A checksum helps identify the exact file used. It does not establish who originally created it, whether reuse is permitted or whether its values are correct. Record provenance uncertainty beside the identifier so it cannot disappear from the final report.
Distinguish identity from validity
A checksum helps identify the exact file used. It does not establish who originally created it, whether reuse is permitted or whether its values are correct. Record provenance uncertainty beside the identifier so it cannot disappear from the final report.
Record the transformation chain
Link source, working and analysis versions. For each transformation retain the rule, reason, author and time. An exclusion needs an eligibility rationale; a corrected code needs evidence. Keep an unresolved original value rather than silently replacing it with a plausible guess.
Worked example: category cleaning
Illustrative example: two text labels appear to represent the same category. The analyst proposes a harmonised code, records the original strings and consults the instrument. If the codebook is unavailable, the ambiguity remains open. A technically consistent recode would still lack a justified meaning.
Original manifest template
Use: Input ID | Source provenance | Permission status | Version/checksum | Transformation/run ID | Output ID | Decision log | Reviewer disposition. For unavailable inputs, state the limitation and its effect on reproducibility rather than pretending that a script alone completes the evidence.
Hand over reproducibility evidence
Include a README describing file relationships, execution order, dependencies and restricted material. A recipient should know what they can rerun and what remains controlled. Statistical validity and independent review still need their own evidence.
Review this research decision
We can review your actual form, dataset and reporting requirement, then scope the smallest service that solves the risk.
Explore the service ↗