Keep an unchanged original
Preserve the original data in an appropriately controlled location and work on documented copies. Use stable file names, dates and versions. Separate participant identifiers from analytical files when possible and avoid including unnecessary personal information in shared material.
Write a data dictionary
Describe variable names, labels, units, category codes, missing-value conventions and derived measures. Record the unit of observation and how records relate across files. A spreadsheet column called “score” is insufficient without the instrument, range and scoring procedure.
Record processing and analysis
Keep scripts or a step log for cleaning, exclusions, transformations and models. Note software versions and relevant settings. Link each final table or figure to its source output so that a later revision can be checked consistently.
Prepare a controlled handover
Include a readme explaining files, dependencies and how to reproduce the reported outputs. State which material can be shared and which is restricted. Check the journal’s data and code policy before submission rather than assuming all material can be uploaded publicly.
Organize supplementary files for a reader
Use a contents list explaining each file, its format and its relationship to the article. Provide a data dictionary or readme where needed, remove unintended identifying details and check that every file opens. Reconcile labels and version numbers with manuscript citations. Follow the journal’s repository, file-size, licensing and access requirements; do not upload restricted data merely because it is supplementary material.