26 October 2009

A Point Resolved in the Hockey Stick Wars

Sometimes seemingly small disputes can tell you a lot. One of a what seems to be a bazillion points of dispute between Steve McIntyre and Michael Mann involves a claim by McIntyre that Mann mistakenly used a number of paleo-climate proxies upside-down. Mann replied by outright rejecting the claim:
The claim that ‘‘upside down’’ data were used is bizarre.
One of the challenges for observers of disputes about climate science is that they are often complicated, technical and nearly impossible to resolve without becoming an expert yourself. So generally people resolve them based on factors other than logic and data. In this case, it looks like this dispute will in fact be resolved unequivocally through the peer-reviewed literature, which for all of its faults, is the media of record for scientific claims and counterclaims.

Evidence pointing to a resolution of the dispute can be found in a newly released correction to a paper by the lead author (Kaufman) that was also claimed to have used proxy data upside-down. Kaufman has written a corrigendum which admits as much (here in PDF). McIntyre discusses the corrigendum today at his site. Here is an excerpt:
We pointed Mann's upside down use of the data (with a worse impact than on Kaufman) in the correct channels. Mann denied it. Once the matter is pointed out, it's not rocket science to determine who was right, but PNAS took no steps to resolve the contradiction. realclimate readers took Mann's denial as being proof that he didn't use it upside down . . . While Kaufman has admitted using the data upside down, Mann hasn't.
On this issue at least it seems that we can now resolve who's claims were correct and who's weren't.

79 comments:

Seneca the Younger said...

It will be fascinating to see how this propagates through the blogs.

Maurice Garoutte said...

Yes the public tends to trust their judgment of the people involved when there is a conflict outside of their own experience. My own first dive into the arcane global warming technology involved the dispute between Steve McIntyre and RealClimate. On RealClimate I found several paragraphs of straw man arguments followed by a final clincher argument that evaluated to an ad hominem. The arguments from ClimateAudit were clear, unimpassioned technology, while the arguments from RealClimate were drawn from the list of logical fallacies.

I fairly concluded that Steve was right since RealClimate from within all the invective did not deny the relevant facts.

Oh, technical reasons? The Team failed to normalize the proxy data over the full range under study and proxies telling a favorable story were weighted hundreds of times higher than other proxies.

For what it’s worth Steve McIntyre has never engaged in a public wrestling match with Joe Romm.

itisi69 said...

In climate science "Robust" and "Peer Reviewed" are probably the most mis-used words...

Fred said...

Dr. Mann gets two minutes in the box for High Sticking + 2 for his Unsportsmanlike Conduct.

His ability or inability to fess up will speak volumes to his integrity.

dljvjbsl said...

The real issue that this bears upon is the credibility of RealClimate. It would better for the climate debate if thse people realized the damage that they are doing to this important issue.

SBVOR said...

The Hockey Stick is pure bunk.

In relation to the hockey stick, let’s note that Arctic ice melt -- especially on the Greenland ice sheet -- is the current hysteria du jour on the AGW alarmist menu. Looking at data from GISP2 -- the Gold Standard in Greenland ice sheet ice core studies -- leads me to wonder…

From the GISP2 web site itself, does anybody find this ice melt/insolation chart “inconvenient” (to the AGW alarmists)?

Does anybody see a Hockey Stick in this chart of GISP2 data?

How about this Greenland ice sheet chart?

How about in the Antarctic data? Anybody see a hockey stick here?

Click here and here for more details, citations, etc. on those last three charts.

Click here for the overview.

the_fatman said...

Leave it to a Canadian to break the hockey stick...i should have known it would go down this way. 10 minute major & a game misconduct for Mann et al follwed up by a suspension with fines.

Tom C said...

Roger -

With due respect, this did not get resolved "through the peer-reviewed literature". Kaufman learned of the problem through Climate Audit, made the correction, but failed to cite the source. Seems to me the "peer-reviewed literature" remains indicted.

Roger Pielke, Jr. said...

-8-Tom C

Kaufman's corrigendum will indeed appear in the peer-reviewed literature. This is what I mean by "resolved through the peer-reviewed literature". Maybe I should have said "in the peer reviewed literature."

But please don't take from my post that I think all is well here.

Brian said...

Roger:

Perhaps, McIntyre's views about "upside-down" treatment by Mann and related peer review are "resolved by proxy".
Perhaps, Andy Revkin will write a follow-up to his recent article in which he reports Mann prodding McIntyre toward "peer reviwed" work (commentaries have alerted Andy about the "upside-down" treatment), and some reporting balance will emerge. Perhaps...

andrewt said...

Roger I think you failed to read the next sentence in Mann et al.'s reply: "Multivariate regression methods are insensitive to the sign of predictors".

Mann et al seem to be saying their methods are invariant to the data's orientation - perhaps to linear translation? - anyway it means the data can't be upside-down.

Now if the mathematical interpretation placed on the data by their methods conflict with physical information from other sources, it does raise questions. Presumably the next sentences in Mann et a's reply l refer to this, I can't see anything in Kaufman et al that illuminates this and it would seem to require detailed knowledge of the field to judge the importance of these issues.

And as you said, we (you, me, Steve McIntyre,...) are not professionally qualified to engage in the substance of such a debate.

Mark B said...

Roger

I have to disagree with your reading of the situation. The situation wasn't "resolved" by Kaufman - it was "resolved" by McIntire. He did the work to show that the proxies were used upside down. Anyone willing to put in the effort could see that he was right. Kaufman just acknowledged that McIntire was correct - he didn't do anything himself. The work was done by a blogger, and the evidence was posted on a blog.

You seem to be making a circular argument. Now that the error has been acknowledged in a journal - although not be "peer review" - it is real. Until someone who had published on the subject recognized the error, it didn't officially exist.

It seems to me that the exact opposite is true. A problem in the literature was pointed out outside the literature. Within the literature, all that came of the published challange McIntire made was a flat denial - with no backing data, no re-analysis, no argument. All the literature gave us was hand-waving and insults.

If someone outside the "team" had published on the subject and pointed out the error, then you could claim that peer review publication had resolved the matter - assuming that the team backed down publicly.

deancrowe said...

I read Mann's reply and don't see where he outright rejected the claim that he used proxies upside down. What he wrote was:

"The claim that "upside down" data was used is bizarre. Multivariate regression methods are insensitive to the sign of the predictors."

It looks pretty clear to me that he wrote that the claim was bizarre because the sign doesn't matter and never outright rejected the claim that he used proxies upside down.

Roger Pielke, Jr. said...

