Josh writes:
Over at Climate Audit, Steve has been looking at the resurfacing of some Hockey Stick science papers, and the fun continues here and here.
Josh writes:
Over at Climate Audit, Steve has been looking at the resurfacing of some Hockey Stick science papers, and the fun continues here and here.
Yes he did. But he in no way said one was the cause of the other. P@ur momisugly Dolan’s characterization is completely false.
Puh-lease. Surely you can recognize satire?
pat says:
April 21, 2013 at 2:06 pm
more work for Steve when this comes out today. we already have Overpeck on ABC Australia radio this morning saying “deniers” aren’t scientists, they work for “special interests” – no link as yet:
22 April: SMH: Tom Arup: 20th century ‘hottest in 1400 years’e
How does that fit with this record?
http://jonova.s3.amazonaws.com/graphs/lappi/gisp-last-10000-new.png
Jeff Alberts @ur momisugly 3:50 pm says…
cn
“Yes he did. But he in no way said one was the cause of the other. P@ur momisugly Dolan’s characterization is completely false.”
Man. Are the warmistas getting desperate. Anybody who Boston Marathon Bombings and climate change in the same sentence are DEFINITELY blaming climat change. Go spin you’re “progressive” cognitive disonance over at SkS or Huffpo.
Sheesh. The stupid,it burns.
David Jones says,
“How does that fit with this record?
http://jonova.s3.amazonaws.com/graphs/lappi/gisp-last-10000-new.png”
In two ways:
1. That is a graph of central Greenland temperature estimates, not global
2. The X axis is mistakenly labeled “Years before present (2000 AD)”
Actually, “present” for the GISP2 reconstruction (as is common with ice cores) is defined as 1950, so the most recent date in the graph is actually 1950-95 = 1855. That is, the end point which looks so cold in this graph really is back in the Little Ice Age. It comes nowhere near the real present.
Steve McIntyre comments at climateaudit:
“So they’re not just screening by correlation, but going full MBH and weighting by correlation. My, my. Since they’re using Neukom’s code, I presume that’s what they did in the Gergis zombie as well.”
—————————————————
And disturbingly, this is published in German owned Nature again.
I wonder if they would not be liable and suable in a way similar to the tobacco industry for costly expenses possibly fed or promoted by their product ?
Manfred says:
“Steve McIntyre comments at climateaudit:
“So they’re not just screening by correlation, but going full MBH and weighting by correlation.””
So it is similar to principal components analysis?
I think that this is the money statement from the posting at REAL CLIMATE
“Cool 30-year periods between the years 830 and 1910 CE were particularly pronounced during times of weak solar activity and strong tropical volcanic eruptions and especially if both phenomena often occurred simultaneously.”
BA says:
April 21, 2013 at 4:37 pm
Manfred says:
“Steve McIntyre comments at climateaudit:
“So they’re not just screening by correlation, but going full MBH and weighting by correlation.””
So it is similar to principal components analysis?
———————————————————————-
Lucia explained it in cartoon form:
http://rankexploits.com/musings/2012/screening-bias-cartoon-form/
You could try that out and generate a hockey stock from the numbers of your telephone book, like by screening for high values on the last digit.
You may, of course, insert principal component analysis to do the weighted averages. That certainly makes it all appear more sophisticated in some circles.
Manfred says:
“Lucia explained it in cartoon form:”
No, that cartoon does not explain anything. Perhaps you could try in your own words? Why should researchers not use some function of the correlations to screen, and to weight, in building a climate proxy? That sounds similar to principal components, which is used for many things–even by Steve McIntyre.
Any set of measurements, say from tree rings or ice cores, might contain a regional climate signal–but it will be mixed in with variable amounts of noise, from microclimate, buffalo leaned on that tree, etc. How could you detect any regional climate signal without looking for patterns of correlation across multiple proxies? And if some proxies don’t correlate with anything, conclude they vary for non-climate reasons?
“You may, of course, insert principal component analysis to do the weighted averages.”
