How to verify the numbers in a published paper — the methods, their limits, and the discipline of separating proof from suspicion.
Why statistical-forensics tools must separate proof from suspicion, what the base rates actually look like, and how to raise an inconsistency with authors without accusing them.
What each statistical-forensics method actually proves, what it needs as input, where it fails, and which combination is worth running on a given paper.
A practical checklist for reviewers, editors and researchers — the five checks that catch most reporting errors, what each one requires, and how long they take.
How a single line of arithmetic can prove that a reported mean is impossible — with a worked example, the exact conditions under which it applies, and where it goes wrong.
Is a reported mean possible for its sample size?
Is a reported standard deviation possible?
Statlint runs statcheck, GRIM, GRIMMER, SPRITE bounds, descriptive recomputation, digit analysis and the Carlisle baseline test in a single pass — free, no signup.
Analyze a paper →