Make your own Michael Mann hockey stick at home

Following yesterday’s first look at emails from the 2 year effort to get NOAA to release emails from FOIA requests, where we learn that some scientists felt ‘Hit on the head with a hockey stick’ and that “The paleodata always got a lot more attention from the general public than it deserved.”.

There’s a great article today at SPPI today Michael Mann — the ghost of climate past which summarizes the whole hockey stick affair quite well. It draws heavily on a post from May of this year on PJ Media by Rand Simberg titled The Death of the Hockey Stick?

In that post I found this little nugget below from December 2009, one that I apparently missed in the furor immediately following Climategate1. I’m now correcting that oversight.

What is it? It’s the hockey stick recreated in Excel, using proxy data and instrumental data freely available on the web,  something you can easily do at home. I figure the more people that explore this themselves using the easy to follow steps, the more people will understand what a statistical abuse the hockey stick is. The source of this tutorial is as surprising to me as it may be to you, and there’s no special software or secret Mannian Excel plugins needed to do the work yourself. I’ll let Rand Simberg explain.

From his May 2012 essay:

Ultimately, in addition to Mann’s claim for the dramatic recent uptick (which we are supposed to presume was a result of the late industrial revolution and equally dramatic increase in carbon dioxide into the atmosphere as a result of the liberation of carbon from burning long-buried fossil fuels), Keith Briffa of the Climate Research Unit (CRU) at the University of East Anglia in England controversially declared, based on Eurasian data, that the well-documented Medieval Warm Period (MWP), from around 950 to 1250 CE — the European Middle Ages — didn’t actually exist.

This claim was important, if not essential, to Mann’s thesis, because his initial formulation only went back to 1400, the beginning of the so-called Little Ice Age. Critics of the theory thus argued immediately upon its presentation that it shouldn’t be surprising that the earth was warming now, given that we are still coming out of it, and that the medieval warming in the absence of late Carolingian SUVs and coal plants argued that the climate naturally cycled, with no need to invoke Demon Carbon. That is to say, to the degree that the hockey stick has a blade in the twentieth century, it would have another a millennium ago.

The theory has continued to take blows over the years since it was first presented. About a decade ago, a paper was published by Willie Soon and Sallie Baliunis claiming that there was good evidence that both the (still extant) MWP and current warming were driven by solar activity rather than carbon emissions. But these initial attacks were beaten back by the climate mafia (as we now know from the leaked emails between Mann and his partners in crime in East Anglia from two and a half years ago). The real damage came when a retired Canadian mining engineer, Steve McIntyre, and a professor at the University of Guelph, Ross McKitrick, started digging into Mann’s methodology, and found flaws in both his statistical analysis and data interpretation, and published a paper describing them in Geophysical Research Letters in 2005. They showed that Mann’s methodology would generate a hockey stick almost independently of the data input, by feeding it spectral noise. Later, Internet satirist (and apparent statistician by day) Iowahawk provided a primer on how to create a hockey stick at home, using a standard spreadsheet program.

Yep, Iowahawk.

I liked this part of Iowahawk’s “how to” primer the best:

Is there anything wrong with this methodology? Not in principle. In fact there’s a lot to recommend it. There’s a strong reason to believe that high resolution proxy variables like tree rings and ice core o-18 are related to temperature. At the very least it’s a more mathematically rigorous approach than the earlier methods for climate reconstruction, which is probably why the hockey stick / AGW conclusion received a lot of endorsements from academic High Society (including the American Statistical Association).

The devil, as they say is in the details. In each of the steps there is some leeway for, shall we say, intervention. The early criticisms of Mann et al.’s analyses were confined to relatively minor points about the presence of autocorrelated errors, linear specification, etc.  But a funny thing happened on the way to Copenhagen: a couple of Canadian researchers, McIntyre and McKitrick, found that when they ran simulations of “red noise” random principal components data into Mann’s reconstruction model, 99% of the time it produced the same hockey stick pattern. They attributed this to Mann’s method / time frame for selecting of principal components.

