Sunday, January 15, 2012

Email 4854, Oct 2003, Phil Jones: "It is rather odd that the email said [M&M] had rerun his (Mann's) exact analysis and got quite different results. I know I couldn't do this, as when Keith, Tim and me wanted to do some comparisons with MBH98 a few years ago a few of the series could not be made available."

Email 4854

subject: Re: CONFIDENTIAL

Thanks Phil, Got your email just as I sent off my latest. I agree fully with what you say--it is very difficult to repeat such an analysis exactly, and the real point here is, who knows what this guy (Steven McIntyre--I don't know who the supposed 2nd author is) actually did. The Mann et al '99 paper was clear that the results were sensitive to a small number of skillful predictors prior to AD 1400, and that non-climate biases had to be corrected for in some of the longer series to get a skillfully cross-validated reconstruction. Without knowing what the guy did, I'm guessing that he doesn't even demonstrate that his alternative "reconstruction" passes cross-validation. If not, its all moot... But more fundamentally, this wasn't submitted to a legitimate peer-reviewed scientific journal. Its a social science journal, and one that has shown a disdain for peer review (e.g. in publishing the Soon et al Climate Research paper essentially in its original unedited form--and see the recent documented comments of the editor). I agree this might blow over, but the folks in DC, such as McCain and Lieberman, who are fighting to represent what the legitimate scientific community has to say, need to be prepared in case the special interests try to use this. Hence, the short response I sent out. [Mike Mann]

[Phil Jones] Mike, Depending exactly on what it says I suggest we should do our best to ignore it. E&E is edited ( a very loose use of the word) by Sonia Boehmer-Christiansen, who's generally involved, in some way, in all skeptic stuff here in Britain. It is rather odd that the email said the two had rerun his (Mann's) exact analysis and got quite different results. I know I couldn't do this, as when Keith, Tim and me wanted to do some comparisons with MBH98 a few years ago a few of the series could not be made available. I'm not trying to make any sort of point here, just to state that repeating an analysis with exactly the same data is normally very difficult.

1 comment:

Anonymous said...

And it never occured to him that this might, just might, be a red flag.