Replication. Part I

Abstract

There has been considerable debate and analysis about the nature and extent of what many believe to be a replication crisis in psychology and other social sci­ences. And perhaps to a lesser degree in the medical sciences. Our approach to the nature and value of replication has been based on the idea that the overriding purpose of replication is to ferret out and identify confounding causes. In other words, to reduce systematic uncertainty. This has led us to understand replication in a broader sense than ordinarily understood. We have developed this approach in considerable detail in three recent books which include both abstract analysis and many case studies drawn primarily, but not exclusively, from physics. We have, for example, considered the difficulty of deciding whether a replication has been successful or has failed, the roles of null experiments, and episodes in which a single experiment has been sufficient to decide, or to further investigate, an issue. In this two-part essay we will review and summarize our approach and results.