To illustrate the nature of that debate through the spreadsheet, try some of the following tests:

  • Run step 3 through step 7, but only use the proxy data up through 1960 instead of 1980.
  • Run step 5 through step 7, but only include the first 2 principal components in the regression.
  • Run step 3 through step 7, but delete the ice core data from the proxy set.
  • Run step 2 through step 7, but pick out a different proxy data set from NOAA.

Or combinations thereof. What you’ll find is that contrary to Mann’s assertion that the hockey stick is “robust,” you’ll find that the reconstructions tend to be sensitive to the data selection. M&M found, for example, that temperature reconstructions for the 1400s were higher or lower than today, depending on whether bristlecone pine tree rings were included in the proxies.

What the leaked emails reveal, among other things, is some of that bit of principal component sausage making. But more disturbing, they reveal that the actual data going into the reconstruction model — the instrumental temperature data and the proxy variables themselves — were rife for manipulation. In the laughable euphemism of Philip Jones, “value added homogenized data.” The data I provided here was the real, value added global temperature and proxy data, because Phil told me so.  Trust me!

I urge  readers to replicate it yourselves. Knowledge is power, especially first hand knowledge. Here’s all you need to do it. I’ll be happy to publish what you learn.


newest oldest most voted
Notify of
Eugene S Conlin

Anthony, the link to “Michael Mann — the ghost of climate past doesn’t work” – it should be


Anthony, I’ve always wondered, why isn’t there a link for the SPPI site on your list of skeptic blogs? I often use WUWT as a base or springboard to visit other skeptic sites, but SPPI is not there.

Ed Scott

Get Lost
By Rich Lowry
August 22, 2012 1:15 P.M.
So, as you might have heard, Michael Mann of Climategate infamy is threatening to sue us.
Mann is upset — very, very upset — with this Mark Steyn Corner post, which had the temerity to call Mann’s hockey stick “fraudulent.” The Steyn post was mild compared with other things that have been said about the notorious hockey stick, and, in fact, it fell considerably short of an item about Mann published elsewhere that Steyn quoted in his post.
So why threaten to sue us? I rather suspect it is because the Steyn post was savagely witty and stung poor Michael.
Possessing not an ounce of Steyn’s wit or eloquence, poor Michael didn’t try to engage him in a debate. He sent a laughably threatening letter and proceeded to write pathetically lame chest-thumping posts on his Facebook page. (Is it too much to ask that world-renowned climate scientists spend less time on Facebook?)
All of this is transparent nonsense, as our letter of response outlines.
In common polemical usage, “fraudulent” doesn’t mean honest-to-goodness criminal fraud. It means intellectually bogus and wrong. I consider Mann’s prospective lawsuit fraudulent. Uh-oh. I guess he now has another reason to sue us.
Usually, you don’t welcome a nuisance lawsuit, because it’s a nuisance. It consumes time. It costs money. But this is a different matter in light of one word: discovery.
If Mann sues us, the materials we will need to mount a full defense will be extremely wide-ranging. So if he files a complaint, we will be doing more than fighting a nuisance lawsuit; we will be embarking on a journalistic project of great interest to us and our readers.
And this is where you come in. If Mann goes through with it, we’re probably going to call on you to help fund our legal fight and our investigation of Mann through discovery. If it gets that far, we may eventually even want to hire a dedicated reporter to comb through the materials and regularly post stories on Mann.
My advice to poor Michael is to go away and bother someone else. If he doesn’t have the good sense to do that, we look forward to teaching him a thing or two about the law and about how free debate works in a free country.
He’s going to go to great trouble and expense to embark on a losing cause that will expose more of his methods and maneuverings to the world. In short, he risks making an ass of himself. But that hasn’t stopped him before.
— Rich Lowry is the editor of National Review.

Andrew W

But, but, where’s the corrected graph, you know, the one that uses all the correct methodology to not produce a hockey stick? Oh, it’s still a hockey stick?

