Last year, Naunyn-Schmiedeberg’s Archives of Pharmacology found itself on the receiving end of what its editor Roland Seifert called a “massive attack of fraudulent papers” that were the product of paper mills.
In response Seifert — who says the journal ultimately will have retracted 10 of those articles and stopped another 30 from being published — has produced a 20-point list of red flags that indicate the possibility of a paper mill in action, and features of these papers in general.
We won’t reproduce the list in its entirety, but here are a few highlights, in no particular order.
Seifert states that his journal’s investigation found that authors linked to paper mills always used commercial instead of academic email addresses:
Often, the email addresses had little relation to the name of the corresponding author. In several cases, authors claimed that their institutions do not provide them with academic email addresses. Evidently, the lack of use of academic email addresses renders identification of the fraud authors much more difficult, if not impossible. In the meantime, we have learned that all “real” academic institutions across the world provide academic email addresses. It has just become a bad global habit that many scientists use commercial email addresses for convenience, and this security gap is aggressively exploited by paper mills. Therefore, our journal does not allow anymore submissions without a valid academic (or pharmaceutical company) email address.
Seifert also notes that email exchanges with the authors of milled papers can be revealing. He advises editors to be on the lookout for:
Puzzling email clusters: Strikingly, in several fraud cases, the Editor-in-Chief received emails from authors from apparently different groups within a short period of time (minutes to hours!) as if the emails were sent by one and the same person under different email addresses. It was also noted that the text was rather similar. Thus, it can be assumed that a single paper mill employee is handling multiple fake submissions in one and the same “work session.”
Sluggish email communication (if any): Sometimes, there was no communication at all, because email addresses (especially of co-authors) did not work or because authors did not respond. Sometimes, we had to wait for up to 6 (!) months to get an answer. Blocking of emails by servers was used as another common “excuse” for not responding quickly. However, we never received any message from email servers that a message could not be delivered, unless an email address was invalid. The sluggish email communication is the major reason why it took so long to publish the retraction notes. We would have liked to proceed much faster, but we could not because the Editor-in-Chief and SpringerNature had to play the formal retraction process by the formal rules. Most unfortunately and to our dismay, fraudulent authors only care about rapid publication of their papers but not about rapid procedurally correct retraction. This is highly unethical and unprofessional conduct.
Another tell:
No ORCID IDs: Very few authors used ORCID IDs. This is an effective method of disguising the identity of an author, particularly when first names and family names are very common in the literature.
And millers generally agree to retraction immediately:
Most surprisingly, when confronted with the suspected fraud, quite often the corresponding authors very quickly (within very few hours and almost simultaneously!) agreed to a retraction without actually admitting the fraud openly. Often, authors mentioned the global term “problem with the data” without being specific what the problem is.
However, that observation seems to be in conflict with another of Seifert’s points:
Painful formal retraction process: The formal retraction process was very sluggish from the side of the authors and lacked professional conduct.
As Seifert writes:
All the above points show that the “success” of paper mills does not rely on a single tactic but a well-balanced portfolio of malicious strategies. These strategies complement each other to lead to the desired result. At the same time, the portfolio of strategies minimizes the risk of the fraud authors to be identified, minimizing the risk of negative professional consequences.
For more coverage of paper mills, which have surfaced thanks to the work of scientific sleuth Elisabeth Bik and others, see these posts:
- ‘Striking’: Journal editor suspects paper mills behind rash of withdrawn manuscripts
- Publisher retracting 68 articles suspected of being paper mill products
- Journal slaps 13 expressions of concern on papers suspected of being from a paper mill
Like Retraction Watch? You can make a tax-deductible contribution to support our work, follow us on Twitter, like us on Facebook, add us to your RSS reader, or subscribe to our daily digest. If you find a retraction that’s not in our database, you can let us know here. For comments or feedback, email us at [email protected].
Seven of the journal’s 10 retractions are entries in a Google spreadsheet, which helped bring the problem to the editors’ attention.