-11-andrewt and -13-deancrowe

Are you both saying that Kaufman's corrigendum is in error?

If Mann, who I assume is fairly proficient in the English language, wanted to say that he had in fact used the data upside down but that it didn't matter, he could have said something like "While I did use the data upside down, it does not affect the analysis because ..." But he did not say anything like this, instead he said that the accusation was "bizarre." Sorry, but you don't call a correct accusation "bizarre" unless you are rejecting it.

This line of argument is a lawyer-like parsing. However, why not just submit a direct question to Michael Mann at Real Climate asking him if he used the data upside down, and see what response you get? ;-)

deancrowe said...

Roger -

Its very simple - Mann did not write that he used the proxies right side up. That would have been an outright rejection of the claim that he used them upside down.

If Mann's English is fairly proficient and he wanted to outright reject the claim that he didn't use them upside down, he could have wrote something like "I did not use the data upside down", could he not? But he didn't, what he wrote was the methods are insensitive to the sign, and that is the reason he gave for calling the claim bizarre.

Call it "lawyer-like parsing" if you like, I really don't care. The thing is, it really doesn't matter. The important point is whether Mann was correct or not in saying that the sign doesn't matter. To that point the Kaufman corrigendum would seem to indicate that Mann was not correct in saying so.

So to answer your question if I'm saying Kaufman's corrigendum is in error, the answer is no.

Roger Pielke, Jr. said...

-15-deancrowe

This is where several recent threads here converge.

Steve McIntyre attempted to address this at Real Climate and he was censored. Had RC simply clarified the record directly in response to McIntyre's perfectly reasonable comment, there would be no ambiguity, would there? Why censor a comment that, if your interpretation here is correct, they could have easily dealt with?

Here is McIntyre's censored comment:

http://community.nytimes.com/comments/dotearth.blogs.nytimes.com/2009/10/05/climate-auditor-challenged-to-do-climate-science/?permid=379#comment379

A good rule is that if you have nothing to hide, then don't hide anything. Score another one for McIntyre's credibility.

The last word on this is yours if you'd like it.

andrewt said...

Roger, I don't understand the remarks about lawyer-like parsing. If your method is invariant to linear transformation of the data then you don't worry about linear factors - its maths not law.

I'm not sure what you believe Mann et al. needs to clarify - do you have some reason to believe Mann et al are incorrect to say their methods are insensitive to the sign of the data? Which words in Kaufman's et al.'s draft suggest that to you - I can't see anything even vaguely like that.

bernie said...

Roger:
Well said. Funny how McIntyre seems to win all these technical debates with Mann et al.

deancrowe said...

Roger -

So you're reducing your claim from outright rejection to speculating on reasons why RC didn't post McIntyre's comment?

I don't know why McIntyre's comment wasn't posted on RC. Given the bad blood between the two, I could hazard a guess, but it would be just that, a guess (and I wouldn't rule out trying to hide something).

I'm not sure what you think my interpretation here is. I said that Mann stated the sign didn't matter - that's a fact.

That's also a fact that McIntyre does not even mention in your Dot Earth link even though Mann's response to him that the sign didn't matter was published in PNAS 8 months before McIntyre's post on Dot Earth:
http://www.pnas.org/content/106/6
Since you're insinuating motives here, why do you suppose that is?

Paul Biggs said...

Andrewt - McIntyre is well qualified in the proxy data debate - it's a statistical problem - so someone with an obvious grasp of statistical methodology and proxy data, which McIntyre has, IS well qualified. Wasn't it Mann who said "I am not a statistician?" I think we have known for some time, if we are honest, what Mann actually is.

Roger Pielke, Jr. said...

-17-andrewt and -19-deancrowe

Let's review.

1. Did Mann deny that he used the proxy upside down?

I say, yes, fairly obviously with his "bizarre" claim.

You say, no, because you argue that Mann says that there is no such thing as up or down in the proxy, so the accusation of upsidedown-ness itself is bizarre.

2. OK, lets pursue your reasoning -- this gets us to the question, is Mann's claim of upsidedown-ness being irrelevant a valid defense?

Well, actually no if you believe the original authors of the proxy study, Atte Kohorla and now Kaufman, all of whom agree that there is an up and an down here.

So your defense of Mann involves arguing that he is wrong about the substance of the matter. This is not a compelling line of argument.

Note that the upsidedown-ness of the proxy is a different question than whether the use of an upside-down proxy "matters" in the context of a reconstruction. Apparently for Kaufman making the correction improved his results. In science getting things right is important.

In this case Mann either denied using the proxy upsidedown (my view, and I think a fairly obvious one) or, if your chose to parse his worse as you have, he did not deny using the proxy upside down but instead says that there is no up or down. However, this argument is substantively incorrect.

Either way, the argument resolves in McIntyre's favor. OK, that is my final last word ;-)

DaveJR said...

deancrowe wrote:
"It looks pretty clear to me that he wrote that the claim was bizarre because the sign doesn't matter and never outright rejected the claim that he used proxies upside down."

Which would be correct so long as the proxy was interpreted one of two ways:

Decreasing proxy values = increasing temperature
Increasing proxy values = increasing temperature

However, the interpretation given to the proxy in the original paper was:

Decreasing proxy values = decreasing temperature

Whereas Mann flipped the sign to give the interpretation that:

Decreasing proxy values = increasing temperature

So his statement was completely wrong.

Using the proxy the right way up by Kaufman didn't have much effect for one very good reason. He had already chopped off the "hockey stick" in the modern part because the authors showed that this region was contaminated by man-made sources. This "hockey stick" otherwise showed temperatures *plummeting* in the modern period. However, when Mann used the proxy he included the contaminated region and reinterpreted it to show that temperatures were rising dramatically! Perhaps an easy mistake to make with confirmation bias and all that, but the fact he not only point blank refused to address the issue when it was pointed out but ridiculed it as well, IMO, speaks volumes about his scientific integrity.

I think it's pretty clear by now that the difference between CA and RC is that CA deals in ugly facts and RC protects beautiful theories.

lkdemott said...

"Sometimes seemingly small disputes can tell you a lot. "

I am not certain what you think the resolution of this dispute can tell us.
Although much of the discussion that occurs over at Climate Audit is way over my head, the point McIntrye makes on upside down proxies seems self-evidently correct to me and easily verified by an examination of the data. That the mistake was not caught in the initial per review suggests a problem. That the journal allowed Mann to deny a problem existed at all without any editorial comment suggests a deeper problem. It was only after McIntrye and others continued to press the point, that the error was admitted in the peer review literature. So is this the lesson: "when a fundamental and obvious mistake is published in peer review literature, it might eventually be corrected if sufficient outside pressure is exerted to embarrass the authors and the journal"?

