Things seem to be hotting up on the Michael Mann front, not least because Steve McIntyre seems to have returned to blogging with a vengeance, assisted as always by his trusty band of followers. Today, the climate auditors have turned up another rather embarrassing problem with Michael Mann's legal submission. This document claims that Mann had nothing to do with the infamous cover graphic for the WMO report of 1999, of hide the decline notoriety. Unfortunately, the claim is directly contradicted by Mann's own CV.
I found myself thinking about another of Mann's claims this morning. This was prompted by a comment on David Friedman's blog about Mann's claim in MBH98 that he had used "conventional" principal components analysis. The author of the comment wondered if this could in fact be true. But readers of the Hockey Stick Illusion will recall that the claim of "conventional" was actually only made about Mann's processing of temperature data. Regarding the tree ring data we were only led to understand that PCA had been used.
At the time there was a question hanging over the eventual use of short-centred PCA for the tree rings: whether the short-centring was an error or a deliberate step. I didn't really go into this in the Hockey Stick Illusion, as we had little way of knowing. However, in the last few years we have some more details.
The most important of these is Mann's own book on the subject, in which he makes it clear that the decision to short-centre was deliberate: he describes this as his "modern" approach to centring the data. This being the case, we are left with a lot of questions. Why would you decide to use a novel approach to data centring? What was wrong with conventional PCA? But more importantly, why, having decided to use such a novel approach would you not mention it in the methods section? Is that not something that readers of the paper would need to understand what was done? Was this not a key methodological choice?
Inquiring minds want to know.