WUWT readers may remember this popular article from August 30th, 2010 New paper makes a hockey sticky wicket of Mann et al 98/99/08 and then The Team’s response RC’s response to McShane and Wyner: a case of orange cones which gave rise to Josh’s cartoon and this cartoon coffee mug:
Patrick Hadley writes in comments today:
OT – The McShane & Wyner discussion is now available at the Annals of Applied Statistics.
There is a lot of fascinating material to read there – the original paper, criticisms from the Hockey Team, support from others and, what seems to me at least, a brilliant rejoinder from McShane and Wyner.
I’ll say. Wow, this paper stirred up a statistical hornet’s nest, just have a look at the table of contents related to this paper. It reads like a who’s who of paleohockey. Each article is fully open, no paywalls; which I see as a testament to the journal integrity. If nothing else, read the editorial by Michael Stein which speaks to the entire table of contents.
There is so much here, I can’t even begin to dig into it all (It’s a workday for me) but if readers wish to place excerpts below of interest, I’ll do a follow up post with them. – Anthony
With all respect to Mr. Steiner, but while his comment about the ‘sharp null’ is reasonable – he did not also talk about the ‘unprovable future assertion’ encompassed by IPCC/GCC ‘projections’ which continue to be wrong in both direction and in magnitude.
It might be useful for Mr. Steiner to elucidate how he objectively analyzes said models and the ‘projective’ performance – and how then 50 and 100 year future behavior is to be based on said models.
Again, it isn’t that alternative energy is not a good objective – but there are vast differences between an intelligent plan for implementing alternative energy which is competitive with existing energy without subsidies as opposed to drowning the problem with cash (and the charlatans lined up at the trough).
Stein finds it implausible “that a large increase in the concentration of CO2 in the atmosphere has exactly zero effect on the global mean temperature.”
And he surprises me by falling into such an obvious trap.
The concentration of CO2 has increased (in round figures) from 300ppm to 400ppm – in other word from 0.03% to 0.04% While a mathematician may speak of a large increase I doubt that a chemist would see a change in dilution of this magnitude as significant or likely to have any relevance in the real world.
Maybe some chemists on here can correct me.
I find the Stein editorial rather unhelpful. He makes the assumption that in this case null hypothesis should be ignored because the effect can’t be null. (why not?) Then he pushes us to act quickly because the great danger of inaction. The second is pure silly resort to the precautionary principle and because we really can’t predict it may be exactly the wrong thing to do.
M&W are scathing of Schmidt, Mann and Rutherford’s statistical abilities. Plus the mistakes SMR make. They are very complementary of M&M: section3 “MM seem to most clearly understand the purpose of this section, and we again recognize their contribution for first pointing out these facts”.
They even get to plug Feynman’s comments on cargo cult science, so hats off to M&W.
RC team response should be interesting.
After reading all the short papers here, and reading the conclusions of longer papers where available, plus scimm-reading them, I will reapeat what I said here what feels about a year ago, when this site linked the paper in discussion as a break through, and the critique in relation to it.
It is evident, though in a very polite, respectful academic manner, that now and then, people are not happy with the method(s) chosen (e.g. questioning use of LASSO, remarking that their Bayesian is actually none, etc).
This goes on to dismiss their argument of random data performing just the same as actual proxies, and in fact granting the Mann reconstruction for the past decade a +90% probability.
The MW may be commeneded for as a great effort, albeit just ‘not quite there yet’, or not cutting it for several statistical short comings. This is the impression I had regarding my last read of initial criticisim as mentioned above, and for which I belive I received a critical comment from Anthony back then (?).
Now, I do not feel that the MW paper has been strengthend in light of all the linked dicussions contributed. I merely think it is commended as an effort which conclusions are relative in light of all the short comings brought forward against it.
I don’t think this is a death nail to the hockey stick at large, and I don’t think the joint input knights the MW paper.
Hundreds of Mann-hours and man-hours have gone into this work. One problem sticks out. The modern temperature record used in all rhe statistical models is from CRU, and is assumed to be accurate. Anthony’s Surface Station Project and satellite data, suggests strongly that the temperature record is wrong. If the CRU data is wrong, ALL that work is invalid.
Having read the rejoinder, I ask myself, if it is not time now for some at CRU, GISS and elsewhere to retire.
If the HockeyStick discussion finally starts using sound mathematics some of those who prefer to collect, massage and gatekeep data, or wander around in exotic places such as Antarctica and Tahiti, or those who like to chat and groupthink via email, all of those may lose interest in such a dry subject.
M&W’s rejoinder to SMR would fall into the category that Phil Jones calls “going to town on”.
Ouch! That will leave a mark.
Stein’s last paragraph is pure post-normal but without any reason or preamble given for him to make the statement. Why not make a substantive comment on the papers or was it because they did not support his obvious bias? You don’t need statistical analysis to come to that particular conclusion.