Andrew said...

I think a confusion is arising because people don't quite know what "Multivariate regression methods are insensitive to the sign of predictors" actually means. If people did, they would realize it is quite unsettling. While it sounds like Mann is saying that even if the proxy was flipped upside down, the method would flip it right side up in reality just the opposite would happen. At this point every body will say "Huh?". Let me explain. Mann's Multivariate Regression technique specifically sifts through the proxy network for "predictors" that can explain the observed temperature history, it does not ask whether such an arrangement makes physical sense. So, if you have a series like the one in question that suddenly spikes in the twentieth century in a manner that would normally indicate cooling, the multivariate method uses the spike to produce the up trend, or part of it, in the observed history-so that as you go back in time, cool becomes warm and vice versa because the regression method turned the proxy upside down.

That's truly disturbing and I can't see how anyone can justify it. It is hard to believe the defense is that it "doesn't matter".

Mark B said...

@Andrew

This is it folks - Andrew gets it right:

"Let me explain. Mann's Multivariate Regression technique specifically sifts through the proxy network for "predictors" that can explain the observed temperature history, it does not ask whether such an arrangement makes physical sense. "

When reading any scientific paper, the first thing you should be asking is "Does it make any sense?" Mann's proxies were hidden behind a wall of secrecy, so no one reading his paper could answer that question. McIntire's demands for data revealed proxies that fail the "Does it make any sense?" test. In science, you need more than correct equations - you need your equations to actually represent something in the physical world. Mann calls this requirement "bizarre."

Tom C said...

Yes. Andrew explained it well in his comment. The real scandal here is that Mann's reply was accepted by Science and (through continued silence) by his dendro colleagues. apparently Kaufman thought it politically risky to follwo suit.

andrewt said...

Roger what you are saying looks like lawyer-like parsing.

Shouldn't you be taking your own advice and not engaging in debate in an area where you are professionally unqualified ?

Roger Pielke, Jr. said...

-27-andrewt

I am indeed professionally qualified to call BS when I see it ;-)

Belette said...

Hmm, interesting stuff.

1) I don't see K admitting to using upside down data, but I may have missed it. Which text are you reading as saying so?

2) Several people have said this already, but you don't seem to have got the point. So let me ask you a question: do you understand that for Mann et al.'s method the sign of the proxy doesn't matter?

Roger Pielke, Jr. said...

-29-Belette

Thanks for dropping by . . .

1) http://www.arcus.org/synthesis2k/synthesis/Kaufman_et_al_SOM_corrected.pdf

"Record 20 was corrected to reflect the original interpretation of Tijander et al. (S32) that X-ray density is related inversely to temperature.

Record 21 was corrected to reflect the interpretation of Haltia-Hovi et al. (S33) that varve thickness is related inversely to temperature."

2)I point you to my comments in #21:

"is Mann's claim of upsidedown-ness being irrelevant a valid defense?

Well, actually no if you believe the original authors of the proxy study, Atte Kohorla and now Kaufman, all of whom agree that there is an up and an down here.

So your defense of Mann involves arguing that he is wrong about the substance of the matter. This is not a compelling line of argument.

Note that the upsidedown-ness of the proxy is a different question than whether the use of an upside-down proxy "matters" in the context of a reconstruction."

I am well aware of Mann's claim that the orientation of the proxy does not "matter" based on his methodology. Let's grant this as true (recognizing that Mann's methodology is contested by SM, but no matter for present purposes).

My response questions are:

1) Are you aware that the original authors of the proxy study, Atte Kohorla and now Kaufman, all that there is a proper up and down in this context and this is at odds with how Mann used the proxy?

2) Can Kaufman's statement (above) about the proper orientation and interpretation of the proxies be reconciled with Mann's statement that orientation does not matter? (If so, please do tell)

andrewt said...

Roger I'm puzzled you've said that people can't engage in the substance of debate over your work unless they are professionally qualified.

Yet you now imply this doesn't apply to the field of paleoclimatology.

What is the difference?

Roger Pielke, Jr. said...

-31-andrewt

You'll be able to resolve your puzzlement by re-reading my comments that you inaccurately paraphrase.

andrewt said...

Sorry Roger I'm still puzzled you said:

"Among these minnows are controversialist bloggers like Tim Lambert, who are professionally unqualified to engage in the substance of most debates (certainly the case with respect to my own work)"

Which suggested to me that, you would believe that professionally qualifications would be needed to debate a paleoclimatology paper.

Now I would not have said myself that professional qualifications were needed, but looking at this discussion I am coming around to your position that they are needed.

steve said...

William Connolley and other defenders of Mann's handling of the Tiljander proxies have no understanding of what's going on here. The recent portion of the Tiljander sediments is not a climate "proxy" - it is contaminated by bridgebuilding and ditches, a point clearly stated in the articles. Mann's "method" seizes on a correlation between increased sedimentation due to nonclimatic ditches and world temperature, so that the sign of the portion of the record believed to be climatic (before 1800 or so) is turned upside down. It's a real defect. Even climate scientists should be able to understand this. It's crazy that people are debating it.

What makes it worse for Mann et al 2008 is that the nonclimatic sedimentation creates a really large blade in the Tiljander proxies. Kaufman cut the series off at AD1800 and doesn't have as bad a problem (with this particular issue).

AMac said...

Thanks to all for a discussion of the "upside-down" controversy that is accessible to a layperson.

Because the PNAS comments are so terse, the parts concerning the "Tiljander proxies" can be quoted in full. (These proxies are a set of measurements of the 'X-Ray Density' of lakebed varves (annual layers of sediments) from Lake Korttajarvi, Finland. Tiljander says that low XRD corresponds to high organic content, which is believed to be a proxy for favorable (warm) climate. She also says that the 20th century XRD record is highly affected by nearby human activity.Link.

In PNAS, McIntyre & McKitrick wrote, "[Mann et al.'s] non-dendro network uses some data with the axes upside down, e.g., Korttajarvi sediments, which are also compromised by agricultural impact."

Mann et al. replied, "The claim that “upside down” data were used is bizarre. Multivariate regression methods are insensitive to the sign of predictors. Screening, when used, employed one-sided tests only when a definite sign could be a priori reasoned on physical grounds. Potential nonclimatic influences on the Tiljander and other proxies were discussed in the SI, which showed that none of our central conclusions relied on their use."

