FAQs

1.  What is the connection of the present studies to your original article in 2003?

     The present articles build on the first article. In that article, we pointed out that there were serious problems with the data set in MBH98 and, in particular, with the tree ring principal component series. We also showed that quite different results could be obtained for the 15^th century under reasonable assumptions using the MBH98 method. Since then, and largely because of the effect of the original article, a great deal of new information about MBH98 has been made available. In July 2004, at the direction of Nature, Mann et al. published a Corrigendum, which included a voluminous archive on data and methods used in MBH98. Earlier, Mann et al. made public the address for the data actually used in MBH98, rather than the address which they had previously provided us. The present articles reflect detailed study of this new material. In particular, we are now able to precisely diagnose the problems with the principal component series in MBH98, which previously were simply noted as being incorrect.  

     We characterized our first article as raising “audit issues”; without co-operation from the original authors, it could not be a complete audit. In retrospect, the issues raised have proved to be very important. In MM03, we reported that we were unable to replicate the MBH98 principal component series. Given the importance of Mann’s PC methods, one would think that this replication would have been attempted by someone else. In MM03, we were not in a position to fully diagnose the problems, but we are now. Also, the simple comparison of archived series versions to versions actually used revealed the unreported editing of the Gaspé series, which also had important consequences. This effect was not specifically analyzed in MM03, but was analyzed here.

     In short, we believe that the present articles are a definitive resolution of issues first raised in MM03.

 2. Has Michael Mann accepted any part of your arguments?

    There is surprising agreement between Mann and ourselves on the effect of differing assumptions on the NH temperature reconstruction – if the assumptions are specified exactly. For example, both of us get high early 15^th century results with centered PC calculations and 2 PCs in the AD1400 North American network and both of us get low early 15^th century results with centered PC calculations and 5 PCs in the North American network. We have tried to canvass these matters in an evenhanded way in our E&E article (see pages 75-76) to show what is agreed and what is not agreed.  

     Mann has categorically denied that his PC method generates hockey stick shaped series from red noise (realclimate#Temperature question #5, but we see no way that he will able to sustain this argument, in the face of the compelling evidence to the contrary in our GRL paper.  

     Mann and ourselves agree that centered PC calculations are “standard” (for this acknowledgement see realclimate#Yet paragraph 3). 

     We had thought that both Mann and ourselves agreed that a valid reconstruction should pass a range of statistical verification tests – this is a position which we endorse and a position which Mann seems to endorse ( see realclimate#Myth 1 3^rd paragraph) .  However, it seems that Mann et al. are inconsistent on this matter and are now arguing that you should only look at one verification test (the RE statistic) to establish significance. We doubt others would endorse this position, but in any case we show in the GRL paper that their reconstruction is not statistically significant, when RE benchmarks are correctly calculated.  

     Both Mann and ourselves have done calculations showing that the MBH98 reconstruction is not robust to the presence or absence of bristlecone pines: this is not under dispute.

    We presume that Mann has done calculations showing that the MBH98 reconstruction is affected by their editing of the Gaspé tree ring series or else they wouldn’t have done the editing. Strangely, the Gaspé series is used twice in the MBH98 model: it appears simultaneously in the North American (“NOAMER”) network and in the “northern treeline” network. But only in the second usage is it edited, whereas in the NOAMER network the archived version was used.  

Are you saying the 15^th century was warmer than the present?  

     No, we are saying that the hockey stick graph used by IPCC provides no statistically significant information about how the current climate compares to that of the 15^th century (and earlier). And notwithstanding that, to the extent readers consider the results informative, if a correct PC method and the unedited version of the Gaspé series are used, the graph used by the IPCC to measure the average temperature of the Northern Hemisphere shows values in the 15^th century exceed those at the end of the 20^th century.  

Does your work disprove global warming?  

    We have not made such a claim. There is considerable evidence that in many locations the late 20^th century was generally warmer than the mid-19^th century. However, there is also considerable evidence that in parts of the Northern Hemisphere, the mid-19^th century was exceptionally cold. We think that a more interesting issue is whether the late 20^th century was warmer than periods of similar length in the 11^th century. We ourselves do not opine on this matter, other than to say that the MBH results relied upon so heavily by the Intergovernmental Panel on Climate Change in its 2001 report are invalid.  

Where can we get further information on principal components methodology?  

     In all our discussions, a principal component series is weighted combination of up to 70 individual tree ring series – some readers may find it helpful to think of the Dow-Jones Index, which is a weighted average of individual stock prices. Principal component series can include negative weights, which result in showing a contrast between different series – picture a series with positive weights for finance stocks and negative weights for tech stocks. In principal components discussions, the weights have forbidding names like eigenvectors or empirical orthogonal functions, but, at the end of the day, these are just weights. The decomposition is prescribed by the matrix algebra. There are canned programs in high level languages so that the principal components decomposition of a matrix X of time series can be obtained in one line. As we discuss in our articles, these decompositions can be highly sensitive to transformations of the data – even if the transformation only seems to be a “standardization”. For a slightly more technical exposition see [to be inserted].  

Who paid for your research?  

     We have neither sought nor received funding for this work. For McKitrick, undertaking the project has required considerable time away from his own economics research. For McIntyre, undertaking this project has required an unpaid leave of absence from his career in mineral exploration financing, at the cost of over a year’s foregone earnings so far.