A very worrying piece was published on Roger Pielke Jr.'s blog about the rising number of retractions in scientific publishing. We know it is worrying because it followed the headline; "Surging Retractions in Scientific Publishing" and was accompanied by this graph;
It was also crossed linked on WTFUWT under the title of "The newest hockey stick". The implication is clear, science is either in crisis or heading toward one. This is supported by the graphed data and it can only be a matter of time with this increasing trend before all research is either in error or a plain fraud. With both these web sites almost exclusively reporting on climate and environmental issues their readerships bias, (that the science is flawed or fraud), is confirmed.
To be fair to Pielke Jr., in his post does state in the last line that most retractions occur in the medical sciences but Watts is not so scrupulous with his title being a clear reference to Dr. Michael Mann's 'Hockey Stick' graph of temperature reconstruction who Watts refers to as "the man who wrote the paper that established the "Hockey Stick" as the icon of misguided climate science", and in the post itself, a slur on Charles Monnett and his Polar Bear article, (which I mentioned in my last post here), by suggesting it could "become a new data point on the graph". The comments on both blogs quickly mention the IPCC and climate scientists etc.
But all this is supported by a graph of hard data isn't it?
Wait a minute...
What does the graph actually mean?
Well we already know, by careful reading of Pielke Jr's blog and by following his links that this has nothing specifically to do with the science of climatology and given Pielke’s long obsession with climate science, I wonder why he hasn't found out how many retracted papers are climate related. Could it be Not Many?
Might I suggest that it would actually appear that if a problem exists with scientific research it may mostly lie within the branch of medical research with most retractions being in that field, one which has many studies carried out or financed by drug companies on their own products?
How big a problem is it?
But the real question a sceptic should be asking is; can the increase be accounted for by other factors and how big a problem is this really? Perspective is needed. By checking the axis on the graph we can see that it is showing retractions per 100,00 publications so it can be dismissed as just being an anomaly due to more papers are being published, but the Pharmalot site cited by Pielke asks the question, is 'there is more fraud or more policing?'
This reminds me of a recent Ben Goldacre piece in the Guardian that makes it clear how important it is to look for extra factors that may influence and better explain the results. He used an example of a study, that suggested people who drink are three times more likely to get lung cancer than people who don't. But then someone asks if maybe this finding is confounded by the fact that drinkers also smoke cigarettes? That could be an alternative explanation for the apparent relationship between drinking and lung cancer. By splitting those surveyed, not only drinkers and non drinkers but also smokers and non smokers, the apparent effect of drinking has been eradicated, and this means that the observed risk of drinking was entirely due to smoking. I believe something similar is responsible the apparent Surging Retractions in Scientific Publishing.
From the data given, 73.5 percent of papers were retracted simply for error so while fraud will always be a problem in any field it seems that better policing is the real answer. It looks to me that this better policing, and even fraud detection, has occurred because of increased access to research papers due to most being published on the World Wide Web. Back in 1977 research would be mostly restricted to the printed version in the journals they appeared, with copies being held on campus libraries. But 21st century research is almost freely available on-line where it will be viewed by many more interested people capable of finding error, flaws or deliberate and unwarranted manipulation.
So I propose that the Surging Retractions in Scientific Publishing can be accounted for simply by being due to better scrutiny before the research becomes obsolete and forgotten.
Again from the graph we know that it is showing retractions per 100,00 publications and the current figure for retractions is 35 per 100,000 papers or 0.035%! Can anyone name any field or profession were serious errors are running at 0.035% and such a figure is enough to suggest a crisis? To me the number of retractions is small enough to be considered uninteresting, and very few papers actually are retracted. This is in fact a good thing, a very good thing. It shows that science, which ever field of it we are interested in, is working well. The fact is almost all retractions are made at the request of authors discovering their error, or in the case of fraud, discoveries by other scientists or scientific literate people who notify the publishers. This is how science is supposed to work.