On page 2 of the Supplemental Information to the paper in question (Mann et al. 2008 PDF), Mann wrote, "...we also examined whether or not potential problems noted for several records (see Dataset S1 for details) might compromise the reconstructions. These records include the four Tijander et al. (12) series used (see Fig. S9) for which the original authors note that human effects over the past few centuries unrelated to climate might impact records (the original paper states 'Natural variability in the sediment record was disrupted by increased human impact in the catchment area at A.D. 1720.' and later, 'In the case of Lake Korttajarvi it is a demanding task to calibrate the physical varve data we have collected against meteorological data, because human impacts have distorted the natural signal to varying extents’)… We therefore performed additional analyses as in Fig. S7, but instead compared the reconstructions both with and without the above seven potentially problematic series, as shown in Fig. S8."

The four Tijander series are XRD, "lightsum", "thickness", and "darksum." Figure 8a of the SI shows Temperature Anomaly vs. Time for Mann's original NH CPS data set (green line) and the data set with the Tijander series and three others removed (black line). The two lines are essentailly identical, including from 1720 through 2000. ("NH CPS" is Northern Hemisphere Composite Plus Scale, derived from 15 data series, including the four from Lake Korttajarvi.)

-- continues below --

deancrowe said...

Roger #21

I didn't parse Mann's words, I took them exactly as he wrote them. It was you that -omitted- the part where he said the sign doesn't matter, removing the "bizarre" sentence from it's context, and inserted meaning into what he did not say. If you want think his saying the sign doesn't matter is tantamount to his denying using the proxy upside down, fine, but that's your interpretation. But the fact still remains that he did not say whether he used the proxy upside down or rightside up. Heck, one could even take his not saying which way he used the proxy as a tacic admission that he did use it upside down.

You point out that Atte Kohorla and Kaufman agree there's an up and down to the proxy - no problem from me - I'm not the one saying it doesn't matter.

Then you say my "defense" of Mann is that "he is wrong about the substance of the matter" - um, no, I'm not defending Mann at all. If you had said he weaseled out of admitting he used the proxy upside down, I would have had no problem with that - understand?

You say "In science getting things right is important" - to which I whole heartily agree. To which then the question obviously is whether Mann used the proxy correctly and if he is correct in saying the sign doesn't matter. Whether he -denied- using the proxy upside down or not doesn't change the results (right or wrong) one whit.

AMac said...

-- continued from above --

With this background, perhaps knowledgable parties could comment on the following points.

A. It seems to me that Mann et al. could have used the Lake Korttajarvi varve XRD series in one or more of three ways:

(1) As a temperature proxy (ie lower XRD indicates warmer or, upside-down, lower XRD indicates colder),

(2) As a measure of reliability or variance of the Temperature Anomaly vs. Time constructions (ie higher XRD leads to larger--or smaller--error bars),

(3) Something else (eg some training or data-filtering function not directly affecting the Temperature Anomaly vs. Time construction.

B. Mann called McIntyre's claim that upside-down data were used "bizarre." Pielke takes this to mean that Mann is dismissing McIntyre's claim as preposterous, i.e. as false. Andrewt and Dean Crowe interpret bizarre to mean that upside-down data wouldn't affect the outcome of Temperature Anomaly vs. Time plots (#15, "what [Mann] wrote was the methods are insensitive to the sign, and that is the reason he gave for calling the claim bizarre.") But Mann et al's paper is titled "Proxy-based reconstructions of hemispheric and global surface temperature variations over the past two millennia." What, then, is the "something-else" function that is insensitive to sign? Tijander recovered and studied the Lake Korttajarvi varve series because she believed it supplied a temperature proxy. If Mann et al discarded the temperature-proxy information in this data: Why? And why is the paper and the SI written in such a way as to imply that the varve series is a temperature proxy?

This just doesn't make sense.

C. Tijander told McIntyre that the Lake Korttajarvi varve series contained instances of severe contamination in the 20th Century, particularly nearby peat ditching in 1930 and bridge construction in 1967. Link. These two events can be seen as the two clear spikes at the extreme left of three of the four Tijander series (including XRD) in Mann's Fig. S9, despite time-averaging.

Despite the human-activity problems with the Tijander varve record and the added upside-down problem of the Tijander XRD series, removing these four series (and three other "problematic series) from the Temperature Anomaly vs. Time graph (Figure S8) makes no discernable difference to the resulting trace of Northern Hemisphere temperature. In other words, a plot of temperature deduced from the set of {eight "good" Northern Hemisphere data series} is virtually identical to the plot deduced from set of {eight "good" series plus three "possibly problematic" series plus three "contaminated" Tijander series plus one "contaminated and upside-down" Tijander series}.

This just doesn't make sense.

A reductio ad absurdum thought experiment: What if Mann et al. had accidentally inverted the fourteen remaining NH CPS time series, in addtion to the Tijander varve XRD series? Would that have affected the Temperature Anomaly graph?

If not: what data are the graphs in Mann et al built on?

"The magnitude of the variance matters, but not its sign" explanation doesn't seem to survive this test.

deancrowe said...

DaveJR #22

"deancrowe wrote:
"It looks pretty clear to me that he wrote that the claim was bizarre because the sign doesn't matter and never outright rejected the claim that he used proxies upside down."

Which would be correct so long as the proxy was interpreted one of two ways:

..."

My point wasn't than Mann used the proxy correctly, and I don't say that he did. My point was that he didn't -deny- the claim, rather -he said- the claim was "bizarre" because the sign didn't matter.

Roger Pielke, Jr. said...

-33-andrewt

Perhaps if you include the sentence that I wrote immediately prior to the one that you select, your puzzlement will be solved:

"But even the big fish apparently see some gutter behavior as not really becoming of professionals (though Romm doesn't seem to care), as to more effectively attack someone's reputation they also rely on the minnows of the blogosphere, people who see it as their sole job to "trash" someone's reputation via innuendo, fabrication and outright misrepresentation."

The issue that I raise in those two sentences is the irony that some professionals cite those who are professionally unqualified as some sort of appeal to authority. It should work the other way, no?

In any case, in this post I am citing Kaufman and Kohorla, who are both professionally qualified to comment on the subject, and who have offered evidence in favor of Steve McIntyre's side of this dispute. In such disputes McIntyre doesn't get the benefit of the doubt like Mann, Kaufman and Kohorla because he is an amateur not a professional (as my colleague Ben Hale has recently argued) and has a high burden of proof to meet in winning such disputes. This does not mean that McIntyre cannot engage in such debates (what is the point of a blog, anyway?), only that it would be odd for professionals to cite his views as authoritative, without meeting this high burden of proof.

