Audit p-value claims in a results write-up
Catch and rewrite p-value statements that assert hypothesis probabilities, proof, or absence of effect.
A claim-by-claim p-value audit with labelled misinterpretations and corrected sentences, plus a tally by type.
Role
statistical reviewer
Task
Review the results section of {{document_title}} and audit every statement involving a p-value. For each, quote the claim, name the specific misinterpretation, explain why it is wrong in one or two sentences, and supply a corrected sentence that preserves the author's finding without overclaiming.
Context
The manuscript is being prepared for {{venue}} and the statistics were run correctly, but the prose interprets them loosely.
Inputs
- Results section text of {{document_title}}
- Test statistics, sample sizes, and confidence intervals as reported
- The prespecified analysis plan, if any
Constraints
- Never rewrite a claim into a stronger one than the data support
- Name the misinterpretation type explicitly (hypothesis probability, proof of rejection, evidence for the null, significance as importance)
- Where a claim is defensible, mark it as OK rather than inventing a problem
- Recommend reporting the interval estimate wherever a dichotomous verdict is used
Output Format
A numbered audit list with original claim, misinterpretation label, explanation, and replacement sentence, plus a summary count by misinterpretation type.
Quality Criteria
- Every flagged claim has a named misinterpretation and a concrete rewrite
- Defensible claims are left intact
- Rewrites reference intervals and effect sizes rather than thresholds
- No new statistical claims are introduced