The weights in principal component analysis are a function of the correlation (or covariance) matrix. So, are the practical results going to be different? I don’t see the gotcha here.
Here is the link to Overpeck interview on Oz ABC RN Breakfast:
http://www.abc.net.au/radionational/programs/breakfast/global-study-reconstructs-temperatures-over-2000-years/4642898
Some challenges dismissed as non-scientist d___ers working for special interests.
@ur momisugly Manfred
“And disturbingly, this is published in German owned Nature again”
what’s wrong with that? are Mercedes and Audi not worldwide renowned cars, even while they are German?
BA:
At April 21, 2013 at 8:40 pm you say
Well, if you don’t like the Lucia’s excellent cartoon then perhaps Lucia’s “own words” will help you to understand. You can read them at
http://rankexploits.com/musings/2009/tricking-yourself-into-cherry-picking/
She wrote that before providing the cartoon which is at
http://rankexploits.com/musings/2012/screening-bias-cartoon-form/
It seemed that people lacking mathematical understanding needed something different, so she provided her informative cartoon. And if you had tried to learn from the cartoon then you would have found the link to Lucia’s “own words”.
As for
Well, it “sounds similar” in the same way that ‘square wheels’ sounds similar to ’round wheels’. And one can discuss the similarities of the wheel’s axles, but that misses the point.
Richard
here is the ‘zombie wood’::
http://www.nature.com/ngeo/journal/vaop/ncurrent/full/ngeo1797.html
and a short introduction by
http://www.realclimate.org/index.php/archives/2013/04/the-pages-2k-synthesis/
also there is a explaining piece by sceptical science:
http://www.skepticalscience.com/news.php?n=1978
BA:
My post at April 22, 2013 at 3:11 am provides a complete explanation of the issue. I now write to demonstrate that – as I suppose Manfred also can – I am able to state the matter in my own words.
The issue is that – when the data is autocorelated – rejecting data which fail the calibration test biases the remaining data to provide ‘hockey sticks’ when the remaining data are averaged.
This introduction of bias is counter-intuitive, but it is real and generates ‘hockey sticks’. In effect, and considering my analogy, the bias selects for ‘square wheels’ and rejects ’round wheels’ from the available data. This is what Lucia’s cartoon shows: removing the time-series which do not correlate with global temperature measurements over the time those measurements are available rejects the time-series which would negate a ‘hockey stick’ and so a ‘hockey stick’ remains in the average of the remaining data.
Richard
Maarten:
re your post at April 22, 2013 at 3:35 am.
Treemometry is bunk. And posting links to propaganda blogs which try to excuse the bunk does not affect that. Please read the links I provided in my post at April 22, 2013 at 3:11 am. One or both of those links will probably explain to you why treemometry is bunk.
Indeed, the fundamental error with treemometry is the same mistake as was made by Rhine when he studied ESP in the 1950s.
Rhine tested subjects for ESP. He then studied those subjects who passed the test. His conclusion was that ESP exists. But his conclusion was an artifact of his selection. And with time the ability of his subjects to exhibit ESP failed.
Treemometry tests tree rings for their ability to match temperature over a calibration period. It then studies the average of those tree rings which pass the test. But the average is an artifact of the selection. And with time the ability of tree rings to indicate temperature fails (i.e. the divergence problem).
Treemometry is pseudoscience.
Richard
@ur momisugly richardscourtney April 22, 2013 at 4:33 am
1.the first link is to a scientific article, not ‘ propaganda’; but for all of you not able to read behind the paywall of Nature Geoscience I provided two summaries;
2. why use the word ‘ proganda’? was I rude and did I call the expression ‘ Zombie Science’ propaganda?
3. If you had read the article and the connecting supplementary materials you would have noticed that it is NOT all about treering science, there are more proxies possible;
It may be full of holes and half-rotten, but it’s alive. Thanks for the inspiration Josh.
http://i38.tinypic.com/2gy36ug.jpg
@ur momisugly richardscourtney April 22, 2013 at 4:33 am
This you wrote:
“Rhine tested subjects for ESP.
