Last week at Retraction Watch (RW) there was a post on a paper in elife by Andrew Stern, Arturo Casadevall, Grant Steen, and Ferric Fang that put a figure on the cost of research misconduct, using misconduct officially identified by the ORI.
A number of comments on RW, including my own, highlighted that the study necessarily (because it had to use hard data) resulted in an underestimate of the financial cost of misconduct by a least a factor of 10. I thought it useful to summarise my views in what follows.
It is obvious from the number of papers that are retracted for what is data fabrication, e.g., image manipulation, data duplication and re-use of data to describe different experiment, that ORI only investigates a small proportion of what is likely misconduct. Moreover, considerable misconduct is beyond the time limit of ORI. There is also misconduct elsewhere on the planet.
A while back I did a rough calculation based on the total number of retractions at RW and the fraction that were for what we can term “correct” reasons, that is a genuine mistake, rather than data fabrication. These are categorised under the “Doing the right thing” tag at Retraction Watch and make up less than 5% of all retractions. Of course RW tracks all retractions, not just ones with US authors, but RW has tags that would allow this calculation to be done afresh for papers with US authors. However, the costs of misconduct to the US do not arise only from papers with US based authors.
So my 10-fold multiplier is a guesstimate, based on less than 5% of papers at RW featuring for the right reasons, rather than misconduct, and the number of papers that are corrected, rather than retracted. Given that several such papers may be authored by the same PI, I ended up with 10x.
So the financial cost is higher than estimated in the elife paper.
In terms of human cost, which in my view is far higher than the financial cost, we can consider (in no particular order):
– Some students and postdocs will accept the poison of misconduct dished out by the PI gladly, since this allows them to progress their careers. The human cost here is that we train charlatans, who go on to teach and train other young people.
– Other students and postdocs will rightly refuse to take this poison. This is likely to drive them out of science, since the weak have no chance against the powerful. There are plenty of instances of this documented in various blogs across the web.
-The desperation induced in a young scientist trying to replicate fraudulent work published in a “top” journal. I have had direct experience of this many years ago at another institution and it really knocks your self-confidence and self-esteem. I would count myself as particularly resilient. My view is that this is another route for driving inexperienced people out of science, unknowingly give them a project that is in fact not possible.
– Send a whole bunch of people in labs around the world barking up the wrong tree, costs resources across many labs and leads also to problems with the actual young researchers attempting to do the experiments. This may also break a PI in terms of acquiring outputs required for subsequent grant funding.
-In the extreme we have suicide. Jonathan Eissen’s post on the subject of the suicide of his father, the result of his being requested to investigate potential misconduct, is incredibly moving. Most recently we have the suicide of one of the authors on the STAP cell papers.
-Fraudulent clinical trials can lead to patient death. There are well-documented cases.
So the cost is higher and may not be possible to quantify. It is a cost that we should not incur.