Making the most of retraction

Retractions are no longer rare, running at over 40,000 journal papers a year according to RetractionWatch, a most commendable blog that takes an interest in these things. A retracted paper is one that has been removed from the scientific record after publication because something is thought to be wrong with it. That much is clear enough; less clear is who – author, editor or publisher – has responsibility for deciding that something is amiss. Just occasionally, the author will seek a retraction, but usually the editor and publisher will work together when a problem arises and then initiate the retraction – sometimes with the author’s approval, more often without, and occasionally without even the author’s knowledge.

If the dominant model of retraction – that the author is to blame – is at all justified, it is in those parts of the sciences where an extra squiggle in an equation or a few more dots on a plot really do make a difference. But not all academic publication is of this sort, a truth often ignored in the sciences. In the arts and humanities, none is of this sort (Heibi and Peroni, 2024); in the social sciences, little. In these other areas, academic papers ponder rather than prove. The leading journals in social work have apparently never retracted a paper (Dunleavy, 2024). To be sure, there are retractions in the non-sciences, most often it seems for plagiarism and overlap with other papers, but it is hard to be sure, as retraction notices are often vague or non-existent. Papers in these subjects are less retracted than simply disappeared.

One issue on which it does not pay to ponder is retraction itself. A research group in Oxford has been issuing a warning to any author it finds has cited a retracted paper, which seems to assume that authors are at fault not just for creating the need for retraction, but also for eroding the scientific record by citing retracted papers ‘as if they were still valid’ (https://www.bennett.
ox.ac.uk/trialstracker/retractobot/). This exercise in black and white legitimacy is organised by the Bennett Institute for Applied Data Science in the Nuffield Department of Primary Care Health Services at Oxford University. Medicine has long had a particular interest in the scientific record, and the appalling quality of much that is added to it (see, for example, Altman, 1994; Smith, 2014). This has nothing to do with retraction. In medicine, authorship is fungible, an entitlement to recognition that need have nothing to do with writing a paper, a task that may well be contracted out to ‘communication’ companies, and organisations that promote the products of the pharmaceutical industry (organisations which include top medical journals and large academic publishers). In medicine, a paper may well have dozens of authors, even hundreds. This increases the paper’s citation and the journal’s impact factor. Author ranks are swollen by numerous honorary and courtesy authors vying for position in a Ruritanian order of authors all obligingly citing a paper they may never have read, much less written. The background of a paper eventually published in Critical Reviews in Toxicology by arrangement with the journal’s publisher, Taylor & Francis, is illustrative:

An option would be to add Greim and Kier or Kirkland to have their names on the publication, but we would be keeping the cost down by us doing the writing and they would just edit & sign their names so to speak. Recall that is how we handled Williams Kroes & Munro, 2000. (email from William Heydens, regulatory product safety assessment lead at Monsanto, February 2015)

And here your general editor must declare a personal interest: he is among those who have received a ‘cease and desist’ notice from the Bennett Institute. ‘Investigations have shown’, declares the Bennett Institute in its email, ‘that the majority [of retractions] are for fraud, misconduct, and error …’. Really? Whose? No evidence is offered. It would seem that the Bennett Institute has simply accepted the model of retraction that holds sway in academic publishing, the one in which the author is the villain, driven to chance her luck by the pressure of publish or perish and caught out in her perfidy by editors and publishers in their role as guardians of academic standards.

Retraction is supposed to be obliteration of the paper, well beyond an errata, a corrigendum or an expression of concern. But obliteration is hard to achieve if retracted papers continue to be cited, which they are. If failure to read a paper has never been much of an obstacle to citing the paper (Simkin and Roychowdhury, 2003), it would seem to be unrealistic to expect retraction to end the prospects of a paper’s citation. Nothing daunted, the Bennett Institute has put its faith in RetractoBot and RetractoBot it was that caught me out. It discovered a retracted paper cited in a piece I published recently on the mess in which medical publishing has landed itself and the danger this poses for the social sciences (Macdonald, 2023). Part of the paper relates to how editors and publishers can arrange the retraction of papers that threaten their own interests. My own paper explores how Séralinia et al. (2012), a biotechnology paper questioning the safety of Roundup, a Monsanto product, came to be retracted. When the ‘Monsanto papers’ were made public on the orders of a US court, it became clear how the editors of leading biotechnology journals had conspired to have a paper that damaged their interests (and those of Monsanto) retracted. My own paper duly cites the paper they targeted, though it is referenced only in a footnote, distinct from the main body of references so as not to confuse. My paper also cites the retraction notice for Séralinia et al. (Hayes, 2014). It even quotes from the retraction notice of Séralinia et al. These precautions did not fool the merciless RetractoBot, which identified me as a bad ’un straight away. A detailed explanation was offered direct to human beings at the Bennett Institute in the hope that this might contribute to their research: the only response was RetractoBot-generated.