Very interesting read until the last paragraph which just deflated me. My head stopped nodding and started to shake vigorously from side to side. What is it about the precautionay principle that convinces so many otherwise sensible people. Yes, we insure houses against fire and our cars, but the cost is proportunate to the value of that which is insured. Not so with the preparations for uncertain climate changes. The costs envisaged are truly enormous and far outweigh benefits. I think that the precautionary principle is responsible for the wholesale persuasion of most of the politicians who support AGW because it is so easy to misrepresent the principle with just a few flawed analogies. This then enables the persuaders to explain to their audience that the numbers are not so important and they need not concern themselves with the science because it is all but settled. As far as the political parties in the uk are concerned, the science is settled and they are all proceeding on that basis.
Truly terrifying (in my opinion).
From McIntyre and McKitrick:
“It also needs to be clearly recognized, that, even though MW results are rather discouraging for the reconstructions using the Mann et al 2008 network, they are, in a sense, a *best case* as they assume that the quality of the data set is satisfactory (thereby not taking a position on prominent controversies over the proxies within this data set).”
They go on to describe many such problematic series. They end with this description:
“In addition, nearly 10% of the Mann et al 2008 network (105 series) are series derived from the Briffa et al, 2001 network, notorious for its late 20th century decline. However, actual data after 1960 has been deleted and replaced by data infilled by a RegEM process (Rutherford et al 2005.) Use of the actual post-1960 data will further erode performance of the proxy reconstruction.”
I did not find the editorial to be that good.
Funny how Gavin A Schmidt, Michael E Mann, and Scott D Rutherford only talk about reducing the MWP if you remove certain series that are deemed unacceptable.
“That additional warmth, as shown above, largely disappears with the use of the more appropriate dataset.”
MW specifically said they did not do quality assessment of the proxies. This brings us back to MM’s point that this is a best case scenario. It actually gets worse if you look at the quality of the series. But give it to Mann and Schmidt wanting to hide not only the decline, but also the MWP.
The statistics of the temperature trends are nice, but here is why engineers really laugh at global warming and the energy balance of Trenberth…
John Kehr
I would suggest a mug with Anthony holding the Ap index….in 50 years, while suffering the deepest of the solar minimums, it would be remembered…
Ed Caryl – no, that does not stick out at all, it hasn’t even been mentioned, or as an aside at best – it is a discussion of the statistical merits of their methods – it is a statistics jouranal, you know. Not even the editor’s preface linked at the top makes a case for either side AT ALL.
Perhaps we need to concentrate more upon the descibing the possible negative impact which could occur if we begin to attempt procedures which may cool the climate when it is already cooling given the dirth of solar activity we are experiencing. I say this as the AGW crowd has been pushing the “what could it hurt” philosophy and the “better safe than sorry” line in their propaganda for many years now. This while the fact is that cold is much more devastating to society than is warm. And this is easily proven in historical records. My favorite idiot scheme is to spray aerosol SO2 into the atmoshpere to cool the planet. Correct me if I am wrong but I believe this one came from someone in the present administration or legislature in Washington, DC.
“tree rings ARE NOT PROXIES FOR TEMPERATURE. Hide the decline.”
I will bet if an elk died next to a tree, it would show increased growth for many years.
Tree ring increases might also be CO2 proxies in certain species, particularly conifers.
This one’s a cracker!
“Before embarking on our discussion of their work, we must mention that,
of the five discussants who performed analyses (DL, Kaplan, SMR, Smerdon,
and Tingley), SMR was the only one who provided an incomplete and
generally unusable repository of data and code. The repository created by
SMR specifically for this discussion was, like that of the other four discussants,
graciously provided and quite usable. However, we lacked clear and
easily implementable code (i) to fit RegEM EIV ourselves and (ii) to draw
new temperatures and pseudoproxies from their simulation model. Code
for these purposes is archived by Mann at:
http://www.meteo.psu.edu/mann/PseudoproxyJGR06/
http://www.meteo.psu.edu/mann/supplements/MultiproxyMeans07/
Among other things, the RegEM EIV fitting procedure cannot be executed
by a straightfoward function call as is typical for statistical code libraries.
Rather, the archives consist of a large number of files layered on top of one
another and, despite a major effort on our part, we were unable to replicate
published results within the publication time constraints of this rejoinder.”
SMR being ….
Schmidt, Mann, and Rutherford!
REJOINDER: A STATISTICAL ANALYSIS OF MULTIPLE
TEMPERATURE PROXIES: ARE RECONSTRUCTIONS OF
SURFACE TEMPERATURES OVER THE LAST 1000 YEARS
RELIABLE?
How to write editorials with huge disclaimer at the end. Did Mr Stein take lessons from Nicolaus Copernicus?
I grew up in a logging community. I have studied many stumps. You can read the history of a tree. You can tell when a neighbor tree was felled or died, when a fire went through, when a nearby stream changed course, and when other trees grew up and closed it in. If a tree has always been isolated, you can clearly see precipitation patterns. Temperature? Not so much.
