Steve McIntyre writes about what he considers another “completely worthless” exercise in statistical data mining, writing:
===============================================================
In today’s post, I will look at a new Naturemag climate reconstruction claiming unprecedentedness (h/t Bishop Hill): “Evolution of the Southern Annular Mode during the past millennium” (Abram et al Nature 2014, pdf). Unfortunately, it is marred by precisely the same sort of data mining and spurious multivariate methodology that has been repeatedly identified in Team paleoclimate studies.
The flawed reconstruction has been breathlessly characterized at the Conversation by Guy Williams, an Australian climate academic, as a demonstration that, rather than indicating lower climate sensitivity, the recent increase in Antarctic sea ice is further evidence that things are worse than we thought. Worse it seems than previously imagined even by Australian climate academics.
the apparent paradox of Antarctic sea ice is telling us that it [climate change] is real and that we are contributing to it. The Antarctic canary is alive, but its feathers are increasingly wind-ruffled.
A Quick Review of Multivariate Errors
Let me start by assuming that CA readers understand the basics of multivariate data mining. In an extreme case, if you do a multiple regression of a sine wave against a large enough network of white noise, you can achieve arbitrarily high correlations. (See an early CA post on this here discussing example from Phillips 1998.)
Read the entire post here: http://wp.me/p6iHb-50Y
===============================================================
For information on the Southern Annular Mode, see this:
http://stratus.astr.ucl.ac.be/textbook/chapter5_node6.html
Interesting.
SAM reconstruction (Figure 2) follows notch-filtered Sunspot Integral over the entire sunspot record.
(But let’s remember: US politics forbids the expression of such ideas.)
‘Unfortunately, it is marred by precisely the same sort of data mining and spurious multivariate methodology that has been repeatedly identified in Team paleoclimate studies.’
In other news bears still find woods a good place to get rid of personal waste .
Although to be fair if all you have is BS then all you can sell is BS so can they actually do any better ?
“The Antarctic canary is alive, but its feathers are increasingly wind-ruffled.”
Oh, great! Just when we thought the Arctic “canary” had was feeling a bit better and starting peck at his bird-seed again, they’ve gone and invented another one to worry about.
They’ve been whining about the Arctic “canary” for a decade. When Arctic ice volume recovers 50% in ONE year, rather than breathing a huge sigh of relief, they suddenly go all quite.and then pop up with : “look over here, this is the canary in the coal mine, it’s worse than we thought , we must act now , etc, etc, etc. ……
Boy cries wolf, once more. He just has not realised no one believes any more.
Sampling on the Dependent Variable – UCLA
http://gabrielr.bol.ucla.edu/soc210a_f09/w9.pdf
210A Week 9 Notes
Sampling on the Dependent Variable
Technically, sampling on the dependent variable is when you select cases on the basis of meeting a criteria and then use those cases as evidence for the criteria. Since we’re usually more interested in associations than distributions we can broaden this problem to something like “sampling on theory affirmation.”
This practice is at the center of Karl Popper’s positivist approach of “falsifiable hypotheses.” Popper complained that Freudians and Marxists just cataloged evidence that supported their theory whereas he argued that real science consists of searching for evidence against your theory and failing. He gave the example of the hypothesis “all swans are white” and said the way to test this hypothesis is not to accumulate a vast catalog of white swans, but to search for a black swan. On so doing you would find that there are in fact black swans (in Australia) and so the hypothesis is false
http://beerbrarian.blogspot.ca/2013/11/the-subtle-joys-of-selecting-on.html
The Subtle Joys of Selecting on the Dependent Variable
Academic research in the social sciences has a variety of aims, but much of it seeks to explain or elucidate phenomena or condition(s) and the relationships therein. In research parlance, this phenomena or condition is the dependent variable. One should not select cases that satisfy the criteria of the dependent variable; doing so is called selection bias and can lead to incorrect conclusions.
Case Selection and Causal Inference in Qualitative Research
Thomas Plümpera, Vera E. Troegerb, and Eric Neumayerc
We show that causal inference from qualitative research becomes more reliable when researchers select cases from a larger sample, maximize the variation in the variable of interest, simultaneously minimize variation of the confounding factors, and ignore all information on the dependent variable. We also demonstrate that causal inferences from qualitative research become much less reliable when the variable of interest is strongly correlated with confounding factors, when the effect of the variable of interest becomes small relative to the effect of the confounding factors, and when researchers analyze dichotomous dependent variables.
