Original Research
Part of PR response and issue handling
Correcting an error in a published company study
Trace a study error through its results, headline and distributed materials, then publish and notify a clear correction.
If a published company study contains an error, identify the result and conclusion it affects. Correct the study, its dependent materials and the account given to earlier recipients. Make the change visible so readers can tell which finding is current.
Confirm the error and its reach
Preserve the published version under question. Ask the research owner to reproduce the affected result from the source data and calculation. Check whether the problem is a label, input, filter, denominator, calculation or inference. Keep a suspected error labelled as suspected until the check is complete.
Trace every claim that depends on the result. A denominator error may affect several percentages, a chart and the headline. A wrong label may leave values intact while changing what readers think they describe. Establish whether the correction changes the conclusion; do not assume it does or does not.
Keep original and corrected values, method versions and review decisions together. If a sound replacement cannot yet be calculated, mark or withdraw the affected finding while checking continues. Do not leave a known unreliable headline in current distribution.
Explain the correction
A useful notice identifies the earlier statement, what was wrong, the corrected position and whether the interpretation changes. Date it and identify the study version it covers. Give readers with a saved copy enough detail to recognise the change.
Error found / Check before release
- Wrong label or period
- Whether the text and chart describe the data correctly.
- Wrong value or base
- Every calculation, comparison and visual that uses it.
- Unsupported inference
- Whether the headline and summary must be narrowed or withdrawn.
- Unresolved underlying data
- Whether the affected finding must remain on hold.
Do not silently replace the file. Describe genuinely new information as an update rather than implying the earlier study was wrong.
Repair the release package
List the outputs that use the result: the study page, download, summary, table, chart, media release, fact sheet, spokesperson wording and short posts. Check each relevant item against the corrected calculation. Put the current version and correction notice where readers encounter the result.
Date or version revised files. Keep an internal record detailed enough to repeat the corrected calculation. If privacy or commercial restrictions prevent release of raw data, explain the method and what can be shared without suggesting the full dataset is public.
Tell recipients of the old result
Use the distribution record to identify known recipients of the affected version. Tell them the old result, the corrected position, which files changed and whether the headline conclusion still holds. Ask them to use the current version. An independent publication controls its own article; provide the corrected evidence rather than assuming its page will change automatically.
Check later enquiries and coverage for continued use of the old figure. If the company supplied a wrong figure that a publication repeated, say so plainly. Record notifications and replies; silence does not show that every copy was fixed.
The correction process is ready for review when the current study and its dependent materials agree, the change is visible, and known recipients have a usable account of it. If the error invalidates the central finding, say that instead of offering an untested replacement.