In this case, it is my view that Kohorla's opinion and Kaufman's corrigendum provide sufficient evidence for me to decide that this point of dispute has been resolved in McIntyre's favor.

You may not agree with my judgment (and that of Kohorla and Kaufman), but that is what it is. Perhaps you'd like to have a go at the two questions I left above for William Connolley? Maybe you can educate me on what I have missed here that seems so blindingly obvious ... here they are:

1) Are you aware that the original authors of the proxy study, Atte Kohorla and now Kaufman, all that there is a proper up and down in this context and this is at odds with how Mann used the proxy?

2) Can Kaufman's statement (above) about the proper orientation and interpretation of the proxies be reconciled with Mann's statement that orientation does not matter? (If so, please do tell)

Andrew said...

"Now I would not have said myself that professional qualifications were needed, but looking at this discussion I am coming around to your position that they are needed."

Steve McIntyre has published papers critiquing multi-proxy methodologies from a statistical point of view. As far as you know, many others here could have any number of qualifications or lack there of.

But all I ever needed was to know what "multivariate regression" means and the nature of the Tiljander and other proxies.

Roger Pielke, Jr. said...

-36-deancrowe

I believe that we have a meeting of the minds:

"If you want think his saying the sign doesn't matter is tantamount to his denying using the proxy upside down, fine, but that's your interpretation. But the fact still remains that he did not say whether he used the proxy upside down or rightside up. Heck, one could even take his not saying which way he used the proxy as a tacic admission that he did use it upside down."

Roger Pielke, Jr. said...

-34-steve (mcintyre?)

How is this for a plain English reinterpretation of your comment?

What you are arguing is that the time period which Mann uses to calibrate the proxy is the time period when the proxy actually diverges from a representation of temperature due to human influences (bridges, ditches, whatnot). It then takes this (incorrect) calibration and applies it to the pre-calibrated record to create a temperature history but this history is upsidedown because of the divergence in the more recent record used to establish the correlation.

So if there is no meaningful overlap between the proxy and the instrumental record this would disqualify it from being used under Mann's methodology?

AMac said...

Roger Pielke #41 --

The key finding of Mann et al. (2008) is stated in its title, "Proxy-based reconstructions of hemispheric and global surface temperature variations over the past two millennia." Figures 2 and 3 are reconstructions of northern hemisphere temperatures, based on a series of 15 proxy records (Figure S9).

How can McIntyre's correct statement that one of these 15 proxies is upside-down be bizarre?

If the upside-downedness of the proxy doesn't affect the reconstruction (it doesn't; Figure S8a), what purpose does it serve in the work that Mann et al. report? Why wasn't the Lake Korttajarvi varve XRD series excluded as trivial?

If the upside-downedness of one proxy doesn't affect the reconstruction: what if two of the 15 series are flipped. Three? Four?...

This "absurd" question forms the basis of a powerful and easily-performed experiment.

Use the Mann et al. code, and re-run the analysis with one, then with two, then with three (and so on) of the 15 proxies flipped, and chart the resulting Temperature vs. Time reconstructions.

Mann and McIntyre should each predict the sensitivity of the derived curves to these intentionally corrupted data sets.

I presume that Mann's computer code has been made public as a condition of public grant support, and again as a condition of publication. Is that indeed the case?

steve said...

#43. You have to watch the pea under the thimble. The distinctive claim of Mann et al 2008 was that they could "get" a stick without dendro proxies. In that reconstruction, they used upside down Tiljander. In their "proof" that the problematic proxies didn't :matter", they re-inserted the dendro proxies - which included a Graybill strip bark dendro proxy of the very type that the NAS panel agreed should be avoided. We also noted that, despite their lipservice to NAS panel recommendations, they flouted them when it came to strip bark bristlecones (not even using the most up-to-date Sheep Mt version from Linah Ababneh which doesn't have a HS shape.) Mann's response to this was equally absurd.

steve said...

Mann's computer code has been made public - though, unfortunately, this is not required as a condition of public support. In this respect, good for him.

Even with the code, it takes quite a bit of work to be able to do sensitivity analyses as the code is awkwardly structured and there are some puzzling aspects to the overall algorithm. PNAS has a very short deadline for comments (90 days and 250 words) and at the time of the deadline, we had a handle on his CPS method, but not his RegEM method. Shortly afterwards, Steig et al came out. That proved to be a much more tractable data set for benchmarking RegEM, which has some odd features. At some point, I'll revisit Mann et al 2008 and finish the assessment of this and other issues.

steve said...

#42. I agree with your point. To the extent that Mann's method requires calibration, these records can't be calibrated.

AMac said...

Re: my comments #35, 37, and 43, I made a stupid mistake. I posted on this subject at a Mann-friendly site, and got a helpful correction.

The following was my reply, and question --

Thanks, Rattus Norvegicus! I see that I'm asking about a smaller issue (the 15 screened proxies available for the NH CPS reconstruction back to the 9th century and earlier, end of pg. 13255 ff). Your point on the broader issue is that there are much, much larger numbers of proxies available for more recent times (red symbols in Fig. 1).

McIntyre claims that all four of the Lake Korttajarvi varve proxies had their signs reversed in Mann et al (2008), not just the XRD one.

Thus, I remain perplexed by Fig. S8a. That depicts the NH Land Temperature Anomaly reconstructions, performed (1) with the full proxy network (green line), and then (2) with the eight proxies remaining after the removal of seven potentially problematic series (black line).

The full proxy network would be the fifteen NH proxy records that pass the screening procedure back to AD 818 or earlier, shown on Fig. S9. Of the seven removed for the black line, four are the four flipped-around Lake Korttajarvi series. Yet the two reconstructions look almost identical, from 200 AD through to 2000 AD.

With so few proxy series, shouldn't this removal have led to a very different-looking curve?

AMac said...

AndrewT replied to my #47 supra over at Stoat. I posted this follow-up, but it was trimmed to a stub by WMC as leading away from the point under discussion (he kindly notified me of that decision).

Steve M, perhaps you could comment on Figure S8a in Mann et al (2008)? I believe it is meant to show the derivation and pedigree of the "CPS land" curve in Figure 3, which is presented as an important long-term Northern Hemisphere temperature reconstruction. If so, it is at the heart of the paper.

- - - - -

andrewt, thanks.

> if a method is insensitive to sign, you should be able to reverse the polarity of any or all of the inputs and the output should remain unchanged.

As a general statement, this is certainly true, and it seems to apply to the four upside-down Lake Korttajarvi varve proxies and the reconstruction in Figure S8a. But that's what puzzles me: what, exactly, does Figure S8a show?