Also, accompanying blogpost about two papermills – the second being the source of the seven papers retracted from Naunyn-Schmiedeberg’s Archives:
http://eusa-riddled.blogspot.com/2020/04/dark-satanic-papermills-2.html
This is just my tiny little area of science, but when I see a paper title with any combination of the word “axis” and two or more of “miRNA”, “lncRNA”, and “circRNA”, I assume the whole thing is bogus and likely part of a paper mill, broadly or narrowly defined.
I predict that we’ll see the same pattern play out for mills as we did for predatory journals.
People raising the problem will achieve some early success in building awareness before the lawyers get going with defamation suits. Librarian-scholars will then spend the next 10 years worrying about whether you’re racist if you condemn paper mills. By 2050, publishers and librarians will declare paper mills to be definitely bad and promise a definition of paper mills by 2060 (to be finally be published in Nature in 2072). A non-profit joint initiative between universities and publishers called Write.Papers.Yourself will
then launch a website with some generic advice on how to say no to using paper mills. As the final nail in the coffin, COPE will release a discussion paper to identify scholarly infrastructure initiatives that should definitely maybe consider doing something about paper mills.
100% correct.
I don’t know that what you flagged is a contradiction– I think they’re saying the authors won’t ever object to the retraction, but when it comes to hashing out the details they’ll suddenly go MIA.
If the reason for the delay in getting the falsified articles retracted is due to “sluggish” email correspondence, the publishers should change the rules. The problem with allowing a delay is not only naiv, the publishers also giving the authors (paper mill) a possibility to produce additional fabricated data to present.
We are talking about obvious falsifications of data and the publishers should be allowed to retract without any feedback from the authors (paper mill). A 48 h deadline for replying and showing raw data if suspicion of fraud is raised should be mandatory.
I was particularly interested in this statement as a potential prevention to paper mill submission: “our journal does not allow anymore submissions without a valid academic (or pharmaceutical company) email address”. I’m not aware of any “list” of institutional email address suffixes, nor am I aware of any peer review system providers that can facilitate this kind of restriction. Does this mean these checks are being done manually for every co-author? How can this be done at scale? Could a Ringgold type solution be the answer to this problem?
The request for original data at submission strikes me as a very positive move not only for papermill prevention but reproducibility too. But how long before this “data” is sold as part of the paper-mill package – do we then need to raise expectations of the reviewing community to start assessing legitimacy of data? This puts pressure on an already strained group of stakeholders.
Lastly, while I am a huge advocate of ORCID, it’s not an identity verification system – anyone with an email address can get an ORCID. As such I’m not currently convinced of its role in mitigating publication fraud.
Does this just happen much more often to poorer journals with poorer editorial boards?
Dear Colleages,
many thanks for the comments. Here is a short reply to some:
1. Yes, the excel spreadsheets were most helpful for nailing down the fraud.
2. The formal COPE rules are a huge problem in terms of solving paper mill issues quickly. The paper mills know and exploit this. I agree that the rules must change.
3. Yes, to this end we check email addresses manually.
4. And yes, ORCID IDs do not prevent fraud. The case exposed in my editorial was an “unsuspicious” ORCID ID case.
5. In an effort to avoid the long-term contamination of science with this type of unprofessional conduct, I have now implemented a class for PhD students in which I teach the principles of proper ethical conduct using the paper mill case.
Re: point 3 and the institutional email address policy.
This seems a blunt instrument. What about authors from countries where academic email addresses aren’t universally used because of issues with e.g. institutional servers blocking legitimate correspondence in an attempt to prevent junk mail (e.g. automatic reminders from journal submission systems, which are often undeliverable to certain addresses)? The author isn’t at fault in being unable to use their institutional email address, but they now can’t submit to your journal unless they happen to have a co-author with an email address from an institution where restrictions aren’t a problem? (Might this in fact encourage gift authorship in some cases?) What about individuals who aren’t currently associated with any institution (e.g. retired or independent)? Would they have no option to submit to the journal anymore? There are various reasons for authors using their personal email address. Just because an institutional address has been provided doesn’t mean that it comes with full functionality for busy academics.
Improved journal policies and checks are important to combat fraud, but a more refined approach would be preferable.