Retraction Watch

Tracking retractions as a window into the scientific process

PLOS ONE’s correction rate is higher than average. Why?

with 17 comments


When a high-profile psychologist reviewed her newly published paper in PLOS ONE, she was dismayed to notice multiple formatting errors.

So she contacted the journal to find out what had gone wrong, especially since checking the page proofs would have spotted the problem immediately. The authors were surprised to learn that it was against the journal’s policy to provide authors page proofs. Could this partly explain PLOS ONE’s high rate of corrections?

Issuing frequent corrections isn’t necessarily a bad thing, since it can indicate that the journal is responsive to fixing published articles. But the rate of corrections at PLOS ONE is notably high.

According to an analysis published yesterday on a blog by Mark Dingemanse, a researcher at the Max Planck Institute for Psycholinguistics, between May and June, 2015, the journal published 8466 articles and 474 corrections. Of course, many of those corrections were issued for articles published before that specific time period, but that still works out to a correction rate of more than 5% of the journal’s total output.

In contrast, Medline data show that more than 800,000 papers were published in the 2015 fiscal year, as well as more than 12,000 errata (again, not all errata were for papers published in 2015); this gives a correction rate of around 1.5%. 

Dingemanse’s post was prompted by a tweet from prominent psychologist Dorothy Bishop about what she experienced at the journal:

When Bishop, who is based at the University of Oxford, UK, complained to the journal about formatting errors in her recently published PLOS ONE paper, she received the following response

Despite any past republication instances and regardless of fault, current PLOS policy does not permit republication unless confidential or copyrighted information has been published and needs to be purged from the publication record. At this time, we do not provide proofs during the production process of our manuscripts.

It went further to say:

…we do not publish corrections for errors that do not affect the scientific content of the article. Because these are not content errors, there is nothing further we can do.

When queried about the policy, Bishop told us:

This is very unusual and out of line with normal publishing practice — and, if my twitter feed is anything to go by, it has led to numerous problems. One person even noted that PLOS One had misspelt their name on a paper and then said it could not be corrected.

After hearing from the journal, Bishop posted this comment under her paper:

I regret that the formatting of the pdf version of this article impairs readability.

The authors had formatted the article so that the 27 statements were distinctive, but we were asked to remove that formatting from the manuscript version of the document.

At this point we did not realise that PLOS One no longer provides authors with page proofs, since this is a change in their previous practice. Had we been able to check proofs, we would have pointed out that in the pdf version, the statements do not stand out and are in some cases run in with the prior heading. The “Supplementary Comments” on each statement have more prominence than the statements themselves.

PLOS One editorial staff tell us they are not able to correct formatting errors.

This isn’t the first PLOS ONE paper with an error stemming from typesetting problems. As Dingemanse writes:

In the period May-July 2015, PLOS ONE published a total of 474 corrections. Over a quarter of these (132) indicate that the error was introduced at the typesetting stage, i.e. beyond the control of the authors.

According to Dingemanse, the numbers for the whole of 2015 are even worse: The journal published 30,790 research papers and issued 1939 corrections in 2015 (again, not necessarily for papers published in 2015), which bring its correction rate to just under 6.3%. Additionally, 415 (21.2%) of those corrections acknowledge publisher errors.

Furthermore, Dingemanse says, the rate of corrections due to publisher errors seem to be rising: 13% of corrections published in 2014 cite a publisher error, and the number is as low as 4% for corrections in 2013.

Statistics released by the US National Library of Medicine suggest that the number of errata increased by 29% — from 9,602 to 12,344 — between the 2014 and 2015 fiscal years. 

However, a 2013 study by Daniele Fanelli, a researcher at the Meta-Research Innovation Center at Stanford University in California, noted that the proportion of published errata has not increased for decades:

…despite a steady increase in the number of publications covered by the [Web of Science], the proportion of errata has remained relatively constant since the 1970s (and arguably since the 1950s).

In response to Bishop, other Twitter users also posted their experiences:

Others have questioned whether PLOS ONE’s massive editorial challenge — publishing more than 30,000 articles — raises the chances that mistakes will happen. (Remember the paper that cited the Creator?) As David Crotty wrote in March in Scholarly Kitchen:

…what about editorial decision-making? PLOS ONE has some 6,100 editors. Rather than funneling everything through an Editor in Chief, the peer review and decision-making process is spread broadly…There is no consistent level of quality control because there are 6,100 different sets of standards being used and no central point where they come together.

PLOS ONE were unable to comment on the situation by press time, but a spokesperson from the organization told us to 

Look for a more thorough, accurate, meaningful and in context blog post from PLOS in the next week or so.

Like Retraction Watch? Consider making a tax-deductible contribution to support our growth. You can also follow us on Twitter, like us on Facebook, add us to your RSS reader, sign up on our homepage for an email every time there’s a new post, or subscribe to our new daily digest. Click here to review our Comments Policy. For a sneak peek at what we’re working on, click here.

Written by Dalmeet Singh Chawla

August 5th, 2016 at 1:05 pm

  • Peter Apps August 5, 2016 at 1:37 pm

    What I cannot understand is why an electronic manuscript being turned into an electronic paper needs to go through “typesetting” at all – this is 2016 isn’t it ?

    • Lee Rudolph August 5, 2016 at 8:25 pm

      Journals (rightly or wrongly) want a consistency of look. Even using a well-designed publisher-supplied (La)TeX template, many authors will find (ingenious, careless, or downright stupid) ways to subvert the editors’/designers’ intent; and many publisher-supplied templates are not well-designed. Moreover, the publisher may use proprietary fonts, so that with the best will and TeXnical skill in the world, the appearance (even at such a coarse level as page length!) of the “printed” electronic paper typeset and distributed by the publisher may differ (if only esthetically) from the typeset version of the (perfectly template-compliant!) electronic manuscript submitted by the author.