Well, nobody (should have) ever said being a “scientist” is a walk in the park. A scientist must be well versed in a lot of disciplines (or surrounded and helped by others that are) and the discipline of “logic” should be the most solid of bases upon which every scientist AND statistician rest. However, even in the case of Michael Stein, that is not always the case. Specifically wherein his editorial he says, An example of an implausible sharp null hypothesis would be that a large increase in the concentration of CO2 in the atmosphere has exactly zero effect on the global mean temperature. Of course! But “zero effect” is a fool’s answer to an non-relevant question, in the wrong argument. And while he continues with clear logic with – When a null hypothesis of no effect is untenable, emphasis should be on estimation and/or prediction along with uncertainty quantification. He then “jumps the shark” by writing – Thus, the testing and attribution questions for climate change seem to me to be irrelevant and the focus needs to be on prediction. AND – Thus, now is the time for individuals and governments to act to limit the consequences of greenhouse gas emissions on the Earth’s climate over the next century and well beyond. WHAT!?! Yeah, IF the effect is NEGATIVE!
Talk about a leap in logic. The “effect” (whatever the unknown amount is) is not zero, so therefore take immediate action to limit the consequences! What if the non-zero effect is only perceived (clearly Stein’s assumption) to be negative but in reality the net consequences “over the next century” are positive? Isn’t THAT the REAL question and what most of us here keep asking? What is this unfathomed penchant in climate scientist AND now (I guess) statisticians for assuming only the worse can happen from a couple of degrees increase in annual average global atmospheric temperature? Talk about an “implausible sharp null hypothesis”….!
Rejoinder:
“As for the indirect ”multivariate calibration” approach suggested by some
of the discussants, we point out that it was designed for highly-controlled
almost laboratory-like settings (e.g., chemistry) with very tight causal relationships.
The relationships between temperature and proxies is considerably
dissimilar. Furthermore, we believe the two approaches, direct and
indirect, ought not differ much in terms of ^y, suggesting that ”both types
of procedures should be able to yield similar results, else we have reason
for skepticism” (Sundberg, 1999)”
I thinks skepticism is reasonable in some cases:
http://www.climateaudit.info/data/uc/briffa98comp.png
If only CO2 affected tree growth it would be one thing, but knowing that nutrients, temperature, and water are more important than CO2 (over the minimum threshold), the only possible reason to insist on using the tree rings is the uncertainty in the temperature, which can be used to hide the decline or force a rise.
Stein: “I particularly object to the testing of sharp null hypotheses when there is no plausible basis for believing the null is true. An example of an implausible sharp null hypothesis would be that a large increase in the concentration of CO2 in the atmosphere has exactly zero effect on the global mean temperature.”
This may seem like common sense to Stein, but he is wrong to adopt a partisan position on a known area of controversy in climatology. Saturation is a plausible reason to test the sharp NH, and a saturated GHE is argued by Miskolczi.
Stein “Any potential benefits of waiting depend in part on estimates of how much our uncertainty is likely to decrease over the next several years.”
The key point is our knowledge about the likelihood of decrease over the several years to come. What if there is no sound reason to expect (any material) decrease in uncertainty anytime soon. In her excellent Respose To The Subcommitte On Energy And Environment Of The United States House Of Representatives, Judith Curry writes:
“Classical decision making theory involves reducing uncertainties before acting. There has been a growing sense of the infeasibility of reducing uncertainties in global climate models owing to the continued emergence of unforeseen complexities and sources of uncertainties.”
IMO Stein condtradicts his own claim that “Statisticians are, by their professional nature, skeptics”. Particularly when he asserts: “now is the time for individuals and governments to act to limit the consequences of greenhouse gas emissions on the Earth’s climate over the next century and well beyond.”
So much potential in the Editorial, but he blew it.
“tree rings ARE NOT PROXIES FOR TEMPERATURE. Hide the decline.”
I think that is your quote of the year.
Jaye Bass says: December 13, 2010 at 9:52 am
Mmmmmmmmm… noted.
——————————————————————-
Also, I’m sure the Ice-CO2 Hockey Stick that Al Gore used to outrageous visual effect, is sitting in people’s subconscious, insisting that – We – Have – Caused – CO2 – Levels – To – Rise (therefore the shamefully abused Precautionary Principle is paraded).
No. I’m highly doubtful that the CO2 rise is our doing (think, slow ocean turnover… slow warming… huge CO2 slow release… biosphere turns over far more CO2 than we emit…) But to prove my doubt, the ice CO2 science needs
examiningauditing.I wish someone would go to town on this, the same way that Steve McIntyre has gone to town on the temperature Hockey Stick. Lonnie Thompson’s non-archived data makes me even more certain that the ice core measurements are just as suspect as the bristlecone pine ring measurements. And despite Thompson’s non-data, the evidence Jaworowski puts forward in the above link is pretty damning. IMHO.
If the ice-CO2 science, and the infamous splice, were put under the microscope as Mann’s Hockey Stick has been, I’m sure those like Stein would no longer get away with reciting a knee-jerk salutation to the Goddess of Insane Precautions.