How many Internet, Publication it is the form of place where it is advisable to be anonymous. There, under randomly assigned taxonomic names, equivalent to Actinopolyspora bistrensis (bacteria) i Hoya camphorifolia (flowering plant), “detectives” meticulously document errors within the scientific literature. While they write about all varieties of bugs, from botched statistics to nonsensical methodology, their collective knowledge is manipulated images: protein clouds which have suspiciously sharp edges or an identical cell arrangements in two supposedly different experiments. Sometimes these irregularities mean nothing greater than that the researcher tried to brighten the figure before submitting it to the journal. But they’re still raising red flags.
PubPeer’s unique community of science detectives has created an unlikely celebrity: Elisabeth Bik, who uses her incredible sharpness to duplicate images which could be invisible to virtually every other observer. Such duplication could allow scientists to conjure up results out of nowhere by Frankensteining parts of multiple images together, or claiming that one image represents two separate experiments that produced similar results. But even Bik’s supernatural eye has limitations: it’s possible to fake experiments without actually using the identical image twice. “If the 2 pictures overlap a bit, I can nail you,” he says. “But should you move the sample just a little further, I am unable to find any overlap.” When the world’s most visible expert cannot all the time discover a scam, fighting it—and even investigating it—could seem unimaginable.
Nevertheless, good scientific practice can effectively limit the impact of fraud – i.e. outright falsification – on science, whether or not it’s ever discovered. Fraud “cannot be excluded from science any greater than murder cannot be excluded in our society,” says Marcel van Assen, principal investigator on the Meta-Research Center on the Tillburg School of Social and Behavioral Sciences. But as researchers and advocates proceed to push science to be more open and unbiased, he says fraud “will likely be less prevalent in the long run.”
Alongside detectives like Bik, “meta-scientists” like van Assen are world experts in fraud. These researchers systematically follow the scientific literature, attempting to be certain that it’s as accurate and robust as possible. Metascience has existed in its present incarnation since 2005, when John Ioannidis, a once-praised professor at Stanford University, recently got a nasty fame for his views on the Covid-19 pandemic, equivalent to fierce opposition to lockdowns – published an article with the provocative title “Why Most Published Research Is False“. Ioannidis argued that small sample sizes and bias mean that erroneous conclusions often find yourself within the literature, and these errors are too rarely discovered because scientists would slightly pursue their very own research programs than try to duplicate the work of colleagues. Since this text, meta-scientists have refined their techniques for studying bias, a term that covers every little thing from so-called “questionable research practices” – not publishing negative results or using statistical tests over and yet again until you discover something interesting, equivalent to… to outright fabricating or falsifying data .
They test the heartbeat of this bias by looking not at individual studies but at general patterns within the literature. When, for instance, smaller studies on a specific topic show more dramatic results than larger studies, it could be bias index. Smaller studies are more diverse, so a few of them will by accident turn into dramatic – and in a world where dramatic results are preferred, these studies will likely be published more often. Other approaches have a look at p-values, the numbers that indicate whether a given result’s statistically significant or not. If in all of the literature on a given research query, too many p-values seem significant and too few don’t, then scientists may use questionable methods to make their results seem more meaningful.
But these patterns don’t indicate how much of this bias will be attributed to fraud slightly than dishonest data evaluation or innocent errors. In some ways, fraud is inherently unmeasurable, says Jennifer Byrne, professor of molecular oncology on the University of Sydney, who worked to discover potentially fraudulent documents within the cancer literature. “Deception is about intent. It’s a psychological frame of mind,” he says. “How do you infer frame of mind and intentions from a printed article?”