      And then there is the cesspool that is Microsoft Word.

  • John Hutchinson August 5, 2016 at 2:02 pm

    It is valuable to learn this. I will not now consider publishing in this journal.
    Like many authors, I take as much care and pride in the few papers I produce as does an artist working on a painting. Usually I find a number of small corrections to make on the proofs, some important for the meaning, some more cosmetic, some my fault, some the copy editor’s. Such errors if left uncorrected would often have really detracted from the pleasure of finally seeing my work in print.

  • Carrie August 5, 2016 at 2:04 pm

    I agree. I would not consider publishing in any journal with this policy. I hope the blog post describes how they will change the policy in the near future rather than trying to justify it.

  • Stanish August 5, 2016 at 2:51 pm

    I recently published with them for the first time. I am a strong supporter of OA and the philosophy behind PLoS. I have also peer-reviewed for the journal. There were some frustrating problems with the editorial process but nothing egregious. However, I learned about their “no correction” policy before submission so I paid for my own professional copyeditor ($500). I also paid the full amount page charge because I have a grant ($1800). So I am a bit torn. I am not sure how far my altruism will extend in the future. It is just as easy to pay for OA in other higher impact journals.

  • Tekija August 5, 2016 at 6:25 pm

    This is no longer scholarly publishing.

  • Steven St. John August 5, 2016 at 8:06 pm

    Some scientific papers are the result of years of work. To have the end of that long, careful process be handled with such carelessness is demoralizing.

  • herr doktor bimler August 5, 2016 at 8:30 pm

    PLOS One editorial staff tell us they are not able to correct formatting errors.

    It is somehow physically impossible to replace a computer file? Or they just prefer not to admit to a mistake?

  • Franc August 7, 2016 at 9:05 am

    Possibly a storm in a teacup?

  • Jose August 7, 2016 at 8:06 pm

    > Possibly a storm in a teacup?

    No seriously, when you’ve spent years on a project, pay significant money, and finally get it ‘out the door’ to only have the final version appear as if you’re an incompetent nincompoop, it matters.

    Many readers may not know it can be/is likely PLOS’s error, and might assume the authors are sloppy- this can quite conceivably torpedo a fellowship, grant, promotion, or job offer. IT SERIOUSLY matters. It’s a sad commentary on how academic has ‘progressed’ but….

    • SB August 8, 2016 at 9:09 am

      I quite agree. Although no longer in academia my job still requires me to occasionally peruse the literature and I’ll hold my hands up and say that articles with poor spelling, grammar, layout etc annoy me and I often think less of them. Having read this I will try to make more of an effort but I’m not the only busy and judgemental person in the world – you wonder how many researchers are getting their work panned because of the typesetting?

  • Franc August 8, 2016 at 8:55 am

    PLOS does publish corrections if such affect the scientific content of the article. If errors are more than cosmetic then a cogent argument should be made to the journal.

    • Carrie August 8, 2016 at 10:57 am

      “One person even noted that PLOS One had misspelt their name on a paper and then said it could not be corrected.” I’m not interested in making arguments to people about why things should be correct. If people are happy with this policy they can submit to PLOS. I will vote with my feet (submissions).

  • Mark D. August 10, 2016 at 8:39 am

    Let me register a prediction here: it is likely that the “more thorough, accurate, meaningful and in context blog post from PLOS” will try to fudge the issue by making it (also) about author-introduced typos in addition to publisher-introduced typesetting errors. One way to do this would be to point to their policy of ‘author proofs’ which states authors should treat their own final ms as definitive, and to give examples of authors seeking to rectify problems they should have (or at least might have) caught before.

    If this were the case (and remember this is only a prediction), there are two replies to that: (1) This would sidestep the real problem, which is that at least 20% of all corrections (or at least 1.5% of the total publication output) actually concern publisher errors, where clueless or careless typesetters introduce errors beyond the control of the authors and so where PLOS violates the promise given to authors to treat their final ms as definitive. I write ‘at least 1.5%’ because god knows how many authors just shrug and forget about it, not filing for a correction; my guess is the majority. (2) Even the ‘author proofs’ system is deeply unfriendly to authors: as everybody knows, no matter how careful one is in the final version of the ms, some typos magicaly become visible only in the page proofs, and the very least a publisher can do is allow authors to have a say in what the final product of months or years of labour will look like.

    On Twitter, Alex Holcombe noted this “software limitation” was “one reason I resigned from their Advisory Board”. A software limitation! This is a problem that’s been solved by, what, 99.9% of other academic publishers? Perhaps there’s some inspiration to be had there.

  • Peter Bloem September 15, 2016 at 1:42 pm

    More than a month in. Has anybody been able to locate that blog post?

    I have a paper with a (lapsed) “revise and resubmit” at PLOS One, but I think I’m going to find another venue.

    • Dalmeet Singh Chawla September 15, 2016 at 2:26 pm

      I asked PLOS ONE on 23rd August whether the blog post had been published. They said it hadn’t but will let me know when it has. No news as of yet.

    • Mark D. September 30, 2016 at 7:51 am

      On Aug 30 a PLOS staff member wrote to me saying “We are still working on that blog post. I will be sure to let you know when it is published.” Haven’t heard from them since. Just sent a reminder.

  • Post a comment

    Threaded commenting powered by interconnect/it code.