“If you torture the data enough, nature will always confess.”
http://en.wikiquote.org/wiki/Ronald_Coase
My goodness. I’ve never known a science discipline so hell bent on finding acorns to hit themselves on the head with.
Even by alarmist journal standards, there are a remarkable number of ‘comments removed by moderator’ in that ‘Conversation’ piece…
“Let me start by assuming that CA readers understand the basics of multivariate data mining.”
Which is why I stopped reading Climate Audit. A roomful of geniuses who gather to be convinced of their own sparkly brilliance while stroking each other’s shiny throbbing swollen egos is what got us into this mess.
“kadaka (KD Knoebel) says:
June 16, 2014 at 8:49 am
“Let me start by assuming that CA readers understand the basics of multivariate data mining.”
Which is why I stopped reading Climate Audit.”
Since you don’t read it, you don’t comment. And for that we are eternally grateful.
Thank you.
Spend time with a text book. read more;comment less
Steve McIntyre’s entire post is a fascinating read. We’ll hear quite a bit more from the CAGW team about Abram et al, 2014. Cold is hot after all, don’t you see?
erratum: Steve’s entire post at Climate Audit…
From Steven Mosher on June 16, 2014 at 9:22 am:
And that would make me more of an elitist bass turd smugly convinced of my superiority thus more like you with your “Mine is bigger! You must respect it! Start stroking, you tiny worm!” attitude?
I’ll pass. You could sit alone in a room with a mirror and debate with your image whose intellect was the most impressive. I don’t want to be infected with such a vile debilitating contagion.
kadaka (KD Knoebel) says:
June 16, 2014 at 8:49 am
A roomful of geniuses who gather to be convinced of their own sparkly brilliance while stroking each other’s shiny throbbing swollen egos is what got us into this mess.
Does “this mess” include the computer you typed this on or all the other achievements of science and engineering? It is important that people who are on at least reasonably friendly terms with multivariate statistics discuss and dissect these artifacts of statistical abuse that the paleo community churn out of their sausage factory. In the meantime, chief bully Michael Mann arrogantly twitters about how PAGES2K confirms the hockey stick, and his admirers retweets and bullies anyone who dares question it.
I know a little multivariate statistics, not enough to follow all the fine points over at CA, but enough that I was completely flabbergasted when I discovered what kind of junk science the paleo guys get through peer review. I’m really grateful that people like Steve McIntyre, Jeff ID and many others have had the guts and time to confront the lunacy with reason and sanity.
From Espen on June 16, 2014 at 1:35 pm:
The (C)AGW-hyping grant-sucking wealth-draining freedom-stealing mess, of course, the default mess examined on this site.
I shall consider that a one-off quip and so-so attempt at humor, rather than your attempting to smear me as an anti-science anti-technology anti-intellectual Luddite with whatever substance in shades of slimy brown you had readily available at the time, until you inform me otherwise.
How very educational of you. Amongst your faults that you missed stating, you are insulting, demeaning, irrational and quite incorrect. Why you seek to badmouth Climate Audit and the people who visit CA, especially here on WUWT is an absurd endeavor.
Steve McIntyre only displays irritation and lack of patience with those commenters who combine bad manners with baseless assumptions or attacks. He snips comments that violate civil discourse and does an excellent job of bringing research to any conclusions that can be made when conclusions are possible given the data.
Your last gasp insults in several of your insults are curious in mendacious ways; claiming self-absorption non-sparkly dullness while thumping your swollen ego for all to read, is exactly what you are attempting to do here in a solo fashion.
I agree with Steve Mosher.
Which is why I stopped reading Climate Audit
=============
if you can’t argue the facts argue the man.
One doesn’t need to be a mathematician to understand the problem with the statistics in a great many climate science papers. Almost everyone knows about the “random sample”. Most people know that if your sample isn’t random, then your statistics are going to be faulty.
Yet climate science paper after paper, the notion of the “random sample” is violated. Time and time again they take the data and apply a filter, such that the sample is no longer random. Typically this filter goes by a name such as “calibration”.
Climate science argues that this “Filter” or “Calibration” makes the data more accurate. And right there is the problem. If you need to filter the data to make it “more accurate”, this is telling you that the data isn’t accurate enough to start with to make any conclusions.
And if you do filter it, then it is no longer random, and cannot be analyzed statistically. To do so violates the underlying mathematical foundations of statistics. Technically the problem is referred to “selection on the dependent variable”, and is written up in many, many places.
http://www.nyu.edu/classes/nbeck/q2/geddes.pdf
Another form of “selection on the dependent variable” is “Weighting”. each of the samples is assigned a “weight”. the weights are selected by some criteria that evaluates how “good” each value is. The actual values are then multiplied by their “weights”, to arrive at new values.
