Replication in education science

Abstract

Since experiments in education often sample from different populations and treatment fidelity can fluctuate, it would seem inevitable that study results of even successful replications might vary slightly. Future replication attempts in education science require practical definitions of replication that incorporate notions of ‘approximate’ replication. We demonstrate how we might operationalize this definition by: 1. defining how we might quantify studies getting ‘almost the same’ results, 2.using Bayesian estimation to assess whether studies approximately replicate, and 3.using this method on an example from the ManyLabs replication project.

Jacob M. Schauer

Assistant Professor

My research interests involve statistical methods for the social and health sciences.