I read an article on Yahoo News this past week about an increase in the number of fraudulent scientific papers published over the past thirty years. The article was written by Seth Borenstein of the Associated Press, and he says that since the mid-1970s, the number of retractions of scientific papers for outright fraud has jumped from around 10 per 1 million papers published in 1976 to over 90 by 2007. On the face of it, that’s quite a jump (nine-fold).
Of course, that’s just the number of papers retracted because of fraud. (I assume it doesn’t count papers retracted for other reasons such as unintentional errors.) How many other fraudulent papers are still out there because they haven’t been discovered is unknown and aren’t counted. I don’t have any real figures to back me up here, but I assume there’s been a substantial increase in the number of scientific papers published in that time, and that could be part of the reason. I seriously doubt the increase in number of papers has been as big as a nine-fold jump, though. The increase may be partly due to better detection of fraudulent papers. Overall, though, there may be a real increase.
The reasons for the increase aren’t immediately obvious, but I have several ideas. As a virology researcher for many years, I never knew anyone I could honestly say committed fraud–of course not, anyone faking results is going to keep it quiet. Apparently, the greatest part of the number of fraudulent papers is composed of a few habitual, incorrigible scientists so the chances of meeting one are slim. I’ve heard of researchers who published dishonest papers while working at one institution, only to move to another one when their deceit was revealed and continue to publish faked results at the new place. But those are rare indeed.
Possibly the major reason for committing fraud in science is the competition for good jobs. When I was still working, competition was fierce. The number of new PhD’s that graduated each year was a lot higher than the number of openings for faculty level, tenure-track positions. For many new PhD’s, that’s the holy grail, and it was for me. Most new PhD’s took post-doctoral training, honing their skills at doing research under an established researcher, either at a medical school or in industry. But for many of us, it stopped there. I took repeated post-doctoral positions (as well as their allied cousin, the research associate) and didn’t get a faculty position until I was in my late thirties. That position, unfortunately, didn’t last very long, and I was back out on the street looking for another post-doc/research associate position. So I know what it’s like being forced to get results in order to get another position. The temptation to get outstanding results is tremendous. Some give in and fake results.
So, too, is the temptation of a faculty member to get recognized. Most people I knew would certainly would have liked to be well-known in their field and perhaps garner a few prizes along the way. The No-bell rings in the head of almost every scientist; the temptation is there. Most don’t capitulate to the temptation and they stay the true course. Only a very few do, but that has the disastrous consequences of giving a black eye to the whole field. As Borenstein points out, the publication of false data can have repercussions well outside science. He gives the example of the publication of a paper that linked autism to childhood vaccinations. That study was later found to be totally fraudulent, and the paper was retracted. The author of the paper was banned from research in Britain. But to this day, mothers of young children still refuse to have their children vaccinated because they’ve heard there’s a link between vaccination and autism. The damage has been done, and nothing anyone says can convince those mothers otherwise.
There will always be unscrupulous scientists, just as there are dishonest practitioners in any occupation, yes even law and politics. Only good peer review of scientific work before it’s published and sent out to the world at large will minimize the effects.