In this fashion “good” values get high weights, and “bad” values get small weights. really bad values for example, get a weight of zero, and disappear. one tree gets a huge weight, while the rest of the trees get weights close to zero. this one tree then becomes your proxy for global temperatures.
However, most people will by now see the problem. What started out as a random sample is no longer random. the weights were not applied randomly, they were applied based on some notion of “goodness”. And in doing so, from that point forward, you can no longer use statistics to analyze the data, because statistics is based on the notion of the random sample. And if you do use statistics, the result will be meaningless. The correlation will be spurious (false).
However, if you are in the business of promoting a false story, then a spurious correlation is what you are looking for.
@ ATheoK on June 16, 2014 at 5:07 pm:
Hey, don’t worry, I know you boys run CA as a private “mutual admiration society” and willfully incidentally accidentally drive away all who don’t “measure up” to your yardstick. Better you guys get your freak on over there rather than in public. I’m actually grateful you make no pretense of being inclusive at all, saves me the trouble of caring what’s over there except for CA Assistant.
It’s much better here in the meeting hall where they try to inform the masses to free them from the propaganda, rather than the well-apportioned drawing rooms where they debate the opposition’s barbaric preparation of noon tea and wearing of unfashionable couture.
@ ferdberple on June 16, 2014 at 5:48 pm and 6:06 pm:
Now that is understandable. Thank you. I prefer it when people are willing to explain, rather than think if you do not understand than you are not worth talking with.
ATheoK says:
June 16, 2014 at 5:07 pm
““““““““““““““`
I did note that Steve Mosher didn’t rate the piece. Damned with faint praise?
kadaka (KD Knoebel) on June 16, 2014 at 2:47 pm
The (C)AGW-hyping grant-sucking wealth-draining freedom-stealing mess, of course, the default mess examined on this site.
Of course – my question was rhetorical to make you think twice about Climate Audit. For science and technology to make progress, some discussions have to be extremely technical and thus not open to everyone. The really tricky part is of course that this may allow for bad scientific subcultures to grow – like (large parts of) the paleo climate community. So your point is not without merit, but IMHO you’re shooting in the wrong direction.
From Espen on June 16, 2014 at 10:49 pm:
“He who fights with monsters should look to it that he himself does not become a monster.” – Nietzsche
As I understand the lore, Steve McIntyre started Climate Audit because basically Mann and associates thought he was an idiot who couldn’t understand their work and they wouldn’t explain what they were doing. Climate skeptics of all types are regularly accused of being idiots who can’t possibly understand what the revered learned climate scientists are doing, even when the skeptics are climate scientists or similar, or have related underlying knowledge like math and statistics.
From CA aficionados, come derisive dismissals, go read a textbook. As if anyone could pick up graduate level subjects from dry printed tomes. And if you can’t, then you’re not worth their time, you’re an idiot.
I do not volunteer my time towards the skeptic cause to trade one set of know-it-all stuck-up elitist intellectuals for another.
As above, the multiproxifiers and “dendrophrenologists” are habituated to and repeatedly return to dubious statistical methods as they usually produce the “right” answers.
And you’ve supposedly experienced or witnessed this? Can you point/link to a specific example?
In the years I’ve been reading Climate Audit I’ve yet to witness anything like what you are claiming.
There have been occasional troll types who show up, ask far lower than High School education questions and then get snotty that they’re not being catered to. I doubt you would be one of those.
There are also occasions where specific topics of physics, regression analysis or similar have been discussed ad nausea in prior threads and comments; many of the regular visitors do not desire wasting time nor space on the discussions again. A quick search can not only take to the relevant Climate Audit threads, but also find other internet postings on the subject.
From ATheoK on June 17, 2014 at 7:39 pm:
Shut mouth, look up.
Sun & SAM
Sunspot Integral & Southern Annular Mode (SAM)
Kadaka:
Another way to look at it is as the fault of selecting data that supports your theory and rejecting data that does not. The rejected data can be used to refute the conclusions of the study, and so the whole study is fruitless.
The language at Climate Audit may be more arcane than the way I have put it, I will agree, but it is the terms of statistical analysis. I think, that if you reflect, you will admit that much good work has been done at Climate Audit. Climate Audit is solely responsible for the exposure of the pseudo-science and the scam science of some of the paleo-climatologists such as Mann, Briffa, Gergis, and others. This is a service of inestimable value, a service to science and to the rest of the world.