parallel wrote:As far as is presently known the BEST temperatures are close to being correct. (Assuming a global temperature means anything.)
That doesn't absolve them of not doing the statistics correctly. Wrong is plain wrong even if the error is small.
As posted before, see: http://www.bishop-hill.net/blog/2011/10/21/keen...
Doug Keena, the supposed statistical expert pretty much said that there was no valid way of building a statistical model from temperature records. Here is the not-very-helpful comment from him :
Doug Keenan wrote:Although the AR(1)-based model is known to be inadequate, no one knows what statistical model should be used. There have been various papers in the peer-reviewed literature that suggest possible resolutions, but so far no alternative model has found much acceptance.
That leaves the door open for him to criticise any and every attempt to create a valid statistical model of temperatures. Now, the question is, how much weight should be given to his criticism of the statistics when he is not prepared or able to state as how modelling such temperature series should be approached?
In other words, how wrong can you actually be when the guy criticising you is actually saying there is no right method.
Such disputes between statisticians is nothing new. In the late 1950's, Fischer, who had made a significant contribution to statistical theory, was a heavy smoker and a paid consultant of the tobacco industry. He accused Bayesians and other statisticians of dishonesty after a 5 year longitudinal study showed lung cancer rates among physicians were 22 to 24x higher for smokers than non-smokers. His theories included one that lung cancer caused smoking, and another that some people had a heriditary disposition for both smoking and lung cancer.
One of the AGW skeptics big misunderstandings is that AGW models are "statistical models".
They roll out statistics types used to soft sciences or (even worse) economics where there are no rules, and these types say modelling is impossible.
But Climate modelling is only partly statistical. Many aspects of the modelling have completely understood physical models which basically solve the problem. Others have physical models which provide goos approximate solutions. Al that stuff, and analysing whether it is OK, is complex and completely beyond the ken of statisticians.
To put it more simply:
A decent climate modeller must also be a decent statistician.
A decent statistician will not usually have the necessary physics skill set to be a climate modeller.
Assumptions: 1) E=1/2CV2
(Only dummies assume this)