Thursday
Jul072011
by 
Bishop Hill
Bishop Hill More bend it by Bayes
Jul 7, 2011 Climate: IPCC Climate: Statistics Nic Lewis's article on the IPCC's "bending by Bayes" of the Foster and Gregory estimate of climate sensitivity got quite a lot of interest a few days back. Nic has followed up with a searching letter to IPCC lead author Gabriele Hegerl.
I think this one will run for a while yet.
Reader Comments (72)
“””””As in "Sun enchanted heating"?!””””
Hoo boy. Reminds me of a bad old "Knock knock" joke--
"Who's there?" "Sam and Janet." "Sam and Janet who?"
...you can guess the punch line.
It would appear that Exeter has taken up sun worshipping:-
“‘scientists at the Met Office and elsewhere are beginning to understand the effect of the 11-year solar cycle on climate. When sunspots and other solar activity are at a minimum, the effect is similar to that of El Niño: more easterly winds and cold winter weather for Britain.”
“We now believe that [the solar cycle] accounts for 50 per cent of the variability from year to year,” says Scaife. With solar physicists predicting a long-term reduction in the intensity of the solar cycle – and possibly its complete disappearance for a few decades, as happened during the so-called Maunder Minimum from 1645 to 1715 – this could be an ominous signal for icy winters ahead, despite global warming.”
But it would also appear that they may have spent just a little too much time in the rays:-
“‘the reputation of the Met Office is higher than ever. “Our trust scores are about 82 per cent, which is phenomenal for any organisation,” says Varley. “I find it heart-warming that, when it comes to the crunch, people trust the Met Office.”
http://www.ft.com/cms/s/2/35145bee-9d38-11e0-997d-00144feabdc0.html#axzz1QSiDNUnF
Even though they possess few Cambridge graduates, the IPCC under Pachauri could field a respectable comedy skit entry for the Edinburgh festival - (indeed possibly this has been the intention all along). My favourites: Pachauri explaining to a journalist (played by Roger Black) that the Himalayas have just evaporated (performed in the Himalayas), and the hammering of a round data peg into a square hole by Michael E. Mann, Jr (the grimly determined facial expressions make this priceless).
I wonder whether the trust score correlates with the improved understanding (or should that be acceptance) of solar effects?
ZT
More likely tte "trust scores" have been subjected to a very local (Exeter area) UHI adjustment.
Or, taking your point further, just what are they trusted for? I stopped using the Met Offices weather forecasts about a decade ago and I am happy with my choice as they appear to be akin to and led by their climate machinations.
One of the best pieces of advice I every recieved was "to stick to doing what you know and do it well". Sadly the Met Office does no tappear to have been in receipt of that insight.
Maybe those trust scores are based on Bayesin stats?
To précis Scaife at the Met Office;
We are playing catch-up in the face of evidence against us. To do this we have had to include information in our machinations that we had previously ignored. We don't like to be derided and point out that 82% is a statistic that supports us.
“‘the reputation of the Met Office is higher than ever. “Our trust scores are about 82 per cent, which is phenomenal for any organisation,” says Varley. “I find it heart-warming that, when it comes to the crunch, people trust the Met Office.”
I can only assume that the 82 per cent are people who work in offices, have no idea of what the weather is doing, and don't much care anyway.
As someone who works outside, amongst my colleagues the Met Office weather predictions have literally become a long standing joke.
"What's the weather doing today?"
"The Met Office says the rain is coming in at 11.33"
[Guffaws of laughter.]
It doesn't take much to amuse us, admittedly.
My experience of working in London is that most people who work indoors are supremely unrealistic about weather. A gentle shower is described as 'Chuckiin' it down', a heavier shower and 'It's raining stair-rods', a couple of centimetres of snow is a blizzard and anything slightly toward the extreme becomes a potential disaster. The supposed 'heat wave' promoted by the Met Office a couple of weeks ago seemed another example of scaring ignorant people about the terrors of what is actually a very mild and gentle climate. Most of the Londoners I have worked with in London believe they suffer the worst climate extremes in existence, yet my relatives in Yorkshire, who experience far greater extremes of rain and snow, see their climate as 'normal'.
Nic Lewis has exposed that the state of climate science is much worse than anyone thought.
I do think this underpins the growing concern that the so called scientific consensus is no more than a system of belief. The science, the data, the methodologies are being made to fit the consensus.
