Diagnosing scientific replicability through probabilistic distinguishability.

AbstractBecause result replicability is important to researchers, and statistical significance does not evaluate replicability, both external and internal replicability analyses are important. Some researchers find true external replication studies impractical, and therefore turn to ‘internal’ methods. Three internal replicability analyses have been proposed: cross‐validation, the jackknife, and the bootstrap.