In 1998, Andrew Wakefield published a study entitled Ileal-lymphoid-nodular hyperplasia, non-specific colitis, and pervasive developmental disorder in children. In the article, Wakefield and his co-authors argue that there is a link between vaccination for mumps, measles and rubella, and autism in children. Years later, it became known that Wakefield had received GBP 480,000 for the results of the study and had concealed this. Today, the study has been withdrawn and Wakefield is no longer a doctor. However, the content of the “study”, which was not a study, still has an impact today.
Withdrawing a paper. Is that possible?
Articles are withdrawn for various reasons. The withholding and concealment of funding bodies or the concealment of conflicts of interest is one of them. The most common reason is honest mistakes in data collection, analysis, or interpretation. As with any other work, these can also happen in research. However, there is also deliberate data manipulation and plagiarism. AI could play a role in the future. People whol request the retracttion of articles are authors who notice inadequacies in their own work (as in the case of Shaawna Williams in The Scientist), colleagues who notice errors, and editors who become aware of inadequacies during public discussions.
10,000 retracted publications
The retraction of articles is a development of recent years and part of the success of the Retraction Watch database. In 2023, more than 10,000 scientific articles were retracted; 10,000! And it’s not getting better. About 60% of retractions, at least according to Jeffrey Brainard and Jia You in a paper on the added value of strategic retraction tracking, are the result of fraudulent intentions by authors. Dimensions already hinted at by Ivan Oransky and Adam Marcus in an article in The Guardian. The bright spots? About 500 authors out of more than 30,000 were responsible for about a quarter of all retractions in 2023. So there are few who produce many retractions. At the same time, more and more publishers have recently jumped on the bandwagon and are now looking for erroneous publications themselves. For example, as Alison McCook reported for Science.org, in 2018 the publishing house IEEE retrospectively reviewed its articles for the period between 2009 and 2010 and removed more than 7,000 publications from its own database.
Retractions: an issue of credibility
With the new verifiability and the demand on publishers to fulfil their obligations, the career pressure on the credibility of scientists has also increased. The Scientist publishes an annual list of the most important retractions in the sciences (2021, 2022). The community has also become more aware and helpful – a “stronger systems hypothesis” advocated by Danielle Fanelli. Titles with particularly spectacular announcements have recently had to face scrutiny. Following the announcement by Ranga Dias et al. that they had found the first superconductor as a solid at room temperature, the article was first picked apart in the preprint within a few months and then withdrawn by the publisher in 2023 at the request of the co-authors. It was Dias’s third retraction, and the first to make waves in the media: Science reported it, the New York Times reported it, the Wall Street Journal reported it. This tarnishes the credibility of individuals, but strengthens science and its community.
What’s next?
It is normal to make mistakes. It is not surprising that fraud occurs. In addition to ethics and morality, there is also competition, careers and unfair means in science. What is not right, and highly problematic for society, is the long wait and lack of information for those who have cited a retracted paper and have not noticed the errors or fraud. Should retracted papers be deleted or made available to posterity in a labelled form?
As a result, there is still room for improvement in retraction processes. Ivan Heibi and Silvio Peroni, for example, in their article published in Digital Scholarship in the Humanities, analyse the retraction notice and find that there are serious differences in both content and metadata. Furthermore, according to the authors, retractions are difficult to find because of their negative connotations. It is therefore advisable to continue to work not only on the review of articles and their content, but also on the approach and publicity. This is a call not only to publishers and the research industry, but also to all those who support the research industry in publishing and disseminating knowledge that is considered reliable.
