Earlier today, we reported on the withdrawal of a paper from Research Policy, an Elsevier journal. The notice didn’t give a reason, just that the “article has been withdrawn at the request of the authors and editor.”
We’ve seen a number of such opaque withdrawals from Elsevier journals, and thought it was worth some exploration. While Elsevier’s policy here on such withdrawals is clear, as it is in other matters, we take some issue with it:
Articles in Press (articles that have been accepted for publication but which have not been formally published and will not yet have the complete volume/issue/page information) that include errors, or are discovered to be accidental duplicates of other published article(s), or are determined to violate our journal publishing ethics guidelines in the view of the editors, may be “Withdrawn” from ScienceDirect. Withdrawn means that the article content (HTML and PDF) is removed and replaced with a HTML page and PDF simply stating that the article has been withdrawn according to the Elsevier Policy on Article in Press Withdrawal with a link to the current policy document.
The policy gives researchers — and journals — a mulligan. It seems to reflect an old-fashioned view of what “publishing” means, namely that the paper has appeared in a print issue. Sorry, folks, publishing online means something is, well, published. We see no reason why the scientific community does not deserve as full an accounting of such withdrawals as it gets from Elsevier’s retraction notices — which, as we have noted before, generally contain more information that the industry average.
We asked Elsevier for an explanation of their rationale. Spokesperson Tom Reller responded with details about
why we make an explicit distinction between ‘articles in press’ and ‘published articles’, and why the procedures for withdrawing the former are different from those for retracting the latter (particularly the amount of information provided in the notice).
In regard to the explicit distinction between ‘articles in press’ and ‘published articles’, we can see how it may appear as old-fashioned, but it’s still practical in many cases. The most common reasons for AIPs being withdrawn are just very simple, administrative errors such as the same paper accidentally being posted twice or the author accidentally submitting the revised version as a new paper and, when accepted, the wrong version being sent for typesetting. Our approach, and the approach with many, (if not most) publishers, is that it’s more efficient to correct this category differently than fully formally published articles.
I suspect some may disagree with the approach, but we think it’s reasonable for all the parties involved and one largely agreed with by our editors and authors. As you’ve noted in the past we work hard to provide as much detail in our notices as possible, which by definition entails more deliberation and approvals among various parties. That said, this approach is worth reviewing from time to time, as we agree in principle that the date of publication is the date an article becomes citable. On this subject, the article in press stage will gradually disappear for many Elsevier journals as technology changes and we introduce our article based publishing for very many titles.
As Tom suggests, we do disagree with the approach, but we appreciate the thoughtful explanation and that the policy is worth reviewing.
Certainly beats this journal disappearing act version of a policy elsewhere.
How about this retraction policy? The American Economic Review, the top American Economic Association journal, still has a retracted 2002 paper posted (with full text available with login).
2002 paper webpage: http://www.aeaweb.org/articles.php?doi=10.1257/000282802762024656
2002 paper at JSTOR: http://www.jstor.org/stable/3083265
2007 retraction: http://www.aeaweb.org/articles.php?doi=10.1257/aer.97.3.1032
Interesting. We post manuscripts within hours of acceptance and make a big deal to authors that it is findable and citable from that point forth. If one is retracted, we publish a retraction in the journal (even though the paper never appeared in the VOR) and we retain the original manuscript as Retracted as an addendum to the retraction online. If we correct a preview paper, we publish an errata and in the final version, we note that changes were made between versions and what those changes were. If an author asks to withdraw, we “retract” in the same way. The record and the paper do not disappear.I am surprised Elsevier lets stuff disappear. Does this mean that they don’t index those papers?
While I generally prefer the fullest disclosure, this one doesn’t bother me much, because:
1) Elsevier says it is moving to a process (ABP) in which “in press” will generally disappear, and cites a description of a fairly coherent plan, clearly not just something created a an excuse for this.
2) Journals that haven’t yet moved to that are still following an older procedure. From experience, big organizations take a while to propagate new processes, and that’s life. If I thought journal X would move to ABP in the near future, I’d very likely to be tempted to leave the processes along until making the ABP change eliminated this particular problem. Hence, the *right* question to ask Elsevier is:
– which journals will *not* move to ABP any time soon?
– and will they continue to handle “in press” the same way?
[This problem is akin to maintenance of big software systems: people find a bug/awkward feature in release N, the vendor has done a major redesign for N+1, and the problem si gone, it’s still in test, but unless the problem is really causing damage, they are unlikely try to patch Release N, but rather roll out N+1.]
I would say that once a paper appears in citable form (online, epub, PMC, etc), it should only be eligible for “withdrawal” assuming there is a technical issue with the files AND it will shortly be replaced with the correct version. If it is not coming back, but has already existed in citable form, it should be retracted.