mechanisms: A multilevel meta-analysis

Deliverable 2.1 is synthesising evidence on three selection mechanisms: school choice by random allocation of students, formal (early) and elite tracking – focusing on both average effects and heterogeneity for low- and high-performing students' outcomes. By concentrating only on causal studies which apply robust experimental or quasi-experimental evidence and using multilevel random-effects meta-regression with cluster-robust standard errors, we avoid the confounding biases of a limited number of datasets applied or multiple outcomes from the same author. Across early tracking systems, mean effects are often reported as negative, and for the random allocation mechanism and elite tracking, positive. Our results show that grand mean effects are not entirely consistent with the state of the art. In the case of elite tracking, we can report only local treatment effects on treated across the cut-off, which is a small positive. In the case of early tracking, the effect is small and negative, but the distributional patterns of the effects show that low and high-achieving students benefit from this selective mechanism. In case of random assignment, the grand mean effect is zero, but top-performers are hurt by lotteries. So, the applications of selective school admission mechanisms always contain political choices – who to grant preferable treatment.