Fake studies abound in scientific journals

Have a “ghostwriter” for cutting-edge scientific publications. It’s possible and even widespread, according to a recent study.

About one in five articles would thus be a potential false publication in the biomedical field, according to a study by Bernhard Sabel, of the Institute of Medical Psychology in Magdeburg, Germany.

The latter, with Emely Knaack, Mirela Bilc but also Gerd Gigerenzer, from the Max-Planck Institute for Human Development in Berlin, has reviewed more than 15,000 publications in biology and medicine. They attempted to more easily identify potential fake publications (referred to as “red flags”) through several indicators such as the absence of a professional email ociated with a hospital for the author and the absence of an international partner. .

“Publish or perish”

As the authors point out, scientists and students are under strong pressure to publish in scientific journals, both by their universities and the public authorities. It is the well-known adage of scientists “publish or perish”. But even more, the number and impact of publications can have a direct effect on salaries, promotions etc. especially in some countries like China.

In this context, “paper mills” (literally “paper mills”) have developed to ist scientists, by producing scientific submissions, for remuneration. Writing help companies abound on the web. And some use artificial intelligence.

“Paper mills”

Researchers quote prices of a few thousand dollars, depending on the services requested, including fake experiments.

The analysis shows that the number of potentially problematic publications has almost doubled between 2010 and 2020. Taking into account a margin of error, Berhnard Sabel therefore estimates that the proportion is around 20%, or more than 300,000 articles per year! “Fake scientific publications are probably the biggest scientific scam of all time, wasting financial resources, slowing medical progress and potentially endangering the health of populations,” says Bernhard Sabel.

The rise of generative AI may further aggravate the phenomenon, according to him. ChatGPT-type conversational robots indeed generate texts that seem credible in a few seconds and sometimes only a careful eye can detect errors, which are nevertheless numerous.

Most of the false publications come from China, Russia, Turkey, knowing that the notion of false covers the data and/or the conclusions. “It’s less common in Europe,” he says.

Why such a deluge of “fake news” in science? The specialist explains it above all by a lack of knowledge of the phenomenon. “The scientists who evaluate the work don’t necessarily realize that,” he says.

Not so surprising, according to Arnaud Mercier, professor at Paris Panthéon-as, who has written extensively on “fake news”: “international journals have very standardized formats, which can generate a premium on formalism. Those who reread will sometimes judge the relevant articulation of the reasoning more than the data contained. And in academia, essment is based on trust”.

We also remember of the “Lancet Gate”, at the height of the health crisis, which had led to the withdrawal of an article on hydroxychloroquine.

A real business

Anyway, there is a real business for these “paper mills”, the study estimating annual revenues of around 3-4 billion dollars. This, without taking into account the so-called “predatory” publications (reviews taking on the appearance of a scientific publication, but without presenting the guarantees).

Several prestigious publishers (Wiley, Elsevier, Springer Nature etc.) have also taken up the subject, launching common detection tools . And, Hindawi closed at the beginning of May four journals likely to be compromised by these “paper mills”.

However, the study by German specialists has not been peer-reviewed and at the “preprint” stage (therefore to be taken with reservations). And, she points out that the red flag is not a definitive indication of fraud, as it can misidentify a substantial number of genuine items.