Reblogged this on Climate Ponderings and commented:
My Favorite Defense of Mr Mann
“Mann’s work has been proven time and time again to not be fraudulent because it has been reproduced by others.”.
My Usual reply:
Why, Yes! It has been reproduced…BUT the question is – Why would you want to reproduce it?
Ha ha ha ha 🙂


Dear Anthony,
Some advertisement is causing your webpage’s main page not to load correctly at least in firefox. Only the first topic is presented, then you can read “ADVERTISEMENT” and then nothing else loads. Happened already 3 times to me.


Ed Scott says:
August 23, 2012 at 11:20 am
“Get Lost
By Rich Lowry”
It’ impolite not to link.

Andrew W,
But… but… Here is the corrected Mann chart. See? There is nothing unusual about current temps.
Here is another chart.
And here is a comparison of results using data that Mann threw out. Where’d Mann’s Hokey Stick go?
And here is the data that Mann hid in an ftp file labeled “censored“.
Here is Mann’s email from Tim Osborne.
And here is Prof Wegman’s comparison in Wegman’s testimony before Congress.
Here, Mann erases the MWP in order to make his hockey stick scarier.
And of course, the late, great John Daly totally deconstructs Mann’s chart here.
Which brings us to the present day.
Now tell the truth, Andrew. Would you still buy snake oil from that Mann?


It appears that Mann’s definition of “robust” is not robust.

Andrew W

Thanks smokey, your first link gives no source, your second link gives no source, your third link is broken, your forth link gives no source, your fifth link means nothing, your sixth link doesn’t actually show climate date at all, it’s just a random noise graph Wegman used to illustrate his argument, your seventh link shows no source, your eighth link is to large to address in its entirely, some bits are out dated others meaningless, your ninth link is I guess some sort of popularity contest, so meaningless with respect to the science.

Andrew W,
May I translate your comment? Thank you:
U R welcome. ☺

george e. smith

Well what I want to know is whether I can make myself a Mann / McIntyre / McKitrick / Mouse / Iowahawk / whatever, Hockey stick from the numbers in the Mann Hattan Telephone book; mebbe that’s the Mad Hatter Telephone book; well either one.
I happen to be among the cognoscenti who are very familiar with the Sally Baliunas / Willy Wei Hock Soon paper referenced above. Read it as Soon as Willy and Sally published it.
What say you Iowahawk ? Your Excel spread sheet robust enough to Hockey Stickefy a Telephone directory ??

Luther Wu

Andrew W,
There’s nothing quite like watching someone flail about in defense of Mann’s hockey stick.
This has been an otherwise humorless afternoon, so many thanks for your efforts.

Paul Penrose

You presume that there is a way to extract temperature data from these proxies and then send readers on a wild goose chase. The fact is, it does not really matter for this discussion. Mann’s methodology has been shown to be flawed given it’s sensitivity to minor input data differences. It’s not incumbent on me to show a “correct” methodology, even if one exists. Mann’s is still wrong and his conclusions worthless. You can’t change this with logical fallacies.


It’s all in how you SAY it…. roBUST! Emphasis on BUST!

Svend Ferdinandsen

I wonder a bit why it should be a problem that the proxies are correlated. In my opinien they should be if they really are good proxies for temperature and then some sort of average would do better.
I have played with SVD on the internet to learn more of what happens.
It takes a little work to format dataseries so that you can copy them into the calculator, but then it works fine. I doubt the results of such treatments.
If you have two series that goes more or less opposite, the first component will resemble the variation of them, the next component will be very small and only have the “noise”, but the first component only shows how they variate, not that they more or less cancel each other if averaged.
The crucial point is to select the number of components and their weights, and in there is the devil in all his might.