He then studied those subjects who passed the test.
His conclusion was that ESP exists.”
how did he do the test from the first line?
Maarten:
re your post addressed to me at April 22, 2013 at 4:50 am.
I would be pleased to read anything substantive which you provide in response to the explanation I provided to you in my post at April 22, 2013 at 4:33 am.
In answer to your existing responses, I say the following.
I said you linked to propoganda blogs, and I did NOT say all your links were.
Yes, as you say, one of your three links was to not to a propaganda blog: it was to the abstract of a paper. Your other two links were to SkS and RC; i.e. propaganda blogs.
And, yes, some other proxies were also used. But, so what? For the reason I explained, the introduction of treemometry creates the ‘hockey stick’ whether or not some other proxy data was included in the average of the proxies.
In summation
Links to propaganda blogs does not provide credibility. And arm-waving in response to an explanation harms credibility. So, I would welcome anything substantive which you wish to say in refutation, dispute or agreement with my explanation.
Richard
richardscourtney says:
“Well, if you don’t like the Lucia’s excellent cartoon then perhaps Lucia’s “own words” will help you to understand. You can read them at
http://rankexploits.com/musings/2009/tricking-yourself-into-cherry-picking/”
Lucia’s made-up example, cartoon and words, are unrelated to the research. Nick Stokes who does know data analysis tries to explain why, but she seems not to get it. Do you?
“Well, it “sounds similar” in the same way that ‘square wheels’ sounds similar to ’round wheels’. And one can discuss the similarities of the wheel’s axles, but that misses the point.”
If you understand, perhaps you can explain it to me using real words, not about ‘square wheels.’ What is the practical difference in this case between “weighting by correlation” as you say Ahmad et al. have done, and weighting with factor score coefficients calculated by principal components?
Maarten:
Your entire post at April 22, 2013 at 4:53 am says
1.
Rhine determined if people accurately predicted cards pulled from a deck better than expected from random chance; i.e. it selected people who showed they had ability to predict cards.
2.
Treemometry determines if a tree’s rings accurately predict temperature as indicated by thermometers: i.e. it selects trees whose rings show they have ability to predict temperature.
I both cases such an apparent ability will be exhibited by some subjects. But it is an error to assume that either case selects for subjects with predictive abilities
There, your ‘red herring’ has been caught, landed and thrown back overboard.
So, I repeat,
I would welcome anything substantive which you wish to say in refutation, dispute or agreement with my explanation provided to you in my post at April 22, 2013 at 4:33 am.
Richard
BA:
I read your post at April 22, 2013 at 5:12 am.
Obviously, you do not understand the issue. However, it is not clear what you fail to understand.
I will try to help if you will explain your problem.
Richard
Richard,
You ake it more serious than it is: Lets quote Josh: Sunday silliness: Zombie Wood;
he or she enters ‘propaganda’, I enter some (missing) info, just to add to the ‘discussion’ on the ‘ best science blog’ (see quotation on homepage)
I is not my intention to start a discussion on tree ring science or ESP
richardscourtney says:
“Obviously, you do not understand the issue. However, it is not clear what you fail to understand.”
I thought this was clear, but I’ll ask it again. I wrote,
“If you understand, perhaps you can explain it to me using real words, not about ‘square wheels.’ What is the practical difference in this case between “weighting by correlation” as you say Ahmad et al. have done, and weighting with factor score coefficients calculated by principal components?”
Friends:
At April 22, 2013 at 3:35 am, Maarten wrote a post which linked to the abstract of a paper and to excuses for treemometry by SkS and RC.
I refuted that and I then explained why treemometry is bunk. I explained that treemometry is an example of the same selection error as was used by Rhine in his ESP studies.
At April 22, 2013 at 4:53 am, Marten replied with a post which only asked for details of the methodology used by Rhine.
I replied to each of Maarten’s posts.
At April 22, 2013 at 4:53 am Maarten wrote a post which ended saying
Conclusion:
Don’t feed the troll.
Richard