An arms race is taking place in science on the border between errors and fraud

Of Betrayers of the Truth William Broad and Nicholas Wade wrote a classic about science fraud in 1982. In their groundbreaking book they made clear how competition, prestige and publication pressure structurally undermine the ideal of science as a rational search for truth. Since the 1980s, there has not only been much discussion and writing about science fraud, internationally and in the Netherlands, but serious work has also been done on prevention and approach. Universities were given codes of conduct, confidential counselors, reporting points and committees for scientific integrity. During that period, the academic world has improved its rules and tightened procedures. Yet the competitive global system remained intact, with rewards for numbers of publications and citations thereof by other researchers. This also meant that the incentives that could lead scientists to stretch or break rules remained.

The forms of fraud that Broad and Wade mainly described – falsified data and fabricated results – still exist, but have taken on new guises, according to the recently published book Sloppy Science. Everything about science fraud: from mistakes to fake conferences by science journalist Stan van Pelt. He describes, among other things, the rise of papermills: companies that produce complete scientific articles for a fee, often with reused or manipulated data, and where researchers can be included as authors for an additional fee. Papermills products look convincing and often pass editorial control effortlessly. Peer review is poorly suited to this: that system is intended to assess the scientific quality of submitted manuscripts and is not designed to unmask deliberate misleading. Because papermills operate anonymously and across borders, they are difficult to tackle. The damage is great: such publications are cited, end up in meta-analyses and can even influence clinical guidelines.

John Bohannan already made it in 2013 Science painfully visible how vulnerable the publication system is to deception: 157 of the open access journals he approached accepted a fabricated biomedical study. Van Pelt repeated that experiment for his book by offering a fabricated article about telepathy generated with ChatGPT and having it published in the (insignificant) medical journal Cases. It can still be read there on the website, despite extensive publicity about this fake publication.

Crosses of pea plants

With another journalistic experiment, Van Pelt makes it amusingly clear how the facade of scientificity is used to extract money from gullible scientists. He attended the conference in Budapest NeuroTalk2024Europeorganized by the Chinese company Bitcongress. An apparently neat neuroscientific conference turned out to be a messy jumble of lectures on completely different topics, from mobile phone use on buses to vibrating socks for Parkinson’s patients. The conference was a typical example of a ‘predatory conference’: high registration costs, a strangely composed list of speakers, misuse of names of renowned scientists who do not come at all and empty halls in which the audience is mainly the speakers themselves. Such fake conferences have been reported previously, including in de Volkskrant and Natureand some Dutch universities are now actively warning their employees about it. Nevertheless, Van Pelt also met two misguided Dutch scientists in Budapest.

The strong point of the book is that Van Pelt, as a PhD neuroscientist, has worked in a competitive environment, on temporary contracts, without the certainty of a permanent position. With this experiential knowledge, he knows how to give his observations good context and he also explains very well to laymen how the academic world works.

He seems less at home in the history and philosophy of science. In the opening chapter he presents Galen, Mendel and Pasteur as early fraudsters. Galen is accused of presenting monkey anatomy as knowledge about the human body, Pasteur of publishing incomplete and misleading data in his anthrax and rabies research, and Mendel of the proportions he found in his crosses of pea plants matching his expectations too perfectly. Van Pelt thus judges them according to contemporary standards, while such standards did not yet exist in their time. He acknowledges this himself at the end of the chapter, but this historical excursion mainly obscures what exactly should be understood by modern scientific dishonesty.

Van Pelt leaves the period before the astonishing fraud of social psychologist Diederik Stapel largely undiscussed

Van Pelt wrote a journalistic book, but you can also expect attention to be paid to the historical and policy context, something that is missing in his treatment of modern integrity cases. He largely leaves undiscussed the period before the astonishing multi-fraud of social psychologist Diederik Stapel (2011), while it was precisely then that the foundation was laid for the current Dutch integrity policy, with codes of conduct (2004, 2012) and the establishment of integrity committees. He makes a loose and unmotivated selection from the years after Stapel. Some controversial cases (such as those surrounding Förster and Kourtit/Nijkamp) are completely absent, others only appear briefly. He dismisses the Van den Boom affair in one paragraph, while it led to a corporate investigation, a national uprising of scientists, an integrity complaint against a scientist who spoke out in the press, the resignation of the LOWI chairman and fierce discussions about the definition of plagiarism. Van Pelt hardly talks about plagiarism and careful source accountability, while this is an urgent theme in the AI ​​era, in which the boundaries between quoting, paraphrasing and drawing on other people’s texts are increasingly blurring.

Van Pelt, on the other hand, devotes almost a tenth of his book to the Kouwenhoven case – surrounding the prematurely claimed discovery of majorana particles, which would have revolutionized quantum technology – with an abundance of details. In that case, an integrity committee ruled that the researchers had presented measurement data selectively and too favorably, possibly as a result of ‘confirmation bias’: a form of carelessness that falls under ‘questionable research practices’. No evidence of deliberate fraud was found. This case was ideal for comparison with the Buck/Goudsmit case from the 1980s, where the researchers were also carried away by the desire to make a grand discovery. However, this historical depth is not provided, including in other cases, leaving Van Pelt’s exploration of the boundaries between culpable or otherwise culpable errors and intentional fraud superficial. It is precisely these nuances that make forming judgments so difficult for integrity committees.

Image manipulation

Since the 1980s, attempts have been made to estimate the extent of science fraud. Broad and Wade stated at the time that each revealed case of fraud symbolizes a hundred thousand others that remain under the radar. However, American information scientist Eugene Garfield warned at the time that there is no reliable ‘epidemiology of fraud’. The slightly harder empirical data that do exist still does not provide a reassuring picture. ‘Fraud detective’ Elisabeth Bik (beautifully portrayed by Van Pelt) found indications of image manipulation in approximately 4 percent of 20,000 biomedical articles. Her fellow fraud detective Ben Willem Mol estimates that in his field, a third of RCTs (randomized controlled trials) are largely fabricated or unreliable. In a large-scale survey by professor of methodology and integrity Lex Bouter, 1 in 8 researchers admitted to having fabricated or adapted data, while more than half were guilty of ‘questionable research practices‘. Van Pelt states that fraud cases are just the tip of a much larger iceberg.

There are also structural problems beneath the surface that cannot be solved with an incident policy. In his final chapter, he argues for binding judgments from integrity committees, independent supervision outside the university, mandatory data disclosure and fewer competitive incentives. The recently published concept of the new Dutch code of conduct for scientists takes a much more moderate line: no inspection or harsh interventions, but more openness about interests, better reporting procedures and protection of whistleblowers. Where the code takes small steps, Van Pelt shows how big the gap is between what is happening now and what he believes is needed.

Despite the above criticisms, to which one can add the lack of an overview of sources at the back Sloppy Science a clearly and smoothly written book about forms of science fraud, which shows how vulnerable science remains to human and structural flaws.

Give a gift





Part





Email the editor