It seems to me to be a graphical depiction of the Northern Hemisphere's temperature anomaly for the past 1,800 years, as calculated based on the use of the 15 (green line) or 8 (black line) very-long-duration NH CPS proxy series that passed screening criteria, and that are represented in Figure S9. This interpretation would seem to be consistent with the Figure's subject matter and legend.

But if that's correct, it seems to take us to very strange territory, when combined with your observation.

I might say, "I have a small number of proxy series that all indicate cooling in certain decadal periods C, and warming in other decades W. From them, I can construct an NH Temperature Anomaly vs Time curve." Sounds plausible.

Then I might say, "To those proxies, I add four additional proxy series that all indicate warming during periods C, and cooling in other periods W. From this new group, I construct an NH Temperature Anomaly vs Time curve that is essentially identical to the first one I made."

That would mean--adding data that is consistent with my a priori view of paleoclimate, supports that view. But--adding data that is inconsistent with my a priori view of paleoclimate, also supports that view.

Surely this sort of non-falsifiablility is a major caveat that should be presented to readers. I don't see anything in the article or SI that discusses Figure S8a along these lines (though I could have missed it). On the other hand--if Fig. 8a is not a Temperature Anomaly vs. Time reconstruction that is derived from long-duration proxy data: then what is it?

edaniel said...

I understand the situation to be as summarized by AMac.

So therefore to me it seems that any arbitrary info can be tossed into the process and out comes a temperature.

Any connections to causality by appropriate physical phenomena and processes are shattered beyond comprehension.

This can't possibly be correct.

Tom C said...

Roger -

To reduce this issue to the simplest form, Mann is (as always) fishing for a correlation. He throws in anything that has a large 20th century uptick whether the methodology was bad (BC pines) or the interpretation is wrong (varve thickness). He says "you can get a stick without trees" (look, nothing in my left hand) as he includes the varves; then he says "I can get a stick without Tiljander" (look, nothing in my right hand after the ball was transferred from the left hand) as he puts the BC pines back in.

When this game is challenged, his fallback is "teleconnection".

Trying to pin him and his assorted defenders down is, as Steve notes, an effort to follow the shell game.

Roger Pielke, Jr. said...

-50-Tom C

William Connolley aka "Stoat" (and "Bellette" above) is plenty smart enough to figure out that he has gotten this one very very wrong.

Lets give him a chance to figure that out and see how he responds.

Xinghua said...

Let all who wonder who is more trustworthy, McIntyre or the entire science of Paleoclimatology, judge based on this one point.

My judgement is this: A temperature reconstruction, by definition, claims to represent temperatures. If the methods used to create that reconstruction can be shown to be negatively correlated to temperature in a given location, then it is very hard to imagine calling this a temperature reconstruction.

Had the medieval Tiljander data indicated warmer temperatures, it would cause the medieval reconstruction to be colder. This is powerful evidence against Mann's curve being a reconstruction.

By treating this not as a mistake, but a deliberate decision, Mann is suggesting that his temperature reconstructions need not represent temperature in order to be valid.

It is incredible to me that he allowed the argument to get to this point (why not just admit the error), but his work and reputation are probably not going to survive this.

AMac said...

Xinghua #52 --

> Had the medieval Tiljander data indicated warmer temperatures, it would cause the medieval reconstruction to be colder.

The validation period for the 15 screened Northern Hemisphere CPS proxy series was 1850-1995 (pg. 13254). Tiljander states that natural signals in the Lake Korttajarvi varve [annual lakebed sediment deposit] proxies are disrupted after ~1720 by nearby human activities (SI, pg. 2). Thick, mineral-rich varves due to ditch-cutting in 1930 and bridge-building in 1967 are prominent in Fig. S9.

The CPS method was applied in three different ways. Full CPS (calibration 1850-1995), Early (1850-1949) Calibration [Validation 1950-1995], and Late (1896-1995) Calibration [Validation 1850-1895] (pg. 13254).

The Full, Early, and Late 15-proxy reconstructions have the same general shape from 400 to 1850, though there are periods of divergence prior to 1700 (Fig. S11a).

For the varve proxy series, all three CPS calibration approaches appear to have led to the correlation of the deposition of thick, mineral-rich varves with higher temperatures.

Thus, it would appear that the four Lake Korttajarvi proxies contributed a "warmer" signal to the 400-1850 CPS reconstructions in the decades when varves were thicker and more mineral-rich. And a "cooler" signal when varves were thinner and included more organic material. (See orientation of Tiljander proxy plots in Fig. S9, and CPS curves in Figs. 2, 3, S7, S8, S10, S11, & others.)

This appears to be the opposite interpretation to the one offered by Tiljander, who correlates thinner, more organic-rich varves with higher temperatures. Kaufman accepts Tiljander in his 10/9/09 correction.

If this description is correct, it would appear that there is a "sensitivity to sign" in the CPS method of analyzing climate proxies to derive reconstructions of hemispheric temperature anomalies.

[Also at Stoat.]

0gOuk4BqqICK5EzCp30f4pxexef3ww-- said...

I went through the Mann 2008 code once again to see what are exact effects on the sign change for Tiljander series.
They are:
1) Anything based on screened network needs to be recalculated as Tiljander proxies do not survive screening (two-pick correlations are negative as they should be >0.106 to pass screening)
2) CPS without screening: sign does not matter; the orientation is always wrong for Tiljander series. CPS includes a step in which all proxies are flipped to the orientation such that correlation in the calibration period is positive,
3) EIV without screening: sign does not matter as it is decided in the RegEM step and may change from reconstruction time step to another. It seems that most of the time it is wrong for Tiljander series.

BRs, Jean S

Roger Pielke, Jr. said...

Thanks Jean S ... Still waiting for WC to do the right thing.

Jo_Calder said...

WC's comment "[Kaufman] is right and Mann is right" (http://scienceblogs.com/stoat/2009/10/oh_dear_oh_dear_oh_dear_oh_dea.php#comment-2029291) is fascinating, because it illustrates a new form of logical operation which depends both on the truth value of the propositions involved and their utility. So, 'p and q' is true just in case either of p or q is true and if either p or q is not true, it has utility. The conditions for 'or' are weaker still, where it suffices for either 'p' or 'q' to have utility. The definition of 'not' is left as an exercise to the reader.

Belette said...

Still waiting on me? Your posting software ate my comment. Ah well, I'll try again.

#1: yes I agree: K say its upside down.

#2: You say: I am well aware of Mann's claim that the orientation of the proxy does not "matter" based on his methodology. Let's grant this as true...

