Deliver a reproducible analysis package
Connect a frozen dataset and code revision to tested calculations, readable instructions and the claims in a manuscript. Test the handoff with a new reader and keep restricted data within its approved access conditions.
Worked decision
Hypothetical decision: A manuscript reports 420 observations, but the frozen output contains 408. The twelve-record discrepancy triggers a hold on release. The analyst traces the claim to its generating script and finds that the manuscript used an earlier data snapshot. The team reruns the approved pipeline, reconciles every affected table and statement, and records the corrected snapshot, code revision and review evidence before handoff.
Record your decision
Can another authorised analyst reconstruct the reported result from the documented inputs? Resolve missing dependencies, unexplained numerical differences and unsupported claims before release. Byte-identical files help establish identity but do not prove scientific correctness. Deposit only permitted materials and state which parts require controlled access.
Work through the evidence
- Freeze the exact analysis dataset and derivation (TXT)
- Prepare a complete analysis handoff
- Define independent expected calculation results (TXT)
- Compare digests of synthetic text locally in the browser
- Test repository instructions with a new reader (TXT)
- Trace manuscript claims to frozen outputs (TXT)
- Review the original secondary-data reporting handoff checklist (TXT)
- Consult the open handbook on reproducible research