Thursday, March 4, 2010

Climate scientists must be ruthlessly honest about data

If we want the public to continue to trust us as scientists, we must be absolutely open and never resort to spin or PR
Porters carry cores of ancient glacial ice down from the 6542-meter summit of Mt. Sajama in Bolivia. Photograph: George Steinmetz/Corbis

I'm not a climate scientist, but I am concerned about the reputation of science and scientists. One motive for going into science for me was that it is one of the few jobs where you get rewarded for telling the truth.
So it was painful to watch the trust of the public in science, already dented, taking another crushing blow when the emails stolen from the University of East Anglia were revealed. We'll probably never know exactly what the emails meant, but we can say that the matter was handled very badly indeed. Phil Jones, head of the Climatic Research Unit, should have been immediately on every TV station, explaining what he meant. By going to ground, and by denying Freedom of Information Act requests, the university gave the impression of guilt, quite regardless of whether there is really anything to hide. That brings the university into disrepute: it is a matter for resignations.
I have never come across anything in my own field that would qualify as fraud, or even dishonesty (well just once it was close), and I have never been asked by an editor to come to a particular decision when reviewing a paper. Our analysis programs are free, on the web.
That is why I was deeply shocked when Jones told the Commons science and technology committee that practices like keeping original data, and analysis programs, secret were "standard practice" among climate scientists. "Maybe it [openness] should be, but it's not." The Institute of Physics submission to the parliamentary inquiry which spoke of "worrying implications ... for the integrity of scientific research in this field" was damning but spot on, and a credit to science.
A recent analysis of verified cases of misconduct in the US suggested that one in 100,000 scientists per year are guilty, but other ways of counting give larger numbers. For example, if asked, around two in 100 scientists claim to be aware of misconduct by someone else. The numbers aren't huge but they are much bigger than they should be.
Anyone can be wrong with no trace of dishonesty. But when that happens, others soon find the mistake. It is that self-correcting characteristic of science that keeps it honest in the long run.
What gives rise to dishonesty? One motive is money. Grants depend on publication and studies funded by industry tend to be biased in favour of the result the sponsor wants. The other reason is presumably the human desire to win fame and promotion.

It is no excuse, but it is perhaps a reason, for misconduct that the pressure to publish and produce results is now enormous in academia. Even in good universities people are judged by the numbers (rather than the quality) of papers they produce and by what journal they happen to be published in. Bibliometrists are the curse of our age.
Vice-chancellors and research councils provide a strong incentive to do poor, over-hurried and occasionally dishonest science. Perhaps the surprising thing in the circumstances is that there is so little fraud. The very measures that aim to improve science actually have just the opposite effect. Application of Thatcherite principles to science results in dishonesty, just as it does among bankers. That is what happens when science is run by people who don't do it. Read more.

No comments:

Post a Comment