Andrew W says:
August 23, 2012 at 12:38 pm

Notice I didn’t include anything Andrew W says because this whole CAGWCF* issue hasn’t been about science from the beginning–so why reference other sources that aren’t about science?
Come clean, Andrew–where is YOUR SCIENCE?
But to the bigger issue, the question remains: Is Mann a criminal or a totally inept scientist?
My answer (and you all chime in now, you hear?) is that he’s both: he’s an inept scientist that has committed criminal behavior to cover his ineptness. And that makes him the worst type of “scientist” possible.
Sorta like a lot of other “climsci” people out there slopped on the trough of taxpayer funding.
*CAGWCF: Catastrophic Anthropogenic Global Warming Control Freaks, aka “watermelons”.

cui bono

Another brilliant post.
Btw, Lucia had a whole series of posts recently creating hockey sticks aplenty out of sticky-back plastic (UK joke) and statistical quirks. It turned out to be so easy Josh cartooned the process:


I urge readers to read what the National Academy of Sciences said. And to read the numerous other studies in scientific journals which found the same relationship with temperature.

See! It’s replicable! What’s not to like?


“using proxy data and instrumental data freely available on the web, something you can easily do at home”
Available where? Which of it has been jiggered in which ways? After following the web log for a while, it’s all a very confusing mess, from my POV. Perhaps if I wanted to set aside all other interests.


Iowahawk is hilarious. OT, but here’s a great blog entry of his recounting a tale of the Reverend John St. Edwards, Lord Plaintiff of Durham, and the Reverend Albert des Gores II, Earl Carbonet of Greenhouse…

Ally E.

I think Mann was playing with words. Clearly “robust” means you get a hockey stick no matter what data you put in.


Schroedinger says:
August 23, 2012 at 1:30 pm

I urge readers to read what the National Academy of Sciences said. And to read the numerous other studies in scientific journals which found the same relationship with temperature.

I’m woundering–did these “other studies in scientific journals” also use equivalent Mannianesque procedures? (The NAS defends Mann and his tactics, don’t they?)
Multiple wrongs (regardless of how many) don’t make a single right.

Mark T

Svend Ferdinandsen says:
August 23, 2012 at 1:07 pm

I wonder a bit why it should be a problem that the proxies are correlated. In my opinien they should be if they really are good proxies for temperature and then some sort of average would do better.

It is not a problem for the proxies to be correlated. Indeed, if they are all linear combinations of the inputs (to tree growth), they should be correlated. What can’t be correlated (for PCA to work properly) are the inputs, i.e., the source signals driving the proxies, e.g., ring width, that PCA is attempting to extract. The components that result from any method (principal components, eigenvectors, etc.) will be uncorrelated by definition. If the inputs are correlated, then clearly any uncorrelated components that were extracted are not representative of the inputs.
Schroedinger says:
August 23, 2012 at 1:30 pm

I urge readers to read what the National Academy of Sciences said.

Many of us have. Besides the fact that they were hardly providing a glowing endorsement, none of those that “said” the NAS position really dug into any of the details.

And to read the numerous other studies in scientific journals which found the same relationship with temperature.

Using the same data and same basic methods*, of course. That’s the problem. Take a subset of the data and the results vary widely. In general terms, this means either the data, the methods, or both, are ill-suited for the purpose for which they are provided.
* If you do not understand what I mean by the “same basic methods,” then you are not capable of understanding any of the criticisms anyway.


Thanks smokey, your first link crashed my computer, your second link sent me to a Russian bride dating site, your third link is only works with Netscape, your forth link gives no destination, your fifth link means nothing to me, your sixth link doesn’t actually show a number to call for a Russian date at all, it’s just a photo of random scribbles from a outhouse wall that Mann used to illustrate his argument, your seventh link shows no sauce only the meat, your eighth link is too large to cut-and-paste in its entirely, some bits are out for a date, your ninth link is I guess some sort of popularity contest, I have to go now Mom is calling me for bed-time.

Mark T

I said:

If the inputs are correlated, then clearly any uncorrelated components that were extracted are not representative of the inputs.

I should have noted that they may be representative of the inputs in such a case, but there would be no way of knowing how. For example, given two inputs that are correlated, x1′ and x2 where x1′ = x1 + f(x2) and x1 and x2 are the uncorrelated pieces, PCA may find x1 and x2, but would be unable to determine x1′ without further information. Using the results of PCA to generate the function f() brings to mind the concept of a self-licking ice cream cone.

Mark T

Funny, I clicked on Smokey’s first four links and they all worked fine. Not sure what your problem is, Andrew30, other than perhaps you really do need to get in for bed ’cause mommy’s calling.

Mark T,
I think Andrew30 was making fun of Andrew W’s complaints about every chart I provided.
[That’s why Anthony asks commenters to put something like “/sarc” after comments like Andrew30’s.]
And you’re right, all the links work. Andrew W = PEBCAK.

Mark T

Oh, if so, then apologies. Yeah, /sarc would have been helpful. 😉


I never had the chance to have a look over these numbers and methodology. So I’ve been playing with the spreadsheet offered by iowahawk.
I see several problems with the methodology that leads to the famous ‘stick’ and the AGW:
1. I don’t understand the advantages of doing a PC analysis and then retaining ALL the factors. This is the same as working with some kind of rescaled original data. By the way, the result of the PC analysis is VERY poor (little variance explained by the main factors). I haven’t tried it, but I’m sure that the full study can be reproduced without the PC analysis. PC analysis are used to extract the main variance source and to discard the ‘noise’ … the name PRINCIPAL refers to the fact that usually you’ll stand with the PRINCIPAL factors and you’ll throw away the rest. But if you have 15 variables and use the 15 factors … where is the advantage ?
2. What the PC analysis says is that the proxy data is very noisy, is so noisy that it can be used to extract from it whatever pattern you can imagine. It’s like a bunch of clouds, everyone can see whatever figure he wants by looking at it. What McIntyre found is obvious to people with pretty good knowledge of Multivariate Analysis.
3. What the linear regression does is to take the high variability of the proxy data to reproduce the increasing trend in the 19 and 20 century. With 15 variables you can do this using almost whatever noise source. If you only had 2 or 3 variables … it would be a very different problem.
Once extracted the ‘right variability’ to reproduce the recent temperatures data … what is left for the rest of the series from 1400 is … noise, the exact same noise that the proxy data is.
So … yes, the stick is constructed. The whole analysis is useless. The guys that did it have absolutely no idea about statistics.


FWIW, that isn’t excel in the picture. It’s Openoffice Calc.


Follow up:
I’ve computed correlation coefficients between temperature and the 15 factors from the PC analysis. The highest correlation is 0.45 for factor 2. This is not high enough. Temperature variation is not a significant source of variability for the proxy dataset.
From the point of view of temperature variation, this proxy dataset is just noise. It should not be used in any way to try to explain the temperature variation.

Steve C

All Smokey’s links work fine for me, too. Including PEBKAC. If you have problems, they’re yours. (Oh, and thanx for the Farage one you referred me to a week or so ago, Smokey. That worked fine, too.)


Don’t forget the upside-down Tiljander over and over…..

Gunga Din

ChE says:
August 23, 2012 at 4:03 pm
FWIW, that isn’t excel in the picture. It’s Openoffice Calc.
If you follow the link IowaHawk says he used Openoffice but gave instructions and a link to do the same with Excel. (I think the link was to a macro Excel would need.)


george e. smith:
Apparently you’re unaware that the methodology hasn’t been shown to make hockey sticks out of white noise (random data) but out of red noise (Drunkard’s Walk). Unless you’d like to make the case that numbers in a phone book are red noise, your proposed test is like trying to find out if squeezing tomatoes will get you orange juice.
Instead of proposing someone else do such-and-such a test, though, why don’t you find some sources of red noise, and put them through this exercise, and let us know your results? That’d be a useful contribution.

