Scientists reveal what they learnt from their biggest mistakes
How retractions can be a way forward.
3 March 2020
Be it a botched experiment or a coding error, mistakes are easily made but harder to handle, particularly if they find their way into a published paper.
Although retracting a paper due to an error may not seem a desirable career milestone, it is seen as important for building trust within the research community and upholding scientific rigor.
A 2017 study found that authors who retract their papers due to a mistake earn praise from peer-reviewers and other researchers for their honesty.
Below are four lessons from researchers who have retracted flawed papers.
1. Pay it forward
Nobel-winning chemist Frances Arnold caused a stir when she announced that she was retracting her own paper in January.
It is painful to admit, but important to do so. I apologize to all. I was a bit busy when this was submitted, and did not do my job well. https://t.co/gJDU0pzlN8— Frances Arnold (@francesarnold) January 2, 2020
The Science paper, published in May 2019, described a novel method for creating beta-lactam rings, which play a key role in antibiotics.
But when Arnold’s group at the California Institute of Technology failed to replicate the results due to data missing from one of her student’s lab notes, she requested a retraction from the journal.
“I did not want anyone to waste their time trying to reproduce our results,” says Arnold.
While she was widely praised for her candid apology, it was another researcher who inspired Arnold to be upfront about her mistakes.
“Someone I admire retracted a very important paper when I was a young scientist,” says Arnold. “I wanted to pay that lesson forward.”
2. Act quickly
Just one month after Diego Forni published a paper on the evolution of two types of herpes viruses, he received an alarming email.
A researcher in the United States pointed out to Forni that the viral gene sequences his team had used were incorrect. The researchers had retrieved the genomes from GenBank, a database containing sequences for over 300,000 organisms. They published their results in Virus Evolution in July 2019.
When Forni realized that his findings were unreliable, he knew he had to retract the paper quickly. “There were already a couple of citations,” says Forni, an evolutionary biologist at the Scientific Institute for Research, Hospitalization and Healthcare (IRCCS) in Italy. “I figured the faster I acted, the better.”
While the faulty sequences were due to an error in the database, Forni says that the experience has reminded him to check public data more carefully, “especially if the results are not what you expected.”
3. Be honest
Kate Laskowski started her year by retracting her 2016 paper on spider social behaviour from The American Naturalist.
Laskowski, a behavioural ecologist at the University of California, Davis, was perplexed when her colleague pointed out sections of duplicated values in data supplied by her co-author Jonathan Pruitt at McMaster University in Canada.
I'm starting the year off with something I didn't expect to ever do: I'm retracting a paper. I recently discovered major problems in the raw data associated with it and so the results shouldn't be trusted. https://t.co/fihzYg7O34 pic.twitter.com/1Rq87kY0QY— Kate Laskowski (@KateLaskowski) January 17, 2020
But in November 2019, Laskowski took a closer look at the files and realized that the duplications could not be explained by Pruitt’s experiment design.
“That was a dark day,” says Laskowski. “If we couldn’t explain it, we couldn’t trust it.”
The same issue also appeared in other work she had published with Pruitt in 2014, leading to two more retractions from Proceedings of the Royal Society B and Biology Letters. Pruitt’s institution is currently investigating allegations that he fabricated data in at least 17 papers he co-authored.
Although Laskowski wrote in a blog post that the experience has been “agonizing”, she says that retracting the papers was the only way forward. It has also reinforced the importance of being transparent at every stage of a research project, she adds.
“The point of a retraction is to correct the scientific record,” says Laskowski. “You can’t get in trouble for being too honest.”
4. Embrace your mistakes
In 2016, Leonhard Schilbach and his team concluded from an experiment that autistic traits influence how individuals make decisions, but not how they read social cues.
But when one of Schilbach’s students detected a coding error while repeating the experiment for another study, the findings didn’t hold up. The paper was retracted from Biological Psychiatry in October 2019.
“It was clear that it was the right thing to do,” says Schilbach, a neuroscientist at the Max Planck Institute of Psychiatry in Munich.
While Schilbach says that retracting a paper can be a difficult decision for researchers, particularly those with young coauthors who have few publications, he adds that scientific rigor should take precedence over prestige.
“Science needs to be about learning new things and understanding how the world works,” says Schilbach. “That’s not happening if our analysis is flawed.”
It’s also important to see mistakes as opportunities to improve work practices, whether quality control in the lab or double-checking data with an outside researcher before publication, says Schilbach.
“You need to know what to do to solve the problem without thinking that it will ruin your career,” says Schilbach. “Find a way to embrace mistakes, as they can help you do better science.”