Sunday, November 21, 2010

The Climate Change Data Scandal You Won't Read Much About in the Telegraph, Mail or the Express

This is a very techey story and much of it is above my knowledge but from what I can understand, but it is very, very interesting. It also shows the inner workings of a denial home-goal. Here's the background. There is this statistics professor Wegman who gets commissioned to write a report examining the famous Hockey Stick (that's the graph that shows global temperatures with a sharp and scary rise at the end of the 20th Century.) Now, climate deniers hate the Hockey Stick and they hate it's most prominent author, Michael Mann. I guess because it's an easy to understand graphic that communicates the serious situation we find ourselves in and also became Mann speaks his mind.

So they went after it. They commission Wegman who produces a big report and it's conclusions tear into the Hockey Stick and all the denialists cheer! Woo! Woo! The meat of the report is an allegation that the computer code that Mann and others used to error-correct the data for the Hockey Stick was doctored to produce a hockey-stick shaped graph, whatever the data that was input. In short they are saying that Mann faked it and the Wegman report proves it.

Except that there a murmurings of problems with this report. Murmurings that wont die. First there is an accusation of plagiarism; the unattributed copying of somebody else's work. This is a serious charge in academic circles. Second the data analysis that 'proves' the fake Hockey Stick result itself looks, well, fake...

[George Mason University] spokesman Daniel Walsch confirms that the university, located in Fairfax, Va., is now investigating allegations that the Wegman report was partly plagiarized and contains fabrications. Last month, a 250-page report on the Deep Climate website written by computer scientist John Mashey of Portola Valley, Calif., raised some of these concerns. Mashey says his analysis shows that 35 of the 91 pages in the 2006 Wegman report contain plagiarized text (with some of the wording taken from a book, Paleoclimatology: Reconstructing Climates of the Quaternary, by Raymond Bradley of the University of Massachusetts) and contain erroneous citations of data, as well.

Some denialists have argued that this is just about a missing attribution and it means nothing. Now remember that a couple (out of thousands of studies) of incorrectly attributed references to a study in an IPCC report was blown out of all proportion by the denialosphere. But now they argue we should not worry about a couple of attribution errors.. (Can't have it both ways!)

Here's the interesting bit; about the Hockey Stick. In essence the data analysis used by Wegamn, which was copied uncritically from the work of a couple of other denialists, generated some noisy data (but meaningless) data then feed it into a version of Mann analysis and out pops a Hockey Stick-like result.

Except that the generation of the input data was far from random - oh no! It generated 10,000 sample sets of data - then selected the 100 from that huge pool that were the most hockey-stick like! These 100 cherry picked results are then fed into the replication of Mann's analysis. Wegman's analysis is far from an objective analysis. It would appear that far from being a random test, Wegman was using loaded dice. It seems the Wegman report may be a huge fudge. (You can read a very detailed breakdown of the full fudgey nature of the fudge here.)

Forget the so-called 'climategate' - this is the Climate Change Data Scandal You Won't Read Much About in the Telegraph, Mail or the Express

No comments: