A Replication of a Quasi-Experimental Approach to Estimating Middle School Structural Transition Effects on Student Learning Trajectories

Abstract

Using a dataset that includes over 17 million students from across all 50 states, we estimate the causal impact of making structural transitions into middle school (in grades 4, 5, 6, or 7) on student math and reading achievement trajectories. This dataset provides an ideal opportunity to engage in the valuable scientific practice of conducting replication studies. Prior research on the impacts of middle school transitions is of high quality and rests on a strong causal warrant, but the study settings vary greatly and use data from a prior decade. We conduct a replication (i.e., using the same methods on different data) using larger, broader, and more recent data. We extend prior analyses in ways that may further strengthen the causal warrant. Finally, we explore heterogeneity of effects across subgroups and states, which may help reconcile differences in the magnitude of estimated effects across studies.