Like many network, PubPeer is where you might want to stay anonymous. There, under randomly assigned taxonomic names like Actinopolyspora biskrensis (bacteria) and Hoya camphor (a flowering plant), the “detectives” meticulously document the mistakes in the scientific literature. Although they write about all kinds of errors, from messy statistics to pointless methods, their common expertise is in manipulated images: protein clouds displaying sharp edges suspicious manner or the identical arrangement of the cells in the two experiments was said to be distinct. Sometimes these anomalies mean nothing more than a researcher trying to beautify an image before submitting it to a journal. But they still raised red flags.
PubPeer’s rare community of scientific detectives has produced an unlikely celebrity: Elisabeth Bik, who uses her uncanny acumen to copy image in place that would practically be invisible to any other observer. Such duplicates could allow scientists to elicit results out of thin air by Frankensteining parts of multiple images together, or claiming that one image represents two separate experiments that produce similar results. But even Bik’s superhuman eye has limitations: It’s possible to fake experiments without actually using the same image twice. She said: “If there is a slight overlap between the two photos, I can understand you. “But if you move the sample a little further, I won’t find any overlap.” While the world’s most famous expert can’t always identify cheating, fighting it—or even studying it—seems impossible.
However, good science practices can effectively reduce the impact of fraud—that is, outright falsification—on science, whether it is detected or not. “Cheating cannot be ruled out from science, just as we cannot rule out murder in the case of science,” said Marcel van Assen, principal investigator at the Center for Meta Research at the Tillburg School of Social and Behavioral Sciences. his society”. But as researchers and advocates continue to push science to become more open and fair, he said, cheating “will be less common in the future.”
Besides detectives like Bik, “supernatural scientists” like van Assen are the world’s expert con artists. These researchers systematically monitor the scientific literature in an effort to ensure it is as accurate and robust as possible. Superscience has existed in its present incarnation since 2005, when John Ioannidis, a once-acclaimed Stanford University professor who recently fall into disrepute for my point of view about the covid-19 pandemicsuch as the outcry against the lockdown—published an article with the provocative title “Why are most published research results false?.” Ioannidis argues that the small sample size and bias mean that incorrect conclusions often end up in the literature and that such errors are rarely discovered, because scientists would rather continue their research program. their own research rather than trying to copy the work of colleagues. Since that paper, metascientists have been honing their techniques for studying bias, a term that includes everything from so-called “suspicious research practices”—such as not publish negative results or repeatedly apply statistical tests until you find something interesting — completely fabricate or falsify the data.
They captured this trend by not looking at individual studies but looking at overall patterns in the literature. For example, when smaller studies on a particular topic tend to show more dramatic results than larger studies, it can be a bias index. Smaller studies are often more variable, so some of them will be dramatic by accident—and in a world where dramatic results are favored, those studies will be published. father more often. Other approaches involve looking at p-values, numbers that indicate whether a certain result is statistically significant. If, in the entire literature on a given research question, too many p-values appear significant and too few do not, scientists may be using dubious methods to try to make their results seem more meaningful.
But those models don’t indicate how much of that bias is due to fraud rather than analysis of dishonest data or innocent errors. Jennifer Byrne, a professor of molecular oncology at the University of Sydney, who has experienced that fraud is intrinsically unmeasurable. worked to identify potentially fraudulent documents in the cancer literature. “Cheating is about intent. It’s a psychological state of mind,” she said. “How do you infer mental state and intention from a published article?”