I don't think that will do. You need to know. Or find someone to tell you. If you don't know, this entire post of yours is meaningless.

Your response questions were:

1) Are you aware that the original authors of the proxy study, Atte Kohorla and now Kaufman, all that there is a proper up and down in this context and this is at odds with how Mann used the proxy?

: Answer: yes, and no.

2) Can Kaufman's statement (above) about the proper orientation and interpretation of the proxies be reconciled with Mann's statement that orientation does not matter? (If so, please do tell)

: Yes. You're confusing the individual proxy reconstruction, where the sign does matter, with the regressed overall proxy reconstruction, where the signs of the individual inputs don't need to matter. From the paper: Reconstructions were performed based on both the “full” proxy data network and on a “screened” network (Table S1) consisting of only those proxies that pass a screening process for a local surface-temperature signal. The screening process requires a statistically significant (P < 0.10) correlation with local instrumental surface-temperature data during the calibration interval. Where the sign of the correlation could a priori be specified (positive for tree-ring data, ice-core oxygen isotopes, lake sediments, and historical documents, and negative for coral oxygen-isotope records), a one-sided significance criterion was used. Otherwise, a two-sided significance criterion was used. Further details of the screening procedure are provided in SI Text. which tells you explicitly that they don't know the sign of some of the proxies.

Jean S's comment #1 above seems likely: if the T proxy is thrown out because of the screening because it has the wrong sign, then it won't be included in the reconstruction. His #2 and #3 is what I've been trying to tell you (did you read it?).

Roger Pielke, Jr. said...

-57-Belette

Now you are dissembling.

Funny that you use Jean S as a point of authority. Are you reading along at CA? Apparently not:

Jean S:

"One thing I have never understood is that how it is possible after all these demonstratably incorrect statements, whose validity everyone can check from the code/papers, we still find relatively many intelligent people (like Connolley) desperately trying to defend Mann. That's truly "bizarre"."

http://www.climateaudit.org/?p=7556#comment-363564

And there is more there, including code if you are so inclined.

Are you going to do the right thing?

MrUsufruct said...

It seems that William has a habit of making...er..."interesting" arguments.

http://en.wikipedia.org/wiki/User_talk:William_M._Connolley/The_science_is_settled

AMac said...

Belette #57 --

Jean S wrote in #54

2) CPS without screening: sign does not matter; the orientation is always wrong for Tiljander series. CPS includes a step in which all proxies are flipped to the orientation such that correlation in the calibration period is positive.

Jean S stated that "the orientation is always wrong" for the Lake Korttajarvi proxies. This is because Mann et al's CPS calibration window (1850-1995) assigns the "warmer" signal to varves that are thicker, more mineral-rich, and exhibit higher X-Ray Densities.

According to Tiljander and corrected Kaufman, this is the opposite of the natural signal, where thicker, more-mineral rich, higher-XRD varves signify cooler temperatures.

Jean S and I agree that Mann et al's CPS protocol gets this wrong because they inadvertantly calibrate varves to local human activity (e.g. ditching and bridge-building), not to climate.

Thus, under this CPS scheme, the Lake Korttajarvi proxies work to corrupt pre-1850 estimates of Temperature Anomalies, contributing "warm" where "cold" is warranted, and "cold" in place of "warm."

You suggest that you agree with Jean S's point 2 at #54. In that case, it would seem that you are in accord with this interpretation.

Is that the case?

(And Jean S, please correct any misstatements of your stance.)

Belette said...

Hi Roger, this is getting weirder and weirder. If you have questions for me, you know where I live. PLease don't rely on me seeing them here or on CA.

I'm not even sure what you mean by "right" (other than confessing that you're always right about everything?)

You asked two supplementary questions. I've answered them. The answers (and Jean S's, sorry if you don't want to hear what he is saying either) confirm what I said before and that you've failed to understand: I can repeat it, but I'm not sure how I can make it any clearer: You're confusing the individual proxy reconstruction, where the sign does matter, with the regressed overall proxy reconstruction, where the signs of the individual inputs don't need to matter.

Mr U: come on over to wiki, if you think you're hard enough.

AMac said...

I offered a response to this comment by WMC/Bellete on Bender's analysis of the meaning of the varve series and its flipped sign. It is in the moderation queue at Stoat and at the 'Kaufman' thread at Climate Audit.

AK said...

I'm a layman, so I don't understand the technical issues here. But it seems that Roger is making a logical error here by confusing "mistake" with "matters." In the Kaufman paper, correcting for whatever mistake Roger and Steve McIntyre are accusing the authors of does not change the shape of the hockey stick:

"The corrected temperature trend through 1900 (green line in Fig. 3C) is - 0.23° ± 0.06°C per 1000 years rather than -0.22°C ± 0.06°C per 1000 years as originally reported. The corrected regional sensitivity of summer (JJA) temperature to orbital forcing inferred from the proxy-based reconstruction is 0.07° ± 0.03°C per W m-2 rather than 0.07° ± 0.02°C per W m-2 as originally reported (Fig. 4A)."

Unless I'm missing something, this seems like a deception tactic - pointing out an error in calculations to create doubt in a conclusion's validity, without acknowledging that the error was not material to the conclusions themselves. Care to enlighten me?

Roger Pielke, Jr. said...

-63-AK

This post is not about the validity of the hockey stick.

Jason S said...

#63,

If you make only the changes in the Corrigendum, it doesn't "matter" in the sense that the graph remains the same.

BUT, if the you replace the questionable series with other peer reviewed studies from substantially similar sources [Yamal with Polar Urals], it DOES matter in this sense.

The bottom line is this: If some valid proxy choices by Kaufman produce a hockey stick, and others do not, Kaufman's result tells us little about historical temperature, and much about his proxy selection process.

Kaufman had to show us that his proxy selection was made without bias. He has to claim that he chose the highest quality proxies, not just the ones which show a hockey stick. The fact that so many of his choices were flawed, casts serious doubt on this.

[BTW, I don't think that Kaufman is actually the bad guy here. This was a big study with many authors. I think he just wanted to get it done, and was forced into compromises which he didn't fully appreciate at the time. But removing the mistakes can not resuscitate the basic result, which is primarily dependent on proxy selection.]

Jason S said...

Connolley is clearly starting to get it.

He was so determined to prove Steve wrong, he got caught up in the semantic details of what "upside down" means instead of looking at the bigger picture.

Unfortunately for him:

1. Steve's already got a post up embarrassing him.

2. Unlike Michael Mann, he's got too much integrity to go on claiming that he is still correct in the face of obvious evidence to the contrary.

edaniel said...

