Direct comparison
Retraction vs. Correction: Key Difference
A retraction withdraws a paper's conclusions; a correction fixes one error while conclusions stay valid. COPE's deciding test, with a comparison table.
Side-by-side comparison
| Dimension | Retraction | Correction (corrigendum / erratum) |
|---|---|---|
| What triggers it | Fabricated/falsified data, plagiarism, duplicate publication, unethical research, or an error so serious it invalidates the paper's main conclusions | An isolated, identifiable error (data point, figure, author list, methods detail, production mistake) that does not undermine overall validity |
| Effect on the paper's conclusions | Conclusions are no longer considered reliable; the paper should not be relied on or cited as valid evidence | Conclusions remain valid once the specific error is fixed |
| Status of the original article | Stays accessible online but is prominently marked 'RETRACTED' and linked to the retraction notice — not deleted | Remains the version of record; the correction notice is linked bidirectionally to it in Crossref/CrossMark metadata |
| Notice published | Retraction notice | Corrigendum (author-originated) or erratum (publisher-originated) — many journals use the two interchangeably |
| Who typically initiates it | Editor, publisher, institution, or the authors themselves — often after a formal investigation | Usually the authors; sometimes the editor or publisher for production-side errors |
| Governing guidance | COPE Retraction Guidelines v2 (Nov 2019, DOI 10.24318/cope.2019.1.4) | Same COPE framework; applied via each journal's own correction policy |
| Typical timeline | Weeks to years — often follows a formal institutional investigation | Days to weeks once the error is confirmed |
| Deciding question editors ask | Would a reader draw materially different conclusions once the error is accounted for? — Yes | Would a reader draw materially different conclusions once the error is accounted for? — No |
| Effect on citing the work | Citation should stop; retraction status is flagged in Crossref metadata and the Retraction Watch Database | Citation continues normally; readers are directed to read the correction alongside the original |
Common questions
FAQ
What's the difference between a corrigendum and an erratum?+
By convention (not a universally enforced rule), a corrigendum corrects an error the authors introduced in their own submitted work, while an erratum corrects a production-side error the publisher introduced, such as a typesetting or figure-reproduction mistake. Many journals use the two terms loosely or interchangeably, or collapse both under a single 'correction' notice type.
Who decides whether an error needs a retraction or just a correction?+
The journal editor decides, guided by COPE's Retraction Guidelines and sometimes an institutional investigation. The deciding factor is always the same: whether the error undermines the paper's overall validity and conclusions — not how large, numerous, or embarrassing the error is.
Can a paper that was already corrected later be retracted?+
Yes. If a correction turns out to be insufficient — the underlying problem is more serious than first understood, or new evidence emerges — a journal can still retract a paper that previously received only a correction.
Does a retracted paper get removed from the journal's website?+
No. COPE guidance requires the retracted article to remain accessible, clearly and permanently marked as retracted, with the retraction notice linked. Retraction corrects the scientific record; it is not intended as deletion or punishment.
Is an expression of concern the same thing as a retraction?+
No. An expression of concern is an interim notice published while an issue is still being investigated, before the editor has enough evidence to decide between retraction, correction, or no further action.
