Wednesday, May 9, 2012

4058.txt

date: Wed Feb 12 09:23:35 1997
from: Keith Briffa <k.briffaatXYZxyz.ac.uk>
subject: the paper
to: Brian

We must put in more details of the calibration/verification stats.
This means showing the regression coefficients for both calibration sub periods
as well as the overall period equations. We must show re,ce stats and the significance levels of the verification rsquared and sign tests - I taqke it
you have all these. What is the point of Table 3. All reconstructions based on regression underestimate the extreme predictand years .This Table just says the most extreme instrumental years produce the biggest residuals. I would drop this. Did you take a note of how the ring-width data were standardised?
I'm making corrections/suggested ammendments to the manuscript. In the meantime you can get all your refs. confirmed and listed. I will sort out which ones of ours
need to go in and rewrite the bit on calibration and data standartdisation if you can fax me back the stats a.s.a.p. this will help. Perhaps you could ring today?
best wishes
Keith

No comments:

Post a Comment