To outsiders, the idea that scientists would forge results might seem counterintuitive. After all, is science not supposed to be about inching ever closer, through endless iterations, to a truer picture of our world? Having worked in academia for ten years I know that the truth is a bit more nuanced. Even so, Fraud in the Lab, written by French investigative journalist Nicolas Chevassus-au-Louis, was a shocking, damning and eye-opening exposé.
Originally published in French in 2016 as Malscience by Editions du Seuil, this book has now been translated into English. As I hope to show below, I think this was a decision we should all be grateful for, as this book deserves wide reading by scientists everywhere.
Fraud in science has been around for a long time, and Chevassus-au-Louis opens with four high-profile cases from the 1970s and ’80s that generated wide public attention. He also touches on one of the most classic cases of data doctoring, that by German zoologist Ernst Haeckel, but also the less well-known question of whether Gregor Mendel’s Law of Segregation, which became a cornerstone of genetics, was based on data that were a bit too good to be true.
But, he claims, judging by the explosion in retractions of articles from scientific journals, the issue seems to have grown worse in recent decades. Although the biomedical sciences seem particularly prone to fraud due to the commercial interests involved and the resulting fierce competition and secrecy between research groups, Chevassus-au-Louis shows the problem occurs in virtually all fields of research. The other indication is the reproducibility crisis. Many papers describe research results that, when attempted, could not be replicated, something which should be the gold standard in science.
“Fraud in science has been around for a long time […] but, [Chevassus-au-Louis] claims, judging by the explosion in retractions of articles from scientific journals, the issue seems to have grown worse in recent decades”
So, what forms can fraud in science take? Many will not be that surprising: the faking or doctoring of data, the selective inclusion of only those data that support your hypothesis, the entanglement of corporate interests with those of science. These are the obvious ones.
But Chevassus-au-Louis probes deeper, forcing the reader, especially those working in academia, to engage in painful introspection by claiming that all scientific papers are cases of storytelling and beautification: “the story is often too good, too logical, too coherent”. There is a large grey area of not-quite-fraud but not-good-practice either that many scientists engage in. Omitting outliers in your data, p-hacking (also known as as “torturing the data”, such as stopping data collection once you have a statistically significant result), or subtly restating your hypotheses after the data has been collected to tell a better story.
He casts the net wider by covering other forms of unethical behaviour and misconduct: plagiarism, the stealing of authorship (sometimes by peer reviewers who anonymously read a manuscript), the publishing of the same results in multiple journals. Equally interesting were the chapters on the dark side of open access publishing where researchers rather than taxpayers foot the bill of publishing a paper. This has led to the rise of so-called predatory journals. Though legit-sounding, they will publish anything for the right price (see also my review of Pseudoscience). These are particularly insidious, as it should be noted that these articles get indexed by Google Scholar and other bibliographic databases. If scientists can barely navigate the jungle of predatory journals, how on earth is the general public supposed to differentiate between good and bogus science?
“The rise in predatory journals is particularly insidious, as these articles get indexed in bibliographic databases. If scientists can barely navigate the jungle of predatory journals, how on earth is the general public supposed to differentiate between good and bogus science?”
The book is enriched with unbelievable examples. Serial cheaters who have had hundreds of their papers retracted. Chinese firms who, for the right price, will add author names to papers already accepted for publication or will find you a writer if you deliver the data. Chevassus-au-Louis repeatedly lashes out at scientific journals, revealing the shocking disregard editors at even high-profile journals such as Nature and Science sometimes show for peer review reports when rushing papers into print. The problems extend to all participants in the process of publishing scientific findings.
This has real-life costs. Take the retracted paper that made a link between the MMR (measles, mumps and rubella) vaccine and autism, which led to deaths that could have been prevented and a vociferous anti-vaccination movement that, perhaps ironically, just refuses to die. But less well known are the clinical studies that rely on poorly executed or false research, leading to an incredible waste of money and time, and sometimes the death of trial participants.
The most obvious reason behind fraud is the enormous pressure to publish (or perish), as scientists are judged by how much research they produce and how often it is cited. Chevassus-au-Louis skewers the simplistic and one-dimensional metrics such as journal impact factors and the h-index (a measure of a researcher’s productivity and citation score).
“Chevassus-au-Louis repeatedly lashes out at scientific journals, revealing the shocking disregard editors at even high-profile journals […] sometimes show for peer review reports when rushing papers into print”
But again, the book moves beyond the easy targets. Reporting fraud when it concerns your superior is challenging to say the least. There is the problem of novelty, resulting in a race to publish first, as journals show little interest in publishing papers that “merely” confirm findings, despite their huge academic value. And there is the problem of significance: studies showing no significant effects are almost impossible to publish, which means that what does not work is rarely reported, leading to the risk of people independently trying to reinvent the wheel. I was pleased to see mention of the grassroots Journal of Negative Results, which was run by some of my departmental colleagues when I was studying at the University of Helsinki.
The discussion on solutions is equally interesting: both the successes and failures of committees established in France and elsewhere to identify and sanction fraud, and the questionable idea of criminalising research misconduct and handing over cases to law enforcement. Above all, Chevassus-au-Louis argues, we need structural reform in how science is planned, executed, and rewarded, and joins philosopher Isabelle Stengers in calling for slow science. Accountability initiatives whereby researchers make public the research questions and hypotheses before doing the research seem particularly promising.
For scientists of any stripe, whether established or beginning, Fraud in the Lab is required reading. But seeing the widespread impact of science on all spheres of our lives, we would all do well to understand the nature and scale of this problem. This eminently readable book will get you up to speed and ventures well beyond the obvious problems.
Disclosure: The publisher provided a review copy of this book. The opinion expressed here is my own, however.
You can support this blog using below affiliate links, as an Amazon Associate I earn from qualifying purchases:
Other recommended books mentioned in this review: