Statisticians Dismember Mann's Hockey Stick

[lawrocket 3]

A week ago, in this post http://www.dropzone.com/cgi-bin/forum/gforum.cgi?post=3922736#3922736 I wrote:

Quote

The Second report (Oxburgh Report) did not find any deliberate scientific malfeasance. However, this report also found that "there would be mutual benefit if there were closer collaboration and interaction between CRU and a much wider scientific group outside the relatively small international circle of temperature specialists."
http://www.uea.ac.uk/...ss/CRUstatements/SAP - p.5

I also wrote:

Quote

Said the Oxburgh Report: "With very noisy data sets a great deal of judgement has to be used. Decisions have to be made on whether to omit pieces of data that appear to be aberrant. These are all matters of experience and judgement. The potential for misleading results arising from selection bias is very great in this area.""

One thing I left out in the Oxburgh Report:

Quote

It is regrettable that so few professional statisticians have been involved in this work because it is fundamentally statistical.

Now, comes news that a couple of statisticians have decided to take a look at Mann's Hockey Stick.
http://www.e-publications.org/ims/submission/index.php/AOAS/user/submissionFile/6695?confirm=63ebfddf

Note: it is to be published in the Annals of Applied Statistics.
Note: one of the authors is at U. Penn - where Mann is also a Professor and where U. Penn cleared Mann of misconduct by basically stating he's an internationally acclaimed scientist who gets a lot of funding.


Read this paper. It's long but not too difficult. I can admit to not having a great understanding of statistics but it's in pretty plain English and their conclusions are quite clearly stated. They specifically attacked the bristlecone proxies.

Quote

“[A] random series that are independent of global temperature are as effective or more effective than the proxies at predicting global annual temperatures in the instrumental period. Again, the proxies are not statistically significant when compared to sophisticated null models.”

random numbers were better correlated with data.

And this one:

Quote

“[T]he proxy record has some ability to predict the final thirty-year block, where temperatures have increased most significantly, better than chance would suggest.”

Now - this last one sounds like, "Wow! The recent cooling we've seen is accurate." What it really implies to me is that the further back one goes, the less predictive it gets. Back 1000 years, there's no idea.

Then this conclusion:

Quote

Climate scientists have greatly underestimated the uncertainty of proxy-based reconstructions and hence have been overconfident in their models.

Checking out their graph (Figure 16), one can see that a flat line fits within the uncertainty. In fact, one can draw a line indicating a downward trend starting at .34 degrees in 1000 and ending at -.25 degrees in 2000.

realclimate isn't discussing it, though Gavin Schmidt stated in a comment:

Quote

The M&W paper will likely take some time to look through (especially since it isn't fully published and the SI does not seem to be available yet), but I'm sure people will indeed be looking. I note that one of their conclusions "If we consider rolling decades, 1997-2006 is the warmest on record; our model gives an 80% chance that it was the warmest in the past thousand years" is completely in line with the analogous IPCC AR4 statement. But this isn't the thread for this, so let's leave discussion for when there is a fuller appreciation for what's been done. - gavin]

Recall my discussion of Schmidt's point - it means that the further back it goes, the less accurate it gets.

Now, to go back on my previous post: "the report pointed to climate scientists like a clique of frat dicks closely guarding their shit against others"

Climate scientists of the alarmist sort have had a well-established history of sticking together, forming their cliques, and not allowing others in on it. They are the climate scientists. RealClimate calls itself "Climate science from climate scientists." I've read the blog pretty consistently for three or four years, and have found that there is the attitude that they define who is qualified and who isn't (as well as Eli Rabbett, etc.)

On the heels of Oxburgh, there are RealStatistics - Statistical science by statistical scientists. They aren't statisticians doing climate science. They are taking on climate scientists doing statistics.

And they have shown, with their peer reviewed and published paper, that Mann's hockey stick went McSorley - it had a bad curve!


Edited to add: My thought is that "real climate scientists" will find themselves hosed by their past. I have frequently mentioned that the techniques that are used by one group will eventually be used against that group.

Here, "real climate scientists" (self-defined) have dismissed others who are not "real climate scientists."
Now, the "real statisticians" may similarly dismiss those who are mere climate scientists.

I go back to the English inquiries, all of which took issue with climate science culture, and one of which stated outright that climate scientists, with their buddy-enemy attitude and cultural unwillingness to cooperate brought this on themselves.

My wife is hotter than your wife.

[rushmc 23]

Thanks for this post
I have said (as a Six Sigma Black Belt) that give me some numbers and tell me what you want it to say, and I can do that.

I have pulled data and built many reports the company I am working for has used. There are two big fears I have every time I start something like this. First, is the report giving a clear indication of what is really going on and second, if I do the first thing or not, will the information (ie report) be used correctly or properly.

It turns into a really big be careful what you ask for, cause you might just get it

Thanks again

"America will never be destroyed from the outside,
if we falter and lose our freedoms,
it will be because we destroyed ourselves."
Abraham Lincoln

[quade 4]

Quote

I have said (as a Six Sigma Black Belt) that give me some numbers and tell me what you want it to say, and I can do that.

So, what you're saying is if you believe the original analysis of the numbers is bullshit . . . this new analysis is just as likely to be bullshit.

Looks like we need another independent analysis . . . or could that also be bullshit? Oh dear. then whoever shall we believe; the people who have no vested interest in saying global warming is happening or people hired by companies trying to disprove it?

My money is on the first and never with the latter.

quade -
The World's Most Boring Skydiver

[lawrocket 3]

Quote

So, what you're saying is if you believe the original analysis of the numbers is bullshit . . . this new analysis is just as likely to be bullshit.

Interestingly - that's exactly what they say, Paul. They aren't coming up with new hypotheses. They aren't saying that there are better proxies. They aren't coming up with new anything.

All they are saying is that the hockey stick graph is statistical bs. They aren't climate scientists. They're statisticians who analyzed the statistics (which is what these models are about).

Had you read the paper you'd have known that. Sometimes it's really neat to see when sarcasm is directly on point. Mann's graph cannot be shown to be more accurate than random data and has little confidence.

My wife is hotter than your wife.

[grimmie 186]

McSorley's hockey stick.

That memory is a kick to the head.


Now back to science...

[lawrocket 3]

Quote

McSorley's hockey stick.

That memory is a kick to the head.

Still a source of bitterness to me, too.

Michael Mann
Marty McSorley.

MM & MM.
Coincidence?

My wife is hotter than your wife.

[quade 4]

I did read it, specifically these parts;

Quote

Our paper is an effort to apply some modern statistical methods to these prob- lems. While our results agree with the climate scientists findings in some respects, our methods of estimating model uncertainty and accuracy are in sharp disagreement.

and

Quote

Nevertheless, the temperatures of the last few decades have been relatively warm compared to many of the thousand year temperature curves sampled from the posterior distribution of our model.

In other words, while they can't support the specific graphing of data based on forensics going as far back as 1000 AD, they do agree that starting in the 1900s there is a sharp upturn of temperatures.

Both the original "hockey stick" graph and theirs correlate quite well when looked at in shorter time periods.

quade -
The World's Most Boring Skydiver

[turtlespeed 226]

Quote

I did read it, specifically these parts;

Quote

Our paper is an effort to apply some modern statistical methods to these prob- lems. While our results agree with the climate scientists findings in some respects, our methods of estimating model uncertainty and accuracy are in sharp disagreement.

and

Quote

Nevertheless, the temperatures of the last few decades have been relatively warm compared to many of the thousand year temperature curves sampled from the posterior distribution of our model.

In other words, while they can't support the specific graphing of data based on forensics going as far back as 1000 AD, they do agree that starting in the 1900s there is a sharp upturn of temperatures.

Both the original "hockey stick" graph and theirs correlate quite well when looked at in shorter time periods.

. . . until when?

I'm not usually into the whole 3-way thing, but you got me a little excited with that. - Skymama
BTR #1 / OTB^5 Official #2 / Hellfish #408 / VSCR #108/Tortuga/Orfun

[quade 4]

See for yourself on page 36 of the article linked.

quade -
The World's Most Boring Skydiver

[rushmc 23]

Quote

Quote

I have said (as a Six Sigma Black Belt) that give me some numbers and tell me what you want it to say, and I can do that.

So, what you're saying is if you believe the original analysis of the numbers is bullshit . . . this new analysis is just as likely to be bullshit.

Looks like we need another independent analysis . . . or could that also be bullshit? Oh dear. then whoever shall we believe; the people who have no vested interest in saying global warming is happening or people hired by companies trying to disprove it?

My money is on the first and never with the latter.

I am not saying anything but you have to assume to make a stupid point I guess.

As for your money? That is your choice. And of course you have to believe that because you have your world view in danger if you dont (and all libs are sooooooo much smarter than conservative too I guess)

The bias is only 1 directional in your worlld. If you agree it is good. If you dont it is biased

Glad so few of you live in that world, but thier seems to be a statistically high number of your kind here

"America will never be destroyed from the outside,
if we falter and lose our freedoms,
it will be because we destroyed ourselves."
Abraham Lincoln

[kallend 2,184]

Quote

The bias is only 1 directional in your worlld. If you agree it is good. If you dont it is biased

Funny coming from someone who believes FOX is unbiased, when their parent company gave $1Million to the GOP and zero to the Dems,

...

The only sure way to survive a canopy collision is not to have one.

[rushmc 23]

Quote

Quote

The bias is only 1 directional in your worlld. If you agree it is good. If you dont it is biased

Funny coming from someone who believes FOX is unbiased, when their parent company gave $1Million to the GOP and zero to the Dems,

Really? I have said that? Show me where please.

"America will never be destroyed from the outside,
if we falter and lose our freedoms,
it will be because we destroyed ourselves."
Abraham Lincoln

[lawrocket 3]

Quote

Our paper is an effort to apply some modern statistical methods to these prob- lems. While our results agree with the climate scientists findings in some respects, our methods of estimating model uncertainty and accuracy are in sharp disagreement.

Yeah. One can certainly find some downward trend in the reconstruction. And an upward trend. And a flat trend. When there's agreement with something that can't be trusted then there's a problem.

Back to it - it's not saying that they've proven that the hockey stick isn't real. just like Verbal Kint's story was actually told. The problem is that the underlying data don't support the confidence in blaming Keaton because it's not as certain as represented.

The 95 percent confidence? The 99 percent confidence? There's low low confidence in that.

[Reply]

Quote

Nevertheless, the temperatures of the last few decades have been relatively warm compared to many of the thousand year temperature curves sampled from the posterior distribution of our model.

In other words, while they can't support the specific graphing of data based on forensics going as far back as 1000 AD, they do agree that starting in the 1900s there is a sharp upturn of temperatures.

[kallend 2,184]

Quote

Quote

Quote

The bias is only 1 directional in your worlld. If you agree it is good. If you dont it is biased

Funny coming from someone who believes FOX is unbiased, when their parent company gave $1Million to the GOP and zero to the Dems,

Really? I have said that? Show me where please.

OK so you admit that FOX is biased. Small steps.

...

The only sure way to survive a canopy collision is not to have one.

[rushmc 23]

Quote

Quote

Quote

Quote

The bias is only 1 directional in your worlld. If you agree it is good. If you dont it is biased

Funny coming from someone who believes FOX is unbiased, when their parent company gave $1Million to the GOP and zero to the Dems,

Really? I have said that? Show me where please.

OK so you admit that FOX is biased. Small steps.

Really, and where did I admit that?

"America will never be destroyed from the outside,
if we falter and lose our freedoms,
it will be because we destroyed ourselves."
Abraham Lincoln

[kallend 2,184]

Make up your mind what you believe.

Take a look at your post Mar 2, 2006, 6:22 PM for starters.

Then Nov 25, 2007, 6:36 AM

And especially Nov 2, 2007, 4:15 PM

I think you forget what you post.

...

The only sure way to survive a canopy collision is not to have one.

[skiskyrock 0]

Quote

Note: it is to be published in the Annals of Applied Statistics.
Note: one of the authors is at U. Penn - where Mann is also a Professor and where U. Penn cleared Mann of misconduct by basically stating he's an internationally acclaimed scientist who gets a lot of funding.

Mann is professor at Penn State, The author is at the University of Pennsylvania. Different schools.

[lawrocket 3]

You are correct. Thanks for pointing out my error.

Yeah, I make mistakes...

My wife is hotter than your wife.

[rushmc 23]

Quote

Make up your mind what you believe.

Take a look at your post Mar 2, 2006, 6:22 PM for starters.

Then Nov 25, 2007, 6:36 AM

And especially Nov 2, 2007, 4:15 PM

I think you forget what you post.

I think you need to provide those exact posts sir

"America will never be destroyed from the outside,
if we falter and lose our freedoms,
it will be because we destroyed ourselves."
Abraham Lincoln

[kallend 2,184]

Quote

Quote

Make up your mind what you believe.

Take a look at your post Mar 2, 2006, 6:22 PM for starters.

Then Nov 25, 2007, 6:36 AM

And especially Nov 2, 2007, 4:15 PM

I think you forget what you post.

I think you need to provide those exact posts sir

I've given you site, date and time to the nearest minute.

...

The only sure way to survive a canopy collision is not to have one.

[rushmc 23]

Quote

Quote

Quote

Make up your mind what you believe.

Take a look at your post Mar 2, 2006, 6:22 PM for starters.

Then Nov 25, 2007, 6:36 AM

And especially Nov 2, 2007, 4:15 PM

I think you forget what you post.

I think you need to provide those exact posts sir

I've given you site, date and time to the nearest minute.

Yep, that is all you did

"America will never be destroyed from the outside,
if we falter and lose our freedoms,
it will be because we destroyed ourselves."
Abraham Lincoln

[kallend 2,184]

Quote

Quote

Quote

Quote

Make up your mind what you believe.

Take a look at your post Mar 2, 2006, 6:22 PM for starters.

Then Nov 25, 2007, 6:36 AM

And especially Nov 2, 2007, 4:15 PM

I think you forget what you post.

I think you need to provide those exact posts sir

I've given you site, date and time to the nearest minute.

Yep, that is all you did

Don't like to be reminded of when you posted garbage, eh?

...

The only sure way to survive a canopy collision is not to have one.

[rushmc 23]

Quote

Quote

Quote

Quote

Quote

Make up your mind what you believe.

Take a look at your post Mar 2, 2006, 6:22 PM for starters.

Then Nov 25, 2007, 6:36 AM

And especially Nov 2, 2007, 4:15 PM

I think you forget what you post.

I think you need to provide those exact posts sir

I've given you site, date and time to the nearest minute.

Yep, that is all you did

Don't like to be reminded of what you posted and when, eh?

"America will never be destroyed from the outside,
if we falter and lose our freedoms,
it will be because we destroyed ourselves."
Abraham Lincoln

[lawrocket 3]

Okay. I'll join in. Haven't you two had enough of the flame war?

My wife is hotter than your wife.

[rushmc 23]

Quote

Okay. I'll join in. Haven't you two had enough of the flame war?

He wanted to go off topic and I let him, my bad
I did a search however of all posts by me. First off, I cant get back to 02 (why I dont know) and I cant find any posts with those specific dates and times (not saying they arent there but I wonder what his game really is)

Back to the topic however, I stand by my point (that quade butchered) stating that I can make stats (givin a data set) say almost anything you want it to. As you have pointed out, biases come into the equation when pulling data and trying to see what they tell you in reality.

So, in the end, the problem is finding what the data realy says with a high confidence level. That is the hard and most frustrating part

"America will never be destroyed from the outside,
if we falter and lose our freedoms,
it will be because we destroyed ourselves."
Abraham Lincoln