When Wiley, one of the largest academic publishers, took the strange decision to acquire Hindawi, an Egyptian publisher of questionable repute, it found itself retracting 2,700 papers on the grounds that their peer review had been deficient. RetractoBot will have a field day. Retraction was not because there was anything wrong with the papers, but because there was something wrong with the system that is supposed to spot that there is something wrong with the papers. Wiley resisted this conclusion for many months before accepting that the problem lay in ‘manipulation of the process rather than author wrongdoing’ (Zhang, 2023). It was easier to lay blame for the debacle on authors, many of whom were suspiciously Chinese, than to accept that editors were obviously corrupt and the strategy of a major academic publisher just plain daft.

From mass retraction to a few individual examples of retraction. In a recent case, the publisher MDPI asked for the retraction of a paper in Research Evaluation, published by Oxford University Press (OUP) (Oviedo-Garcia, 2021). The paper was duly retracted in September 2021 and replaced by a new version created by the author and OUP that was kinder to MDPI:

Oviedo Garcia told us [RetractionWatch] that she did not ‘have full details’ about who raised concerns about the article. ‘The revision of the article was a joint work between myself and the publisher’, she said. (Kincaid, 2023)

The editor of Research Evaluation played no part in this attempt of one publisher to assuage another by schooling an author in what to say. The publisher also seems to have ignored the editor in the case of Chris Brook (Brook et al., 2021). Brook was given three days’ notice that his paper on expert witnesses was to be retracted from the Australian Journal of Forensic Sciences (personal correspondence, 2 March 2020), published by Taylor & Francis. The retraction notice is typically succinct: ‘This article has been removed for legal reasons’. It seems that anonymous individuals had expressed their displeasure about Brook’s paper to the Australian Academy of Forensic Sciences, which owns the Australian Journal of Forensic Sciences. This was quite enough to secure its retraction.

… the Academy is unique in bringing together persons of professional standing from the legal, medical and scientific professions whom [sic] have contributed to the advancement or practice of forensic science. Membership is by invitation. (https://forensicacademy.com.au/, accessed April 2024)

Readers may recall the trouble Prometheus had publishing a special issue on the role of expert witnesses in shaken baby syndrome trials. Anonymous expert witnesses complained to our publisher, also Taylor & Francis at that time, which employed lawyers to spend months trawling through all eleven papers. No accusation was ever made that anything in the papers – most written by medics and lawyers, expert witnesses themselves – was wrong, but changes to detail and emphasis were expected throughout. No sooner had the publisher’s lawyers expressed themselves satisfied with the end result than Taylor & Francis recruited yet more lawyers. These found fault – again unspecified – in all eleven papers. We gave up with Taylor & Francis, published the whole issue in the United States (Macdonald et al., 2019) and found a new publisher in Pluto Journals. We were left in no doubt, as was Chris Brook, that anonymous individuals are able to use their influence to prevent publication of academic papers that are not to their liking, and that retraction on unspecified grounds is one means by which this is achieved. Peer review is undermined when publication of academic papers is ultimately subject to a retraction system not unlike the blackballing of unwelcome applicants for membership of a gentlemans’ club (see Macdonald, 2019; Brook et al., 2021)

The problem here is not just that accepting the dominant retraction paradigm denies the complexity of the situation, but that this acceptance deters the serious research into retraction that is required to understand what is going on. It seems well worth testing the extent to which editors and publishers exploit retraction for their own purposes, not least to distract attention from their activities in the murky and highly profitable world of academic publishing. RetractoBot is just not up to the task. One cannot conduct serious research into retraction without mentioning papers that have been retracted. The scientific record certainly needs protecting, but the citing of retracted papers is a tiny threat compared with the citing of papers that have no purpose other than to be cited, a practice particularly common in medicine.

