Larry Hedges will discuss his belief that using an ensemble of studies to evaluate replication in research findings is simplistic and its methodology understudied. He asks whether replication requires underlying effects to be identical, or if some variation in true effects should be allowed. He will explore using different possible analyses -- analyses that have different decision theoretic properties -- to show that a single replication study is almost never adequate to determine whether a research finding replicates.

This lecture is part of WCER's Interdisciplinary Training Program in Education Sciences.