Most of us are no longer surprised to learn that people from politicians to priests lie and cheat. Scientists do too. Take for example, the notorious case of Hwang Woo-Suk, who claimed to have cloned human embryonic stem cells, a lie so spectacular it was sure to be exposed with time.
But low-level fudging is harder to detect: the subtraction of a data point to enhance the appearance of a graph or to render results statistically significant, for example, or Photoshopping a jpeg to enhance a key feature.
Daniele Fanelli of the University of Edinburgh decided to quantify the pervasiveness of minor fraud among scientists, and concluded it is more common than most would have suspected.
Fanelli reviewed 18 published surveys of the matter. He found that only 2% of scientists confessed to outright fraud (such has modifying data to improve study results) at least once in their careers.
But nearly 10% fessed-up to dicey practices like “dropping data points based on a gut feeling” or “failing to present data that contradict one’s previous research.”
And survey participants proved more willing to point fingers at their colleagues than at themselves. Fully 14% of them said they’d seen colleagues falsify or fabricate data.
When the question was phrased in general terms like carrying-out experiments using substandard techniques or misrepresenting data, 46% of those queried said they saw their colleagues engaged in such misconduct.
Surprisingly, survey responders chose to intervene only half the time when they witnessed the shenanigans.
Come to think of it, the findings of Gregor Mendel, generally considered to be the father of modern genetics, were impossibly accurate given what we now know of his methods.
Fanelli’s write-up appears in the Public Library of Science.