Most warm weekends, you can find me in Central Park by the Tavern on the Green playing the beautiful game of petanque. This is the French version of bocci, only unlike the Italian game, which uses effeminate wooden balls, we use manly balls of steel.
The game goes like this: each team takes turns throwing 800g balls six to ten meters towards a small colored ball, called the cochonette. The goal is to get as close to the cochonette as possible. When each team has thrown their six balls, we walk up to the cochonette and try to see which team’s balls is closer. Often, of course, it’s a narrow call whether my ball or my opponent’s ball is nearest.
Now, I have stood over the cochonette literally thousands of times—it helps to understand that I have perfect vision and have never needed glasses—and in a large fraction of those times I would have sworn, on my soul, that my ball was the closer of the two. Sometimes, of course, it is, but if you know me as a player, you know that is a rarity. Usually, my ball is the furthest, but it is often manifest, I pledge on my honor, that mine is best! Not only does my ball appear closer, but it is so obviously closer, that I cannot for the life of me see why there is an argument from my opponent.
But there is invariably a dispute, so out comes the stick, usually a telescoping radio antenna stripped from its base. Somebody bends down and measures the distance between all the balls and the cochonette. Once the objective results are in, there are usually groans from one side and calls of “It was obvious” from the other.
Psychologists are well familiar with this phenomena; in science it is called the experimenter effect. It describes what happens when an honest scientist carries out an experiment in the absolutely fairest way possible, looks at the results, and sees exactly what he expected to see, only to find that, later, other scientists have shown his result to be a statistical artifact or due to a forgotten, unaccounted variable. This is why, for example, double-blind trials in medicine are required, else the doctors would always find that the “active” pill beats the placebo.
You must understand that our scientist is a nice person, is kind to small animals, pays his taxes, and votes the proper way. He, with the best, and most honest, intentions carries out his experiments in the most meticulous manner he knows how. He is unbiased and exceptionally bright and in no way delusional or politically motivated. Only, it turns out, he is far too confident in his results. This is no dig against our scientist: most people in most things are overconfident; this is another thing that psychologists well understand.
It is true that greater than 99% of all climatologists are like our scientist, forthright, incredibly bright, and diligent. Too many “climate skeptics” have accused climate researchers as being driven by politics or by money (in the form of grants), and so seek to disregard results from these scientists on that account. But this is no different than the “green activist” denigrating findings from scientists whose research was funded by private companies. All results have to be analyzed on their own merit.
Climatologists are, I believe, too confident in their results: if there is any political temptation here, it is towards the tendency to make public statements that convey more certainty than research warrants; but there is no attempt to mislead. Without question, “activists” are annoyingly precise in their pronouncements, and since theirs is a political life, there is no temptation to which they will not give in. But many skeptics, too, could use a dose of humility. To say, for example, that “global warming is a hoax” is carrying constructive criticism too far.
Love your blog.
I doubt the incentive introduced by scientific funding would have come up if the AGW activists had not attempted to dismiss critics by claiming that they have tenuous connections to the oil industry.
Personally, I think that bias due to egos are a bigger problem than funding. Many key individuals have invested too much in the AGW hypothesis and cannot back down to matter what the evidence.
I see examples of the ego bias when I see AGW scientists trashing the work of other scientists suggest that suggests that GW may not be as serious as claimed. In an ideal world the AGW scientists should be happy to be proven wrong since much human suffering is ahead if they are right. Alas, we do not live in an ideal world.
One thing I noticed while arguing about climate trends on Tamino’s blog is that he sees a way of statistically representing what he believes is true and looks no further. I have also seen this happen in hurricane papers where, to take one example, Bayesian methods (ie rule-based) had been used to tease out a trend that clearly wasn’t there before. Of course the inference rules were biased towards the expected behaviour – that warming causes more storms – which is now not that certain. Hence stats has become more of a mathematical wrapper for preset opinions. I notice that Armstrong and others are set against significance testing as being anti-science and pro-bias. I’d agree. In fact the phrase “statistical significance” is so overused and abused, it has become a byline for “not actually obvious until you massage the numbers”. Has Statistics now fully dropped out of the Maths prospectus and should it be regarded more as a social science? Do we have to ask someones political leaning in order to determine the credibility of their stats? That you have come out to be politically to the right certainly makes me distrust your numbers as much as the left-leaning Tamino’s and I’ve not even read your papers yet. Personally I think that being apolitical helps a lot in BS detection.
JamesG:
I too would wish that the term “statistical significance” be forever banished from the mathematical lexicon. This is because, read in isolation, the term has very little meaning, and that it could be, and often has been, used to mislead.
For the same reason, I also do not love “significance” testing and would instead like to see the estimated probability that a statement is true (see my postings about “Why most statistics don’t mean what you think they do”; I am slow and have not finished Part III yet).
A person’s politics should not necessarily dissuade you from examining their results. For myself, I would say that I am not “right”, and I am most definitely not “left.” I am a conservative in the sense that philosopher David Stove used that word, and for the same reasons, the first of which is that people have always, and probably will always, forecast future events poorly; and that they have too much confidence in their beliefs.
Technical note: Bayesian statistics does not mean “rule based” in the sense that you are able to prove whatever preconceptions you have. You can do that using classical, frequentist or Bayesian methods.
Pingback: William M. Briggs, Statistician » Is climatology a pseudoscience?
This is a great post. I have made a similar point (less adroitly) to dhogza at Tammy’s place, when he got all kerfuffled about accusations of fraud. And the funny thing is that my skeptics, especially hoi polloi, do froth too quickly and harshly. However, dhogza gets it to the point, where any type of inference of bias is read as deliberate lying.
Also, I would note that a LOT of the skeptics (Watts, McIntyre) are inclined both to beleive what helps them…and to refuse to admit converses when shown them. Very weasely, very pussyish.