FILE SELECTED // FILE 00
Replication and Reproducibility
“Recomputing the same files and repeating a study with new data test different things.”
01 / SUMMARY
In short
Under the National Academies terminology, reproducibility means consistent results from the same data and computational steps; replicability means addressing the same question with new data.
03 / SCENE
In the wild
Running published code tests reproducibility. Recruiting another sample and repeating the protocol tests replicability.
04 / LIMIT
The limit
A differing replication does not automatically prove fraud or single-handedly refute the entire original idea.
FINAL NOTE
What remains
Ask exactly what was repeated: computation, measurement, sample, or question.