To this issue of Prometheus, late as ever. Alejandro Agafonow from the ESSCA School of Management in Paris is annoyed by the widespread belief that some people are genetically predisposed to be entrepreneurs. This is bad news for all those management schools that teach their students how to be entrepreneurs and for all those governments with policies to develop entrepreneurialism. Agafonow is particularly enraged by the Global University Entrepreneurial Spirit survey, which asked students whether they were entrepreneurs and whether they suffered from attention deficit/hyperactivity disorder (ADHD). As the students self-reported their circumstances, it was no surprise that a connexion was established. From there it is but a short step to presenting the link as causal: ADHD makes the individual an entrepreneur. But does being an entrepreneur predispose the individual to ADHD? Talshyn Tokyzhanova and Susanne Durst, from Tallin and Reykjavik respectively, write about knowledge hiding. Those who are interested in knowledge tend to focus on its diffusion and consequently on the means by which information travels. Theirs is a world of networks and of all manner of exchange relationships. The obverse – the hiding of information – has been much neglected until recently. Tokyzhanova and Durst examine this literature, not to review it, but rather to develop thematic maps by identifying key antecedents, boundary conditions, and outcomes of knowledge hiding.

There are only four book reviews in this issue, and two of these are of the same book. Would that this were a clever attempt to provide contrasting views of the book rather that an embarrassing mistake on the part of the general editor.

Altman, D. (1994) ‘The scandal of poor medical research’, BMJ, 29 January, 308, pp.283–4.
Brook, C. (2019) ‘Is there an evidentiary basis for shaken baby syndrome? The conviction of Joby
Rowe’, Australian Journal of Forensic Science, 53, 1, pp.1–9.
Brook, C., Lynöe, N., Eriksson, A. and Balding, D. (2021) ‘Retraction of a peer reviewed article suggests ongoing problems with Australian forensic science’, Forensic Science International:
Synergy, 3, paper 100208.
Dunleavy, D. (2024) ‘On the dearth of retractions in social work: a preliminary analysis of ten leading journals’, preprint available at https://osf.io/preprints/metaarxiv/y9pvm (accessed April 2024).
Hayes, A. (2014) ‘Retraction notice to “Long term toxicity of a Roundup herbicide and a Rounduptolerant genetically modified maize”’, Food and Chemical Toxicology, 63, 1, p.244.
Heibe, I. and Peroni, S. (2024) ‘Retractions in arts and humanities: an analysis of the retraction
notices’, Digital Scholarship in the Humanities, forthcoming.
Kincaid, E. (2023) ‘Article that assessed MDPI journals as “predatory” retracted and replaced’,
RetractionWatch, 8, May.
Macdonald, S. (2019) ‘The shaken baby debate – editorial’, in Macdonald, S., Luttner, S., Kirkwood,
H., Lynöe, N. … Wilmshurst, P. (2019) Prometheus Shaken Baby Debate, Researchers.One, March,
pp,1–21, available at https://www.researchgate.net/publication/331963471 (accessed April 2024).
Macdonald, S. (2023) ‘The gaming of citation and authorship in academic journals: a warning from
medicine’, Social Science Information, 61, 4, pp.457–80.
Macdonald, S., Luttner, S., Kirkwood, H., Lynöe, N. … Wilmshurst, P. (2019) Prometheus Shaken Baby
Debate, Researchers.One, March, available at https://www.researchgate.net/publication/331963471
(accessed April 2024).
Oviedo-Garcia, M. (2021) ‘Journal citation reports and the definition of a predatory journal: the
case of the Multidisciplinary Digital Publishing Institute (MDPI)’, Research Evaluation, 30, 3,
Séralinia, G.-E., Claira, E., Mesnagea, R. et al. (2012) ‘Long term toxicity of a Roundup herbicide and a Roundup-tolerant genetically modified maize’, Food and Chemical Toxicology, 50, 11,
Simkin, M. and Roychowdhury, V. (2003) ‘Read before you cite!’ Complex Systems, 14, pp.269–74.
Smith, R. (2014) ‘Medical research – still a scandal’, BMJ Opinion, 31 January, available at https://
blogs.bmj.com/bmj/2014/01/31/richard-smith-medical-research-still-a-scandal/ (accessed May
Zhang, H. (2023) ‘Is Hindawi “well-positioned for revitalization”?’ BishopBlog, available at http://
deevybee.blogspot.com/2023/07/is-hindawi-well-positioned-for.html (accessed July 2023).

Stuart Macdonald
General editor