By Michael Banks
Everyone hears the big stories of fraud in science. Indeed, a feature in last month’s Physics World (May 2009 pp24–29) documented the rise and fall of Jan Hendrik Schön who published a number of papers in prestigious journals such as Nature and Science that have since been shown to include fabricated data.
But how common are the smaller cases of misconduct? It is not easy to get accurate data about how common misconduct is within the research community. One could, for example, look at the number of paper retractions in journals, but these only include cases that have been discovered, and possibly not all retractions are based on fraudulent work.
So who better to ask than the researchers themselves if they have ever fabricated of falsified data? Well, Daniele Fanelli from the University of Edinburgh in the UK has analysed over 20 surveys in which scientists were asked a number of questions about scientific misconduct including if they had ever made up data points or distorted their results.
Fanelli found that, on average, 2% of scientists admitted to fabricating, falsifying or have modified data at least once during their careers. While over a third of researchers said they have published papers with “questionable research practises” such as not including data in a publication that may counter their conclusions or dropping data points from analysis because they were deemed “inaccurate”.
Fanelli also analysed surveys that asked researchers about the practises of their colleagues. He found that 14% said they knew someone who had fabricated data, while a massive 72% said they knew someone who has published papers with “questionable research practises”.
As Fanelli points out, it is sometimes difficult to interpret what researchers may term as misconduct, “the fuzzy boundary between removing noise from results and biasing them towards a desired outcome might be unknowingly crossed by many researchers”.