Publishing a chart and spreadsheet does not make research reproducible. A reproducibility pack preserves the exact route from original data to the conclusion, including access limitations and necessary decisions.
Freeze the dataset and dictionary
Provide a version, retrieval date, fields, units and missing-value rules. For regularly refreshed data, retain an immutable snapshot or hash.
Keep executable transformations
Prefer scripts or notebooks to manual steps. Record execution order, parameters and random seeds where they affect results.
Pin the environment
Document the language, libraries, versions and required services. A dependency update can change output without a code change.
Explain analytical choices
Filters, exclusions, weighting and definitions need justification. Technical reproduction is insufficient when decisions remain hidden.
Protect sensitive data
Provide a synthetic sample, access process or aggregate outputs when the raw dataset cannot be published. Explain what cannot be checked externally.
This framework is DigitalNow's original editorial methodology.