by Glen Wright
David Tennant: any relation?
According to the restraining order, no.
RETRACTIONS
Even a paper that has passed the rigorous review process may later turn out to be fundamentally flawed. In such cases, a paper can be formally retracted from the literature.*24 One of the first English language retractions was self-submitted by Benjamin Wilson to the Philosophical Transactions of the Royal Society on 24 June 1756.† It reads:25
Gentlemen,
I think it necessary to retract an opinion concerning the explication of the Leyden experiment, which I troubled this Society with in the year 1746, and afterwards published more at large in a Treatise upon Electricity, in the year 1750; as I have lately made some farther discoveries relative to that experiment, and the minus electricity of Mr Franklin, which shew I was then mistaken in my notions about it…
I shall be very glad to have this acknowledgement made public, and to answer that end the effectually, I wish that it may have a place in the Transactions of the Royal Society.
Wilson had been locked in a public debate with Mr Franklin* on the question of whether lightning conductors should be round or pointed at the top, and had previously arranged an audacious demonstration before King George III at the Pantheon on Oxford Street in London to prove his point.†26
Retractions are an important part of the scientific process, yet they generally receive scant coverage. There has, however, been increasing interest in improving documentation of retractions in recent years. Leading the charge is Retraction Watch, once called the ‘Garbage Collectors of Science’ by a Swiss radio station.27 Retraction Watch looks out for retraction notices, follows up on tips regarding faulty science, and aims to improve the overall transparency of the scientific publishing process.
Co-founders Ivan Oransky and Adam Marcus say that ‘retractions are born of many mothers’ and, while outright fraud is quite rare, such cases are especially damaging to both science and to the career of the perpetrator.28 Anaesthesiologist Scott Reuben spent six months in prison for faking data and was ordered to pay back $360,000 in restitution for misusing grant money.29 Dong-Pyou Han, a former researcher at Iowa State University, received a 57-month prison sentence and an order to repay $7m in grants after he spiked samples of rabbit blood with antibodies to make a potential vaccine against HIV appear more effective than it truly was.30
Caught on camera
Retraction notices posted by journals are typically terse affairs. For example, a notice retracting a 1994 paper from Nature simply read: ‘We wish to retract this Article owing to an inability to reproduce the results.’ Yet the real story is closer to spy vs. spy than science.31
Karel Bezouska was one of the foremost biochemists in the Czech Republic, until an ethics committee at Charles University in Prague found that he had probably committed repeated acts of scientific misconduct. In one absurd instance, Bezouska realised that his results could not be replicated, so he broke into a lab where another team was attempting to replicate his results and adulterated the samples in an attempt to change the outcome of the experiments. A student working in the lab tested the samples and found that they’d been handled without authorisation. The lab installed CCTV cameras and caught Bezouska breaking into the room and surreptitiously rummaging around in their fridge.
Fake it until you make it
Faked peer review is one of the more egregious violations of academic integrity leading to retractions. In August 2012, Korean researcher Hyung-In Moon had several papers retracted because he himself had peer-reviewed them.32 Moon suggested preferred reviewers during the submission process who were either himself or bogus colleagues. In some cases, he simply invented names, but on other occasions he used the names of real researchers (so that a web search would verify their legitimacy) and created email accounts that could be used to provide the peer-review comments. To make the reviews appear more realistic, he submitted favourable comments, but provided some critical feedback or suggestions on how the paper might be improved.
Similarly, in August 2014 SAGE Publishers retracted 60 articles from the Journal of Vibration and Control after a 14-month investigation revealed a similar scam.33 The scandal centred on Peter Chen, formerly of the National Pingtung University of Education in Taiwan, who had created various aliases to enable himself to peer-review and cite his own papers. The publisher admitted that it could not definitively determine the number of individuals involved as their attempts to contact 130 suspicious email addresses resulted in precisely zero responses. The publisher and editor of the journal confronted Chen with the allegations in late 2013. When they were unsatisfied with Chen’s explanation, they alerted his University. Chen resigned in February 2014, and in May the editor retired and resigned from the journal. The fallout didn’t stop there. Taiwan’s then education minister, Chiang Wei-ling, had supervised the thesis of Chen’s twin brother and appeared on several of the retracted papers. Ultimately Wei-ling also resigned over the scandal.34
Plagiarism
Good old-fashioned plagiarism is no doubt common, but one paper in particular could easily have been dismissed as an April Fool’s joke. The Indian Journal of Dermatology retracted a paper on plagiarism . . . for plagiarism.35 The paper included definitions and strategies to detect and prevent plagiarism, but was itself found to have been copied from a master’s dissertation. The author of the retracted paper, Thorakkal Shamim, had been part of a panel of experts on plagiarism consulted by a student a few years earlier. Shamim had copies of the responses to a questionnaire the experts had answered and decided to publish the results, spelling mistakes and all, simply adding an introduction and a conclusion. To make matters worse, Shamim had previously taken a hard line on plagiarism, writing an article suggesting that plagiarising authors should be blacklisted and banned for submitting an article for at least five years, and that the head of the author’s department and institution must to be notified.36
In a similar incident, the author of an article on reincarnation sought to reincarnate the Wikipedia page on reincarnation, copying and pasting considerable chunks of text directly into the manuscript.37 The retraction notice states that the paper was being pulled because of ‘duplicity of text’.38
Calling bullshit
A Washington State University investigation found that a researcher studying how to turn cow manure into natural gas fabricated data in a journal article (and also failed to declare a commercial conflict of interest).39 Rather than admit to the falsification, the researcher told the investigation that he had lost the data. He claimed that a wind storm dumped his notebook into a manure pit during a visit to a dairy farm, and that photocopied pages of the notebook were lost at his sister’s house. He neglected to provide an explanation for the loss of all the data files stored on his office computer.
