#Nanmean matlab 2009 archive#
#Nanmean matlab 2009 verification#
#Nanmean matlab 2009 code#
#Nanmean matlab 2009 series#

IвЂ™d think that confidence intervals should follow standard methodologies and it would be very hard for Mann to justify a roll-your-own method, in fact CI should be about the easiest thing to reverse engineer, given the data.Д«ut harking back to the the gadfly post, I do have to fault you on your snark at the end of the post. I am quite interested in seeing your comment to Mann 2008, of course I guessed way wrong on your cited references, but you evidently focused your comments on your area of expertise, statistics, rather than MannвЂ™s bailiwick paleo-climatology.

#Nanmean matlab 2009 series#

Unless both the instumental series and the reconstruction has the same mean over the calibration interval, these are not the same.

#Nanmean matlab 2009 verification#

If IвЂ™m not mistaken, the formula for verification RE (given in the same form as in the NAS report) is 1-MSE(y(verif),yhat(verif))/MSE(y(verif),y_mean(calib)), and now Mann is calculating 1-MSE(y(verif)-y_mean(calib),yhat(verif)-yhat_mean(calib))/MSE(y(verif),y_mean(calib)).

So the (calibration) means are subtracted from the instrumental and the reconstruction (_reconmean, _actmean) before calculating the difference (diffmean). Meansq=nansum((e_actmean(vals1:vale1)).^2) Дiffmean=l_actmean(vals2:vale2)-l_reconmean(vals2:vale2) L_reconmean=l_recon-nanmean(l_recon(cals2:cale2)) Дiffmean=e_actmean(vals1:vale1)-e_reconmean(vals1:vale1)

#Nanmean matlab 2009 code#

Well, I think (PLEASE correct me if IвЂ™m mistaken) the REs and CEs are calculated wrong, hereвЂ™s the relevant code (verification REs):Д®_reconmean=e_recon-nanmean(e_recon(cals1:cale1)) Psu.edu/~mann/supplements/MultiproxyMeans07/code/ The method of uncertainty estimation (use of calibration/validation residuals) is conventional (3, 4) and was described explicitly in ref. Update: Their reply, needless to say, did not acknowledge that they placed the code online AFTER our comment: One thing that seems вЂњvery likelyвЂќ to me: IвЂ™ll bet that, in their Reply to our Comment, Mann will say that we were вЂњwrongвЂќ in our statement that the archived source code did not contain this calculation because if you go to their website, you can see that the calculation is there. UC and Jean S have taken a quick look at this code and we should have something to report in the next week or two. Although their cover webpage reports some other changes, this change was not reported there. Jean S just noticed that, on Dec 15, 2008, Mann added what appears to be the relevant source code as. In our comment (submitted on Dec 8, 2008), we commented on the calculation of confidence intervals, noting that, contrary to assurances in Mann et al 2008, the source code did not contain their calculation of their Figure 3 confidence intervals. (But again in fairness to Gerry North, IвЂ™ll bet that our PNAS comment was sent to him for screening and heвЂ™s the sort of person who would be fair.) I emailed Gerry North, said to have reviewed Mann 2008, seeking an explanation, but he told me to вЂњmove onвЂќ.

#Nanmean matlab 2009 archive#

And while, to his credit, Mann archived a lot of source code вЂ“ a fairly brave undertaking given that the code is a lot like GISTEMP on a scale of 1 to needles-in-your-eyeballs, he didnвЂ™t archive the source code for confidence interval calculations вЂ“ so we had another mystery.

You canвЂ™t really tell from the running text or the SI. I was very interested to see how Mann 2008 calculated confidence intervals.

Mannian confidence intervals have always been a mystery with MBH99 confidence interval methodology remaining an intractable mystery that has defeated all reverse engineering (and engineering) efforts by UC, Jean S and myself to date (though we havenвЂ™t picked up this file for a while.)