What's with all the retractions?

The Times Higher Education reports that there's been a tenfold increase in journal articles over the past two decades.

"The growth has been particularly pronounced in the past few years, even factoring out 22 retracted papers authored by Jan Hendrik Schon, the disgraced German physicist, earlier this decade," the Times reports.

According to analysis completed by Thompson Reuters, there were 94 retractions among 1.4 million papers published last year compared to approximately 300 papers retracted over the last 20 years. According to the study, 25% of retractions were due to plagiarism or falsified data, while 26% were due to errors.

But why the increase? Are scientists today under greater pressures to publish, no matter what the data say? Or do we simply have better ways of spotting fraud and plagiarism today than twenty years ago? Share your thoughts.

-- Edyta Zielinska, Associate Editor, The Scientist

I write as having had ~ 40 years observations, experience as a referee (for Js, grant giving bodies, including the US National Science Foundation).

As already emphasised/pointed out to many highly distinguished 'practising scientists', acting as Editor-in-Chief of Js (such as, Dr Cindy Dunbar- BLOOD) and editors of other Js (SCIENCE, NATURE, NATURE Med., Lancet, BMJ, The Scientist, NEJM), the serious problems arise from:

(a) Over influential scientists-referees not being hypercritical of MSS by well known 'names' and too trusting of well known names submitting MSS,

(b) anxious for fast - business like turn over-the over reliance of busy editors-in-chiefs on the preliminary screeners- sadly not quite so widely read- limited to literature going back to 10 years of MEDLINE at best. In the old days, a MSS came, the referee was assigned and the decision really depended upon a widely read person. Even in that system though there were problems of course- such as the 3,-4 paper by the 23 yr old Joule (working in his father's Salford brewery back yard) submitted to the Royal Society was not accepted for a full paper (but a communication) with the thought " how could such important conclusions be put in such a short 4 page paper" and was eventually published in Camb. Phil Soc.

Many years later when Joule was asked what he thought of all that affair, he said:" I can imagine those gentlemen sitting in London thinking that nothing could come out of a place where they dine at mid day!' (For the benefit of the people in the USA, Europe, elsewhere, the 'working class' people eat 'dinner' at lunch 12 noon-1p.m., whereas 'gentlemen' of higher status society eat lunch at 1 p.m. and dinner in the evening).

Francis Crick thought that refereeing must remain anonymous/confidential because the scientists being 'human' might get vindictive.

Solution: (i) even if it increases publishing time, the MSS must be looked at by senior real scientists, chosen because of their special expertise in that field of the MSS, must be asked to declare competition and if currently working in the field. I know of cases when the referee have delayed sending in the report while completing the work, sending in a preliminary report to establish priority! Great bitterness expressed to me

(ii) if too much conflict of opinion between the authors and the referee- some time of an equally high expertise, experience, status, publish the MSS with the referees comments side by side.

Don't reveal the name of the referees if desired.

(iii) ALL Novel findings, if significance not grasped after refereeing, something bizarre could well be published. This will prevent major discoveries not being published quickly enough.

(iv) The senior authors must declare that close supervision was exercised.

Bearing in mind the amount of rubbish that every pracising scientist is obliged to read every day, 94 out of 1.4 million is a dreadfully low retraction rate.

Regards

My top five list of reasons for the increase in retractions:

5. Erosion of the referee review process, caused by fundamental changes in the scientific publication industry, resulting in relaxation of the standards for publication.

4. Erosion of training and supervision of research students, especially in large research groups where the chain of command often results in inexperienced workers supervising newcomers, resulting in mixed signals on scientific standards.

3. The increase of multi-author and multi-lab publications, where no one author is in a position to evaluate the entire manuscript, and co-authors often sign on without ever reading the manuscript prior to submission.

2. Increased pressure to publish, caused by intense competition for scarce funding, exacerbated by rigid numbers-driven institutional criteria for tenure and promotion, and resulting in corner-cutting at all stages of data acquisition, interpretation, and presentation.

1. The intrusion of legal standards in place of scientific standards in the evaluation of new science, leading to the criminalization of scientific error, resulting in an increase in self-censorship. When questions arise, it is safer to simply retract a paper, rather than risk multiple rounds of prosecutorial investigation and possible criminal sanctions.

#2 – #5 lower the threshhold for publication of defective manuscripts, while #1 lowers the threshhold for retraction of a defective manuscript once it has been published.

Major unanswered question: Why isn't #1 a deterent up front? My guess: nobody ever expects to be caught.

I'd add to JON199271's top five a kind of influence and infiltration that could be described loosely as industrial lobbying, though unfortunately this is probably not a major reason for retractions.

Such influences can be more subtle than the publication of a complete spoof medical/pharmaceutical journal (Elsevier). In my own field I suspect that there may be a kind of tacit consensus (pharmacepeias, regulatory agencies, industry, literature) designed to keep scientific standards at a reasonably low level in the analytical control of pharmaceutical chemistry and manufacturing. The motive here would be industrial expediency, because if you let scientific rigour take over completely, the uncertainties would cause R&D and manufacturing to grind to a halt.

Most likely, similar influences afflict other fields where industries need to maintain momentum and limit contestation to survive. Lead in petrol and CFCs are gone, but food additives and agrochemicals of dubious safety are still with us.

EdytaTS48532 wrote:there were 94 retractions among 1.4 million papers published last year compared to approximately 300 papers retracted over the last 20 years.

So what's the percentage increase then?

It seems that Germany is the country in which a fraudist receives the right punishment and has to live with it. Just to retract a publication and not to be prosecuted does not bring changes to the behavior of the fraudist as demonstrated in the article by The Scientist July 13th, 2009. One can be proud of Germany for their consequence in dealing with such fraudists by publishing their names, demand of retraction of publication and attempt to derecognize the doctorate title.

There has been an exponential increase in published papers on any subject since about 1990, with apparently no end in growth. It is also now rare to get a single author paper. That the retractions are so low is a compliment - though I confess - the peer review process is creaking under the strain!