Obese
Peculiar circumstances precipitated the retraction of a paper on obesity treatment from Biochemical and Biophysical Research Communications. The authors were all affiliated with the University of Thessaly, a real university in Greece. The authors were however not real: no trace of them can be found online, and the correspondence address is not an official institutional one. The names of the second, third and fourth authors appear to have been sloppily copied and pasted from a real paper,* with the other two being copied from a different paper. Bruce Spiegelman, a cell biologist at Harvard, said that he had presented similar findings at various research meetings and was preparing to submit them for publication. The particular proteins being studied had not previously been the subject of any paper looking at their role in obesity, so Spiegelman was suspicious.
The real intrigue here is why anyone would want to pull such a move. Spiegelman is, in addition to his academic posting, a co-founder of a company developing therapeutics for metabolic disorders, and he reckons that premature publication of his results was a malicious act intended to complicate future patent applications relating to the results.40 Luckily, Spiegelman had already applied for the paten
ts.
Hearts and minds
Michael LaCour had struck academic gold. His study, entitled ‘When contact changes minds: An experiment on transmission of support for gay equality’41 challenged the conventional wisdom that attempts to win hearts and minds only entrenches existing views. The paper was a hit on social media, and This American Life dedicated a whole podcast to it.42 I listened with anticipation as the findings were described: a single instance of personal contact with someone affected by the ban on gay marriage could change a person’s opinion on the issue. The result seemed too good to be true. It was: LaCour had faked the data.
The unravelling began when Joshua Kalla and David Broockman from the University of California, Berkeley pored over the numbers. Noticing some inconsistencies, they published a damning report describing the multiple reasons they suspected something shady.43 They realised that the baseline ‘feeling thermometer’, which was supposed to be calibrated to local samples, was instead identical to a freely available national dataset. In addition, the changes in participants’ feeling thermometer scores were perfectly normally distributed – i.e. not a single participant changed their mind in a way that meaningfully deviated from the distribution – a highly unlikely result in the real world.
The researchers reached out to a senior co-author of the paper, Donald Green, to alert him to their discovery. Green agreed that unless LaCour had a good explanation, a retraction was in order. LaCour provided no such explanation. At first, he claimed that he’d simply lost the data. Later, he would claim that he had destroyed the data to comply with privacy and confidentiality protocols.
Green recalls: ‘I sent off my retraction, and I went to sleep and I woke up in the morning at 5:30 and there was a lot of email.’44
An editorial in the Wall Street Journal snidely suggested that the LaCour paper was so popular because it ‘flattered the ideological sensibilities of liberals’.45 As a sensitive liberal snowflake myself, I was certainly happy to hear of the findings, and equally disappointed to learn that people are just as set in their ways as we always knew them to be.
But there is a heartening twist. The two whistle-blowers were themselves in the middle of conducting a similar study, with opinions on transgender people as the subject. Their study found that the canvassing strategy really can change people’s minds.46
Treefinder
While many are working hard to reduce prejudice, one academic’s attempt to further entrench outmoded attitudes led to a 2015 retraction from BioMed Central. The journal retracted a highly cited paper describing the software Treefinder (software that creates trees showing potential evolutionary relationships between species) because the lead author and software developer changed the licence terms to make it unavailable in certain countries.47 Firstly, in February 2015, creator Gangolf Jobb prohibited US users from using the software, citing the country’s imperialism. Then in October 2015, he prohibited its use in countries he viewed as too immigrant-friendly, bringing the paper into conflict with the journal’s policy that all software discussed in papers be freely available.
Jobb told Retraction Watch that the software is still available to any scientist willing to travel to non-banned countries:
Every scientist can use Treefinder, as long as he or she does it in one of the allowed countries and is personally present there. However, having to travel to a neighbouring country is inconvenient, I admit. I don’t care.
