Few tasks come as thankless as asking a scientific journal editor to correct a bad study. “I think that it [is] time to stop this never-ending story,” the editor in chief of a diabetes journal told us last year, refusing our request for the data behind a study that we deemed to have fatal statistical errors. As noted by Retraction Watch, in an earlier e-mail to us, the corresponding author of the paper said, “sharing data with a third party would breach the study [participants’] consent and European rules on data protection.” But such errors would invalidate preliminary positive results for a Web-based clinical decision support tool intended to help diabetes patients. Those are the people we were concerned about.
Unfortunately, such irresponsibility is all too common.
Science informs our medical care, flies us through the skies, and keeps us safe while driving. And yet, because it’s done by people, mistakes can be made. After all, to err is human.
On supporting science journalism
If you’re enjoying this article, consider supporting our award-winning journalism by subscribing. By purchasing a subscription you are helping to ensure the future of impactful stories about the discoveries and ideas shaping our world today.
Scientific papers are riddled with errors; for example, as many as 30 percent of cluster randomized controlled trials for childhood obesity may have been analyzed incorrectly. In psychology, 20 percent of studies or more may have inconsistencies with their basic descriptive statistics that shed doubt on the reliability of their findings.
When mistakes happen, the costs can be high. Researchers say that science is self-correcting, but often it isn’t or it is unforgivably slow. For example, in the early 19th century, based on obviously bad sampling, physicians misdiagnosed many healthy infants with “enlarged” thymuses and mistakenly believed they were responsible for sudden infant death syndrome (SIDS) by occluding the trachea, leading to a century-long medical misconception. Radiation therapy was then used to shrink thymuses for thousands of children, which ultimately caused increased rates of thyroid and breast cancer, resulting in over 10,000 deaths. The error was finally recognized and corrected in the 1940s. However, a nationwide campaign was only launched in 1977 by the National Cancer Institute to warn the medical community and alert the public.
Serious errors occur in social science publications, too. Economists Carmen Reinhart and Kenneth Rogoff published influential research in 2010 suggesting that when a country’s debt surpasses 90 percent of its gross domestic product, economic growth declines. Broadly cited by political leaders, this finding famously supported austerity policies worldwide. Independent researchers later uncovered significant errors in Reinhart and Rogoff’s calculations. The blunder omitted key data from five countries, reversing the findings to show an average growth increase, rather than decline, at high debt levels. Reinhart and Rogoff acknowledged the error but argued against other criticisms of their analysis and maintained that their main findings generally held. This case raises another all-too-common irresponsibility we encountered in our diabetes study correction quest, where authors do not proactively disclose data and editors fail to require its availability. And while Reinhart and Rogoff had initially shared some of their data analysis details publicly, it was insufficient to fully reproduce and check their findings. Had they shared their data completely, the error could have been identified and corrected sooner, potentially prior to publication.
We end on an egregious saga of scientific falsification and fabrication that went uncorrected for decades in the fields of Alzheimer’s and Parkinson’s disease research. Eliezer Masliah, former director of the National Institute of Aging’s neuroscience division and a neurodegenerative disease specialist, was found guilty of scientific misconduct following a government investigation. As a result, two papers were retracted because of what appeared to be doctored images. A news story in the journal Science after an independent investigation reported that as many as 132 papers from 1997 to 2023 are suspect; several of those papers were very influential in the neurodegenerative disease field and in therapeutic drug development. A quarter century of Alzheimer’s research may have gone down the wrong trail to potential remedies because of this disaster.
Many of us have had or will watch a loved one go through the slow, debilitating, and heartbreaking loss of self and others that Alzheimer’s brings. Tragically, current treatments offer limited relief. Despite countless researchers’ efforts and millions of dollars, progress has been hindered by chasing false leads while people are lost to the scourge of Alzheimer’s.
The odiousness of knowingly delaying the correction of error-contaminated science came into play in the 2023 case of Marc Tessier-Lavigne, a distinguished scientist in the neurodegenerative field and a former president of Stanford University. Tessier-Lavigne resigned after an expert panel review found several of his group’s published papers to contain manipulated images. Although Tessier-Lavigne himself was exonerated of research misconduct, the review concluded that as supervisor of the research group Tessier-Lavigne “failed to decisively and forthrightly correct mistakes in the scientific record.” This case highlights the dangers when scientists (especially senior scientific leaders) do not fully stand up for the rigor and integrity of science. Just as we experienced, it also highlights the failure of editors and journals to maintain the scientific integrity of the scientific literature. It turned out that Tessier-Lavigne attempted to make corrections for papers in Cell and Science, but Cell initially said it was not necessary, and the correction in Science initially was not published, consequent to an editorial error.
Such delays and failures to correct known errors not only harm the public and stall progress but can also erode public trust in scientific research. This is why scientists and the journals in which they publish must unwaveringly commit to correcting errors. Scientists must uncompromisingly commit to rigorous research and accurate and trustworthy reporting; all of our welfare hangs in the balance.
It’s not all bad. We know of the errors cited here in part because science as an enterprise takes them seriously, even if individuals did not. Science has always been about self-improvement. It is a wellspring of innovations that learn about the world and leverage those insights to help us all.
We are still hopeful. When bad science is left unchecked, it harms us all. And we need to do better. How much more progress could we make and how many more lives could we save if we do?
This is an opinion and analysis article, and the views expressed by the author or authors are not necessarily those of Scientific American, the authors’ institution, or any other organization.
