We are seeking feedback on a draft proposal for establishing policies and procedures on error correction. This is the first project of a committee charged with the task of creating policies and procedures on publication ethics at Peaceful Science. The committee’s current membership is @cwhenderson, @Jordan, and @sfmatheson.
To provide suggestions and critiques, or to ask questions, please respond here in the forum. Note that the conversation will be carefully moderated to stay on topic (so that the committee gets clear feedback), and note that while the proposal is focused on a pretty specific topic, an “unanswered questions” section points to future projects and topics on which we welcome input. If you would prefer to provide comments privately, that’s fine–use the private messaging facility here or feel free to contact Stephen Matheson by email (which you can find at sfmatheson.com).
Proposed Policy and Process for Error Correction
Pre-publication processes can mitigate errors, and those processes are part of a separate set of principles and standards. But errors will happen, and these basic principles apply whenever an error is noted. Peaceful Science expects and welcomes error correction.
Error correction will be comprehensive, transparent, and public. Even the correction of a typo or misspelling will be noted. Correction mechanisms will range from editing in place (with full disclosure) to retraction.
Categories of error
There are two overall categories of error that can happen in an article.
Minor errors include typos, misspellings, inaccuracies in citations or attribution, grammatical or format-related errors that hinder clarity or obscure the author’s intended meaning. Correcting a minor mistake leaves the intended meaning unchanged.
Errors of fact, whether they seem major or minor, are almost always a separate category from minor errors. Correcting an error of fact may not change the overall message of an article, or affect its conclusions, but the correction changes the meaning of at least a portion of the article.
What is not an error
In the context of ongoing scholarship (science, specifically), articles regularly become “wrong” when new discoveries happen. This is why science is often correctly described as “self-correcting,” and the scientific literature has no other mechanism for correction of this kind of wrongness. The policies in this proposal do not address this kind of ongoing scientific self-correction, but future proposals will.
Processes for correcting errors
Minor errors are corrected in place. This means that the article is edited, and the original version containing the error is replaced. All changes are disclosed in a note attached to the article, and the changes are specifically identified.
Errors of fact result in either correction via a corrigendum, or retraction.
When correction of errors in an article leaves the article’s main points and conclusions intact, a corrigendum is a reasonable process (more details below). There are at least two related factors that can result in the more dramatic step of retraction, and both involve editorial judgment. The first factor is the extent to which the error undermines the article. If an author’s main points rely significantly on false claims, then correction of the error renders the article untenable and retraction is a likely outcome. The second factor is misconduct. If an article is tainted by scholarly misconduct (gross misrepresentation of facts or other scholarship, selective citation, quotemining and other misrepresentation of the words or work of others, and so on), retraction is a likely and reasonable outcome.
A corrigendum is a separate article (or an addendum to the original article) that identifies the error, its source (either the author or, in rare cases, the editor), and the steps taken to correct the error. The corrigendum points to the now-corrected article, and the article points to the corrigendum. The original article can be retained and linked in the corrigendum. If this is done, the original article must be very clearly marked to alert readers to the fact that the article has been corrected.
A retraction notice is attached to the original article and specifies the reasons why the article is being retracted and the parties who have agreed to its retraction. Retracted articles should remain accessible to readers, in a form that very clearly indicates that the article has been retracted and why.
This draft document is fully open for discussion. Some very important questions are not addressed, but must be settled (through discussion) before the policy is complete.
Error correction requires oversight and judgment from a person or committee, which is not addressed in this draft.
Editorial judgment regarding type of error (minor error or error of fact) and scholarly conduct requires standards and criteria, which are not specified in this draft.
Pre-publication review is not discussed in this draft.
Retention of articles after correction, through a system of versioning, is hinted at here but not discussed in detail.
Retraction due to misconduct is likely not workable without explicit agreement by authors to be subject to editorial judgment. This draft says nothing about how to navigate that.
Correction of wrongness that results from ongoing discovery is not addressed in this draft. One suggestion is to create a set of “living documents” that comprise summaries or discussions of scientific topics, organized so that the reader can see the development of ideas in time without the “correction” or removal of articles. Another suggestion is to annotate articles with pertinent updates, similar to the process for annotating articles with corrections.
COPE (the Committee on Publication Ethics) sets standards for publication practice. The first reference is a PDF summary document of all aspects of publication practice and covers all the topics in our draft. The second is specifically about retraction.