His co-authors, who had no say in the decision, readily supported the retraction (though I imagine that losing a paper cited over 700 times must have hurt a bit). Sandra Baldauf, a biologist at Uppsala University in Sweden, was one scientist that was happy to go back to the drawing board: ‘I would stop using [Treefinder] just on general principle, even if we had to resort to using pencil and paper.’48
Con Man
Diedrek Stapel, a social psychologist from the Netherlands, was something of a star in his homeland. Stapel wrote many well-regarded studies on human attitudes and behaviour, and his results, like those of LaCour, often told us what we wanted to hear (or at least expected to hear) about human nature. Stapel also precipitated his own dramatic downfall by perpetrating a bold academic fraud over the course of a decade, fabricating results and ultimately notching up over 50 retractions.
One of Stapel’s much-publicised studies, appearing in Science, purported to show that a dirty environment brought out people’s latent racist tendencies. Stapel supposedly conducted a study at Utrecht train station that showed that white people tended to sit further away from a black person on a bench when the surrounding area was strewn with litter compared to when it was tidy. Years later, in the midst of the self-initiated unspooling of his career, Stapel visited the train station and realised that there was no location there that matched the fictional one he had meticulously described in the paper.
Stapel has never denied that his deceit was driven by ambition, a common thread among high-flying fraudsters. However, he was also obsessed with order and had long been driven to frustration by what he saw as the imperfect nature of experimental data. Instead of crunching the cumbersome numbers of the real world, Stapel concocted results that were pleasing to the eye. ‘It was a quest for aesthetics, for beauty – instead of the truth,’ he said in a tell-all interview with the New York Times.49
Another of Stapel’s creations highlights his questionable quest for order. He designed a study to test the hypothesis that people presented with a bowl of M&Ms will eat more if they are primed with the idea of capitalism. Subjects would answer a questionnaire: half would do it sitting in front of an M&M-filled mug emblazoned with the word ‘kapitalisme’ and the other half would have a mug adorned with jumbled letters. Stapel had a student load the mugs, M&Ms, and questionnaires into his car, saying that he’d conduct the study at a local high school. Instead, he drove home, binned the majority of the questionnaires, and set about simulating the experiment. Eating what he believed to be a reasonable quantity of M&Ms, he filled out the questionnaire and built a dataset around that estimate.*50
The long-running and wide-ranging nature of Stapel’s fraud provided the perfect opportunity for a couple of language experts to investigate the linguistic fingerprints of fraud. They analysed patterns in 24 of Stapel’s fraudulent papers (170,008 words) and compared them with 25 of his genuine publications (189,705 words). They found that the writing style matched known patterns of deception in language, including, for example, the use of fewer adjectives in fraudulent papers. The fraudulent papers also contained a greater number of words pertaining to methods, investigation, and certainty.† This is the painful irony of Stapel’s search for perfection: he unwittingly wrote the hallmarks of deception into his otherwise perfect papers.
Foiled
Finally, here is a retraction that was quite close to home. Colleagues at my research institute had recently published a paper about ocean warming and acidification in Science51 when I learned of a conference paper pulled from ‘Heat Transfer 2014’.52 The climate-sceptic author claimed to have single-handedly debunked ocean warming with a home-made experiment using tin foil and cling film.
THE GARBAGE COLLECTOR OF SCIENCE
Ivan Oransky is a co-founder of Retraction Watch. He is the vice president and global editorial director of MedPage Today, Distinguished Writer in Residence at New York University’s Carter Journalism Institute, and vice president of the Association of Health Care Journalists.
How did you first become aware of the world of retractions?
I was deputy editor at the Scientist magazine for six years (2002–08). Retractions were rare, but when they happened there was often an interesting story behind them.
How many retractions are there?
Around 500–600 per year, 5,000–6,000 in total, although there were close to 700 in 2015. The rate has gone up dramatically in the last 15 years.
How did Retraction Watch start?
I got to know Adam Marcus, a medical journalist who had broken a few big retraction stories
, in particular that of anaesthesiologist Scott Reuben. We’d share details about different cases we saw, about the stories, the ethics and the fallout. I said, ‘Let’s start a blog,’ to which Adam replied, ‘Sure, whatever that means.’
What’s next?
Our next big project is creating a database of retractions. A lot of people are amazed that one doesn’t already exist. You could of course cobble together your own, but you wouldn’t have consistency or, importantly, the real reasons for the retractions.
For example, it used to be thought that fewer than half of retractions were due to fraud or misconduct, but we now know that’s not the case because estimates were relying on retraction notices. A 2012 paper used RW and other sources to estimate that two thirds of retractions are down to misconduct, which has changed our understanding.53 The database will allow new work like that to take place and let us analyse patterns.
How are retraction notices misleading?
Retraction notices are often simply unreliable. They vary greatly from journal to journal. Sometimes they say literally nothing, other times they obfuscate the true reasons for the retraction. Overall they don’t give a clear picture, so when people look at retraction notices and try to understand the phenomenon, they are likely to be misled.
What’s the best retraction notice you’ve seen?
There are plenty of amusing instances where journals dance around the truth – we have even published a couple of lists of ‘plagiarism euphemisms’. Our favourite was a clear case of plagiarism where the journal ventured that this was ‘an approach to writing’. Adam commented that this was an approach to writing in the same way that showing up to a bank with a gun is an approach to banking.