IDEA FOUND // IDEA 119
Replication and Reproducibility
“Recomputing the same files and repeating a study with new data test different things.”
01 / PLAINLY
What it means, plainly
Reproducing means obtaining the result again with the same data and computational steps. Replicating means addressing the same question with new data.
02 / CONTEXT
A little more
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 / WHY IT MATTERS
Why it matters
The distinction separates checking the original computation from asking whether compatible results appear when the same question is studied with new data.
04 / EXAMPLE
A familiar example
Running published code tests reproducibility. Recruiting another sample and repeating the protocol tests replicability.
05 / LIMIT
What it does not mean
A differing replication does not automatically prove fraud or single-handedly refute the entire original idea.
06 / NOTICE
Notice it in your day
Read a description of a repeated study and underline whether it reused the original data or collected new data. Classify it from that detail.
FINAL NOTE
The idea worth keeping
Ask exactly what was repeated: computation, measurement, sample, or question.
QUESTIONS / 02
Questions people still have
What stays the same in a reproducibility check?
The original data and the computational steps used to obtain the result stay the same.
What changes in a replication?
New data are used to address the same question.