The WG1 authors have behaved in a way that would make Alistair Campbell proud.
Green Sand
A more worrying thought is that the Met Office is sticking to what it knows and is doing it well!
I don't dispute the 82% trust figure. At least 82% of the people I know trust the Met Office to get it wrong at least 82% of the time.
Alexander,
The reality is that so called heatwave was merely one hot day (last Tuesday) and that was it. Come Wednesday and it was back to being lovely and cool (quite a relief from the previous day).
Regards
Mailman
Theo
Your post yesterday struck a chord with me. I am no statistician by any stretch of the imagination, but I have have had a fair bit to do with them over the years.. I researched the Bayesian stuff briefly and am quite surprised that climate scientists actually use what is essentially Game Theory to make meaning from their very strange take on what they see as the measurement of traces of artifacts left behind by transient phenomena connected with climate in some way. I too have come to the conclusion that such methods have more to do with opinion and faith than they do with actual measurement.
The more I discover about science the more I admire engineers.
For those readers who are not resident in the UK, it is of great importance that attacks on the credibility of the IPCC and their acolytes continue until such time as they are totally discredited.
This is the only way that you can avoid the problems developing in the UK as a result of a government taking on board the stuff published by those bureaucrats and "scientists". You may be spared the dreadful consequences of "fighting global warming" being embodied in legislation in your countries.
Unfortunately for readers who are voters and tax payers in the UK, it does not matter whether or not the IPCC is closed down, or that every paper supporting AGW (or even CAGW), are withdrawn. It is already too late for the UK.
The IPCC reports and those papers resulted in the UK government putting on to the statute book the Climate Change Act 2008 which, by the law of unintended consequences, commits in the very near future a large proportion of the UK population to fuel poverty and needless premature deaths by freezing.
This sorry state of affairs will continue unabated in the UK, irrespective of the discrediting of the the IPCC and associated "scientists", until this horrendous piece of legislation is repealed.
This is unlikely to happen any time soon as can be seen by reading the Hansard report of proceedings on Thursday, 7 July 2011, in the House of Commons, Westminster, London, UK:
(see http://tinyurl.com/4yyrbs)
"Oral Answers to Questions, Energy and Climate Change"
Here are three contributions on the subject of fuel costs and fuel poverty as recorded by Hansard:
Mr Mike Weir (Angus) (SNP): Last week I received my gas bill, which showed that my consumption had been 25% lower than it had been during the same period last year. At the same time, however, the direct debit payments requested by the company were rising substantially.
Barbara Keeley: The charity Age UK predicts that rising energy bills will take 250,000 more pensioners into fuel poverty, and those pensioners are under-heating their homes by rationing their consumption of fuel and thereby increasing their exposure to potential ill health, misery and depression.
Luciana Berger (Liverpool, Wavertree) (Lab/Co-op): Yesterday, a report by uSwitch showed that fuel poverty levels in the UK are spiralling, with 6.3 million households—almost a quarter of all UK homes—now classed as being fuel poor.
It is well worth reading the whole thing, even if you do not live in the UK, just to see the antics of the so called "representatives of the people" as they inhabit what appears to be a parallel universe stuffed with windmills, CCS projects and other fantasies.
For the readers who live in the rest of the world, what is being suffered by the UK now and in the near future, could be your fate unless you can succeed in bring down not only the IPCC but also any "representatives of the people" in your area who appear to be infected by this madness.
Good luck.
Brownedoff
Check tomorrow's news from Australia.
It will begin something like: "The price of carbon has been finally set at..."
Theo Goodwin, 5:11 PM yesterday, was a bit harsh on Bayesian analysis. Here is my shot at defending it from such a broad dismissal:
The key role of applied statistics is, in my view, to help with both the testing and the encouragement of ideas about what has happened in a system, and what has caused or contributed to it. The statistical toolbox has a lot of stuff in it, including a whole section on Bayesian methods. In my own career in industrial statistics, I never used that section at all, partly because I never felt any need for it, and partly because I had never studied it in sufficient depth. But I know it is one that is greatly prized by those experienced in its use. It, along with other statistical methods, shares the virtue of allowing the investigator to expose assumptions, models, and other reasoning devices, along with the data, to help establish the credentials of a piece of analytical work. As McIntyre has so able demonstrated, results based on statistical methods are open to investigation – their results should be reproducible, the chain of reasonings exposable to critical review. The work of Lewis which has caught our attention here is a another example of just this. One of the most inspiring of statisticians for industrial workers is G E P Box. In his book on Bayesian methods, co-authored with Tiao, they wrote this:
‘Inferences that are unacceptable must come from inappropriate assumption and not from inadequacies of the inferential system. Thus all parts of the model, including the prior distribution, are exposed to appropriate criticism.’