It seems that maybe two ( or, very likely, more ) issues have gotten tangled up.

As I understand the situation, Steve McIntyre originally used the phrase 'upside-down'. The statement being made relative to the physical causality of temperature for the characteristics of the proxy.

Somewhere along the line, 'wrong sign' seems to have been substituted for 'upside down'. Not the same thing, but at the same time a somewhat useful characterization.

And now, someone has noted that in the maths used for processing the data, the sign attached to the data doesn't make any difference relative to the calculated numbers.

The latter, IMO, is not the issue. It might even be construed to be a straw man thrown out as a distraction from the original issue.

The issue, IMO, is that a proxy has been used in a way that does not correctly reflect the physical causality that has been assigned to the proxy by the original investigators. Or, more directly, the proxy is being incorrectly used.

The mathematical methods, assuming that they have been properly verified, don't even enter into the discussions of the original issue.

Does is make any sense at all to be concerned with wether or not a mathematical method is being correctly applied to incorrect information?

Corrections will be appreciated.

AMac said...

The comment I referred to at #62 supra does not appear to have passed moderation at Stoat. A version can be read as #86 at Climate Audit.

AK #63 --

I am a layman too. I walk through one technical issue I can understand in this Cruel Mistress comment.

The upside-down nature of the Lake Korttajarvi varve proxies that McIntyre alleged (see links in the body of the post) is now established, to my mind.

Bender (and, earlier, McIntyre, and others) have explained why this matters to the paleoclimate reconstruction of Mann et al (2008).

I explain in the linked comment why I think that this discovery calls for a hard look at Mann's Figure S8a. The figure was put in the manuscript to demonstrate the robustness of the proxies and of the CPS method, and thus of the calculated reconstruction curves.

In light of the mishandling of the Lake Korttajarvi varve proxy calibration, Figure S8a instead suggests that there may be serious errors in the way Mann et al performed their CPS-based temperature anomaly reconstruction analysis.

Jason S said...

#67,

That's really not a fair characterization.

The sign of the correlation between Mann's reconstruction and Tiljander _IS_ inverted relative to the sign of the correlation between local Temperature and Tiljander. This is due to a spurious correlation between non climactic elements of Tiljander (related to bridge building and other human activities), and recent temperature.

What happened here is that Connolley (a smart guy with a strong bias towards Mann [his former RC collaborator]) got caught up in mathematical minutia and proving McIntyre wrong, and never made a good faith effort to understand the actual argument (until he had already embarrassed himself).

AK said...

Well think about it. Mann's data is in line with other analyses that confirm the existence of a medieval warm period and "little ice age." As Robert Grumbine points out, if Mann used the data upside down, then using it "right side up" would invert his results, creating a "medieval cool period" and "little warm age," not to mention cooling in the present day in contradiction to the instrumental record. So if Mann made an "error" in using data "upside down," it doesn't matter to the conclusions.

http://moregrumbinescience.blogspot.com/2009/10/sound-and-fury-at-wuwt.html#more

Jason S said...

#70,

Some Mann proxies are upside down. Others are right side up. The two cancel each other out.

What you are left with is a straight line for the period preceding the instrumental temperature record, and a rapidly rising line during the recent warming.

A straight line followed by a rapidly rising line looks like what?

Captain Curt said...

One of the side issues here that amuses/angers me is the claim (e.g. by andrewt and Connelly) that only professional climate scientists can understand mathematical issues with multi-variate regression, as if the technique had been invented for paleoclimate reconstructions, and no one else really understands it.

Roger, I presume you have background in multi-variate regression analysis. Correct? Is it even possible to get a doctorate in Political Science without one?

Roger Pielke, Jr. said...

-72-Captian Curt

Yes, basic statistics is required in just about every advanced social sciences degree program, including the one I was in.

In my case it was a bit redundant as I also have a degree in mathematics ;-)

MrUsufruct said...

William (#61),

"Mr U: come on over to wiki, if you think you're hard enough."

Since Roger is still holding out hope that you are capable of behaving with some class, I'll will throttle back on my response, only to note that you appear to be challenging me to a "fight" at a closed discussion board.

Well, I will add one more thing. Perhaps you're referring to another subject site at Wikipedia where you appear to be proving your manhood:

"(note, BTW, how the poor dear septics don't even bother trying to edit the page any more, having been crushed so often by the Mailed Fist)"

http://scienceblogs.com/stoat/2009/01/does_anyone_care_about_theon.php

I wonder if there are any psychologists following these discussions.

Jaye said...

"non-climatic ditches"...heh, heh, that's pretty funny, sad but funny.

andrewt said...

Captain Curt you have me confused with Roger. He was putting the view that professional qualifications are needed to debate such issues. He has since said this was was irony - exactly how escapes me - but amateur discussions of paleoclimate are apparently OK.

Captain Curt said...

#76 andrewt: Are you being deliberately obtuse? Roger introduced this topic by saying:

"One of the challenges for observers of disputes about climate science is that they are often complicated, technical and nearly impossible to resolve without becoming an expert yourself."

This is called the "setup" to the post, the point being that this dispute is not one of those issues. This particular issue is one of basic mathematics, obvious to anyone who has taken (and understood) an undergraduate statistics course. It is an issue of how to interpret results of a multi-variate regression, a type of analysis used in many different fields, and probably more in the social sciences than in climate science. You really don't need to know anything in particular about paleoclimatology to judge this one. And Roger is no "amateur" on the subject of regression analysis -- he certainly understands it better than Mann and Connelly!

AMac said...

(Five weeks after this post went online...)

I have worked through the first of the three Stoat threads on the use of the Lake Korttajarvi (Tiljander) proxies by Mann et al (PNAS, 2008), Oh dear oh dear oh dear oh dear. In this comment, blogger William Connolley derided Roger Pielke Jr. for "not understanding" how Mann et al. used these lakebed sediment records. A read of this abridged and annotated version of the Stoat thread shows why I conclude that RPjr's understanding of this matter was (and is) correct.

AMac said...

Four months after this post went up, I am appending another comment.

Part of the response to the airing of the upside-down use of the uncalibratable Tiljander proxies in Mann et al (PNAS, 2008) has been to say,

"Even if these data sets were used incorrectly, Mann shows that his analysis holds up when they are excluded. Thus, No Correction Needed."

I think this analysis is backwards. Instead, the inadvertent use of the Tiljander proxies constitutes an important test of the paper's methods. A discussion of this topic is here.

Post a Comment

Note: Only a member of this blog may post a comment.