Gee, Andrew30’s Mom really was calling him for bed. When Andrew30’s Mom tucks him in, she’ll have to remind him again about why he shouldn’t be viewing some of those unsavory sites on the web. Smokey’s links work fine for me. I particularly like some of the pre-climategate links. Very interesting when you consider what has transpired since those astonishing days.

Mark T

RDCII: maybe you should investigate the terms white noise, red noise, and random data. Your assumptions regarding these concepts are clearly incorrect.

Sorry, I don’t know my Andrus’s from a hole in the ground.

jonathan frodsham

Rich Lowry: You are now a hero of mine. If you wish to set up a fund to fight Mann just let me know through WUWT or the National Review. I am sure you will have no trouble getting a “Fight Mann Fund” togethe;, yes sir the Mann’s of this world need to taught a lesson in freedom. Well done Rich!

Thank you very much Anthony.

John Blake

Thanks to Iowahawk for an illuminating basic treatise, and to AW for posting this in its entirety. Valid technique is critical, but so of course are comprehensive, real inputs. As Lewis Carroll’s Mad Hatter says to the March Hare at their tea party in Wonderland, however nicely done oiling your watch with butter, dipping it in tea, does not improve performance: “I told you that butter wouldn’t suit the works!”
“But it was the best butter,” the March Hare replied.
In concocting Hockey Sticks, Briffa, Hansen, Jones, Mann, Trenberth et al. “use only the best butter.”

Lee Wells

Let’s who should I believe? NASA or a guy that runs a website for a living? NASA has very good numbers on climate change in the last century
They figured out space flight, launched how many satellites? Made the Space Shuttle & piloted it. Should I believe them or some guy with solar cells on his roof? hmmmmmm……..

Lee Wells says:
“Let’s who should I believe? NASA or a guy that runs a website for a living?”
You probably don’t know that Anthony Watts is a businessman and a peer reviewed, published author on climate studies. He does not “run a website for a living”. Now you know.
“Should I believe [NASA] or some guy with solar cells on his roof? hmmmmmm……”
NASA used to be a great organization, before it decided its top priority was ‘Muslim Outreach’, and before Hansen hijacked GISS for crazy CAGW propaganda. I think I’ll believe the guy who put solar cells on his roof, over a know-nothing.

Lee Wells says:
August 24, 2012 at 11:32 am
Note the climatology in your reference. Then explain what, to your mind, problem exists?
Please remove your head from your arse before attempting the removal of your feet from your mouth.

Lee Wells, there is an organization called NASA JPL located in Pasadena, California. NASA JPL puts humans on the moon and rovers on Mars. These are the “right stuff” people.
There is another organization called NASA GISS located in New York, New York. NASA GISS collects more geophysical data than the rest of the world put together. Some GISS people analyze this data and attempt to understand how this planet works. They also have a (gasp) website so the logical deduction is that, like Anthony Watts, most of their work must be the “wrong stuff.”

Gunga Din

Lee Wells says:
August 24, 2012 at 11:32 am
Let’s who should I believe? NASA or a guy that runs a website for a living? NASA has very good numbers on climate change in the last century
They figured out space flight, launched how many satellites? Made the Space Shuttle & piloted it. Should I believe them or some guy with solar cells on his roof? hmmmmmm……..
I thought you guys said solar power was a good thing?
As far as the space guys looking at surface temperatures, as Mike Mellor says: August 24, 2012 at 1:26 pm, just pointed out, the “rocket scientist” aren’t the same ones who have their heads in the clouds and somewhere else that never ackowleges or even sees the Sun shine. (ie. Hansen)
If you want to see how “robust” the surface temperature record really is, I suggest you read, and seriously consider, Watts et al. 2012 Draft Paper. Conclusions are only as the data they are based on.

Gunga Din

AHHHH! I did it again.
“Conclusions are only as the data they are based on.”
Should be, “Conclusions are only AS VALID as the data they are based on.”
(Good thing I’m not one of the “rocket scientist”. 😎

Tim Clark