Box & Tiao, Bayesian Inference in Statistical Analysis, Wiley Classics (1992)
Jul 9, 2011 at 2:05 PM | Roger Carr
That is yesterday's news in the UK.
Or rather, 23 March 2011:
From Hansard:
The Chancellor of the Exchequer (Mr George Osborne):
Blah, blah, then ......
23 Mar 2011 : Column 960:
First, as I have long argued, investment in green energy will never be certain unless we bring some stability to the price of carbon. Today we become the first country in the world to introduce a carbon price floor for the power sector.
See Hansard:
http://tinyurl.com/3bsbxn5
John Shade,
I regret painting with too broad a brush, if that is what I did. I know that Bayesian statistics are widely used and can be very valuable. However, I was talking about the specific context of establishing a degree of belief for scientific hypotheses. That is not similar to diagnosing cancer or bad welds in pipes.
My criticism of Bayesian statistics - as a means of establishing a degree of belief in scientific hypotheses - can be made clear as a bell simply by looking at the case in point, by looking at what the IPCC did. Some scientists wrote an article in which they argued that climate sensitivity is rather low by IPCC standards. They used the "Prior" X. Some people at the IPCC believed that the "Prior" should by Y and they changed X to Y with the result that climate sensitivity shows up as much closer to what the IPCC believes. That lays the groundwork for description of the problem.
If there is in Bayesian statistics no rational method for choosing between "Priors" X and Y, then using Bayesian statistics - in this case - means employing irrational methods to establish a degree of belief in a scientific claim. I can see no rational method. It seems that the choice of Prior is nothing more than a matter of preference; that is, the authors wanted this one and the IPCC wanted that one and, because the IPCC is the publisher, the IPCC got their preference and the authors were left with no basis for objection. Can you imagine anything more irrational? If I have made an error, please correct me.
James Evans
You sound like some of my neighbours! Many of us in the Met Office live in the villages around Exeter, and have friends who work in farming who don't let us get away with a poor forecast. OK it doesn't always work out, but generally speaking I think we do pretty well. I do remember putting my reputation on the line when I got the whole village to move an outdoor barn dance indoors on the strength of the forecast - it was a beautiful morning but I insisted, and we spent all day clearing out my neighbour's working barn, and with no sign of rain coming I was getting pretty nervous - but then as expected the heavens opened in the early evening and my credibility was saved! (On that occasion anyway!)
BTW nice to see recognition (albeit sarcastically) by simpleseekeraftertruth that we *do* revise our approach based on evidence - just like scientists are supposed to ... :-)
Theo Goodwin - thank you for your response. I suspect I share your horror at this most recently exposed machination by the IPCC, but I would not throw away Bayesian methods with the dirty bathwater. I tend to get defensive about statistical methods, even ones I poorly grasp, because I think the fault, when they are attacked, often lies elsewhere. I came relatively late to the subject, and developed a huge admiration for it and for its best practitioners. I hope I have not come across as too dogmatic as a result. But coming back to choice of prior, I think the proper users of this methodology do not regard it as merely a matter of preference - I think they devote a fair amount of agonising to it, and it some cases at least the prior can be based strongly on observational data. It will be interesting to see whether or not there is a substantial reply to Lewis, and if there is, what view they will take on this.
John Shade,
Thank You for your response. I do not care to criticize Bayesian as such and I am glad that you pointed out that I had gone too far. I am quite happy to limit my criticism to the particular circumstances under discussion. If the the IPCC can get away with the claim that there is a rational reason for choosing the Prior that they substituted then I would like to see it. However, I would like to make the logical point (not methodological) that the means for choosing among competing Priors is not found in Bayesian statistics, but (pardon the pun) prior to use of Bayesian statistics.
It is understandable that cash makes us free. But how to act if one does not have cash? The only one way is to receive the personal loans and just secured loan.