Deep Climate continues his examination of the Wegman report. It would seem that Wegman's "reproduction" of McIntyre's results amounted to nothing more than running McIntyre's code without understanding what it did. And while Mann's "short centring" method does tend to produce a hockey-stick McIntyre greatly exaggerated the extent that it does so.
Without bona fide climate scientists advising and educating the statisticians, which would quite probably be what the denialsphere would regard as preferential (climate scientists will corrupt the statisticians somehow), we're doomed.
The Wegman Report must be the most highly regarded statistical analysis amongst contrarians, sceptics and deniers alike, but all it's turning out to be is an exercise in mass confirmation bias.
No, it's much worse than that.
Once the WR is widely and officially recognised for what it was, M&M's entire intellectual underpinning (such as it is)of the denialist machine goes down with it.
Not that that will stop it completely, but it's a great start.
Looks like the Wegman and M&M scandal is beginning to look like the Bre-X scandal. Not surprising since both were orchestrated by mining executives. Cherry picking is just as dishonest as salting mining samples.
On a related note, a random blog commenter was recently touting a recent paper (2010) by McIntyre and McKitrick, so I asked him what, in his understanding, their method actually was. I still haven't got a reply.
Frank, keep in mind that said commenter also refers to Pat Michaels as his "hero".
I thought the thread between you, Timothy Chase, and the CA guys was going to go on forever...don't they have day jobs?...but as soon as Tim argued that McIntyre "stokes paranoid delusions on the part of rightwing fanatics that â judging from the death threats and at least one dead rat that someone found on their doorstep just after the doorbell had been rung one night â may end up getting someone killed some day", they disappeared. It was really weird.
1) Read SSWR, pp.134-135:
"Given all this, I had resolved to avoid the real statistics analysis in the WR, but eventually realized there was none."
"Using McIntyreâs code and help, they reproduced MM05b-like charts. Re-running code does not prove that its assumptions are correct."
"The WR contains no actual new statistical analysis of MBH itself, just reworks of MM material."
Of course, I talk to DC occasionally.
2) But do not be deterred by the stats discussion in DC's opus.
To me, the real zinger is showing that WR Figure 4.4
a) Was produced by McIntyre code whose output hadn't been published.
b) Was effectively presented as a sample of 12 from 10,000, that showed that MBH "mined" random data for positive hockey sticks.
c) Whereas it was a sample of 12 from the 100 (1%) sorted as having the highest positive hockey sticks... MM screwed up various places in the sequence of steps to generate that, but this takes cherry-picking to a new height. Does anyone have any idea how serious statisticians might view this?
d) In the US, this would be like declaring the average male height to be 6'6", without bothering to mention the sample was taken on an NBA basketball court.
Does anyone know where GMU is at with their investigation of Wegman et al?
Now we are all eagerly awaiting Judith Curry's reaction.
Well, whatever the cause of his leaving, it's clear that his 'arguments' were getting less and less defensible.
Re 8 Jeremy c
A) VP Research takes quick look at complaint. If substance:
B) inquiry committe named, only job is to decide if an actual investigation warranted, writes yes/no report.
C) If warranted, investigation committee formed, investigates in detail.
I've surveyed policies at 6 schools, all of which require report back to complainant at end of B) and C). At GMU, each committe recommends whether or not to give report to complainant.
In Bradley's case B) should consist of locking at DC's side-by-side comparisons (of Bradley text and a month later, the SNA text) and reporting whether or not they look enough like plagiarism to investigate. Personally , I would estimate this to take 5 minutes, but what do I know?
Sadly, wheels turn slowly at GMU.
It took them 5 months to have the first meeting of inquiry committee.
It has been almost 3 more months, and at least as of 11/15/10, no *inquiry* report had been given to Bradley. Again, their process seems to allow them to say nothing, ever....
Sorry this is kinda off topic (but on metatopic);
but I think it's almost as good as the "What if we make the world a better place for nothing" cartoon.
Yes, McIntyre's exaggeration of the effect of Mann's algorithm is one clue that he is not some honest participant in the debate (as if you needed more clues than his obsession with himself and certain climate scientists who don't take him seriously)
Another example of McIntyre's sketchy relationship with the truth comes from his recent reply in PNAS to Mann 2008. McIntyre claimed that:
Their CPS reconstruction screens proxies by calibration-period correlation, a procedure known to generate âhockey sticksâ from red noise (4)
Ref. 4 is:
Stockwell D. (2006) Reconstruction of past climate using series with red noise. AIG News 8:314.
(Ignore for the moment that this ref is not peer reviewed and was published in a newsletter)
If you take a look at figure 1 you will notice that what McIntyre calls a "hockey stick" is a record with an MWP almost as warm as late 20th century warming. The only similarity to the MBH graph is the twentieth century warmth--and that is already known.
McIntyre loves playing language games. Watch the pea under the thimble, as he loves to say. :)
So Stockwell generated a set of random stochastic series with red noise, smoothed with a 50-years filter, selected those that had a distinct upwards blade (i.e. significant correlation with temperatures from 1885-1990) and found that they tend to look like hockeysticks.
FYI: Wegman plagiarism story going mainstream - USA Today, today.