New Article in Chronicle of Higher Education

Here is a link to a new article in Chronicle of Higher Education about the HS debate in congress. This link should work for the next 5 days. A sidebar is available at the following link. There will be an online discussion Wednesday Sept. 6 at noon eastern time with Gerald North about the hockey stick and the broader issue of politics and science at

The tone is less antagonistic towards me than an article last year. A couple of obvious and annoying errors though. The realclimate blog started first and devoted much of its initial energy to trying to preempt our ctiticim of the HS, which was then about to be published. was started in response.

They also say in the sidebar that other recent studies do not use the contested bristlecones. In fact, the two most recent studies, Osborn and Briffa 2006 and Hegerl et al 2006, in small populations of 14/12 sites each use Mann’s PC1 plus a separate foxtail series.


  1. Steve McIntyre
    Posted Sep 5, 2006 at 3:57 PM | Permalink

    I wrote to Monastersky with the above criticisms. He replied:

    I did not state in the article that Mann and others started the RealClimate blog after your blog. I said that they started it in part to respond to comments about their work.

    With reference to the NRC report, its recommendations about the use of bristlecone pines, and recent reconstructions, I would refer you to Franco Biondi, a dendrochronologist, who participated in producing the report.

    By discussing this blog before realclimate, he certainly gives the impression – intentionally or unintentionally – that realclimate was responding to us, rather than the other way around.

    With respect to Biondi, it’s not a dendrochronological issue whether Osborn and Briffa 2006 used Mann’s PC1 (and a separate foxtail series); it’s a matter of fact. If Biondi thinks otherwise, he’s wrong and that’s the trouble with these fly in-fly out panels, where nobody actually checks the data.

  2. Dave B
    Posted Sep 5, 2006 at 4:19 PM | Permalink

    THANK YOU for your wotk on this, steve. material like the HS getting into textbooks is a good example of bender’s idea that the team will just keep inventing new ways of building a consensus upon a pile of sand. it is disgraceful, and unfair to present and future students.

  3. Jeff Weffer
    Posted Sep 5, 2006 at 6:49 PM | Permalink

    I love how the article says that its not fair to exclude the error bar “shades” in the hockey stick graph.

    The article notes “Those who doubt that humans are warming the climate substantially also often print the curve without its error bars, and then disingenuously attack it for being inaccurate. The editorial page of The Wall Street Journal recently published a naked copy of the hockey stick as part of a critique of it.”

    But including the error bars just reinforces how inaccurate the chart really is. The error shades just gives the reader the ultimate incorrect visible interpretation of the record.

    The shading only accentuates an uninformed reader’s gut reaction that today’s temperatures are the hottest in a millenium. The bolded hockey stick within the error bars simply attached to the 20th Century temperature record just solidifies the reader’s interpretation that today’s temperatures are way out of the climate norm. The increase must be caused by AGW.

    The purpose of the hockey stick is to rewrite climate history. The objective is to get rid of the LIA and the MWP and the dark ages. The purpose is to prove that natural climate variability does not exist.

    The purpose is to rewrite the climate record so that AGW is proved simply by a graph. All by way of manipulating data.

    The article rewrites the debate (like the hockey stick rewrites the climate) to prove a point which is so far unproven. The article is just like the error bars, manipulation.

  4. Pat Frank
    Posted Sep 5, 2006 at 6:55 PM | Permalink

    I’ve submitted these questions for the Chronicle’s conversation with Gerald North:

    The original hockey stick has been shown not just flawed but wrong. Why was the NAS committee unable to clearly state that?

    A more basic question: There is no analytical theory that can extract a growth temperature from tree ring widths or tree ring densities. On what scientifically valid grounds, therefore, can anyone possibly “calibrate” a tree ring series using temperature vs. time data?

  5. Ken Fritsch
    Posted Sep 5, 2006 at 6:59 PM | Permalink

    Gerald North, a professor of atmospheric sciences at Texas A&M University at College Station, served as chairman of the National Research Council committee that has investigated the hockey-stick curve. Despite finding some problems in the seven-year-old study that was the basis of the curve, the panel determined that it is basically correct in its conclusions, which have been corroborated by more-recent work. The scrutiny by Congress “is kind of a delaying tactic to find little things like this to slow down government action on greenhouse-gas limitations,” says Mr. North.

    That is not a statement that an apolitical or nonpolitical and disinterested scientist would make, much less one who headed a panel to pass judgment on the issue. Of course, the general approach of the NAS panel in quickly defending AGW despite being obligated to criticize the HS was a dead give away from the beginning. Why would not North (and any disinterested scientist sans an agenda) say that there was too much uncertainty in the data and the methodology involved to determine temperatures relative to the present going back more than 400 years? And that would be for all temperature reconstructions not just the HS.

  6. Steve McIntyre
    Posted Sep 5, 2006 at 7:14 PM | Permalink

    Actually the error bars may be the MOST inaccurate calculation in MBH, which is quote a statement. The error bars are calculated from the calibraiton period residuals. If you look at the graphs plotted up in this post, Mann (in effect and without knowing what he was doing) did a partial least squares regression on a large number of series. In the example, red noise series produce just as accurate a fit (i.e. calibration period residuals) as real data. Hence the error bars are totally meaningless. The error bars don’t bracket the real errors any more than red noise does. UC’s clever diagram is the only available reconstruction from MBH proxies. Essentially a straight line with huge error bars.

  7. Pat Frank
    Posted Sep 5, 2006 at 8:29 PM | Permalink

    #6, and the error bars are scientifically meaningless anyway, because they do not reflect an estimation from a theory competent to derive temperatures using tree-ring variables.

  8. Steve McIntyre
    Posted Sep 5, 2006 at 9:47 PM | Permalink

    PAt, also note Gerry North seminar presentation here

  9. bender
    Posted Sep 5, 2006 at 9:48 PM | Permalink

    Re: graphic presentation. Along with the “error bars” and “flavor uncertainty envelopes” issues raised here, I think the curve itself should be shaded along its length, with the degree of contrast (dark line on white paper) proportional to the degree of certainty. I want to see that line fading off into oblivion – as the data clearly suggest – as the time-series recedes back from AD1600. To plot that line in a single shade, or thickness, along its full length creates the illusion that the mean itself is as certain 1000 years ago as it is today. Gerald North tries to fix this by shading the background darker back in time. Shade the line itself. When media people are squinting to resolve the location of the line AD1000, that’s when you know you’ve got the right shading gradient. (I’ve mentioned this device before, but it seemed apt to mention it again here.)

  10. Posted Sep 6, 2006 at 10:40 AM | Permalink


    True, presentation makes a difference. Added MBH98 reconstruction to the figure here. Does it look like hockey stick anymore?

    (ref. topic Snooping. I wont say that the hockey stick fits in to the model, due to #10, too metaphoric 🙂 )

  11. bender
    Posted Sep 6, 2006 at 11:30 AM | Permalink

    Looks like a fly buzzing around inside a whiskey jug.

  12. Pat Frank
    Posted Sep 7, 2006 at 12:43 AM | Permalink

    #8 — I’ve now had a chance to listen to North’s seminar. I really have to say that I found his general attitude and all that avuncular ‘ha-ha-ha-ing’ irritatingly patronizing. The attitude comes out right away with the slide showing “Enter the amateurs (M&M)” He was even patronizing towards Wegman, the cause being Wegman’s very reasonable suggestion that statisticians be included among workers doing proxy reconstructions (as though it were a patent ‘make work for statisticians‘ gambit). Then North extended his patronizing dismissal to statisticians in general, wanting to exclude them from science because they tend to be “anal.” He also disparaged you (Steve M.) and Ross M. as ‘half-fools’ implied Ross is non-credible, and rambled toward the 30-yard line by completely mis-stating the history leading to ClimateAudit. He finally told a story that one of Wegman’s reviewers sent him (North) an email saying she only got 3 days for her review, thus casting general aspersions on Wegman’s claim of independent review.

    The feeling one gets is that he considers most oppositional people as zanies, and most of his audience was clearly in agreement. He offered empirical evidence for AGW in the apparent correspondence between the ‘hockey stick’ trend of rising CO2 over the last millennium juxtaposed with the MBH hockey stick trend. One wonders if he knows the difference between ‘association’ and ‘causality.’ North promised at the beginning of his talk to tell us whether the M&M analysis “makes any difference,” but instead later passed over the statistical questions as “too arcane,” and ended up never addressing the question at all. He said there are problems with PC analysis ala’ MBH but never said what they are, and never mentioned that PC’s are numerical constructs without inherent physical meaning.

    His arguments concerning the environmental determinants of tree-ring widths were totally hand-waving. There was nothing analytical in them, or theoretical, or quantitative, even in reference. He passed quickly over those lacunae to go directly to splicing the tree-ring widths to the temperature record (a grotesque miscarriage of science). He talked about the physically confonding inputs affecting tree-ring widths, such as out-of-range growth and non-linear effects, but then never discussed them again as though they had no effect. That whole part is a ludicrous display of non-science.

    In his discussion of ice-cores, right after he finished saying that dO-18 temperatures from tropical ice-cores are confounded by possible changes in hydrology (as opposed to those from the poles), he went on to exclaim over the 20th century HS shape of dO-18 temps derived from ice cores taken from Tibet and the Andes! I.e., ice-cores from the very places he just finished saying were untrustworthy!

    At the end of his talk, as he discussed all the various flavors of published reconstructions, North not only said that the error bars, more than just statistical, are also physical, but he went on to say that we don’t really know what the errors are. That is, after all the song-and-dance, and displays of other proxies that look like hockey sticks, he said in effect that we don’t really know, after all, whether the hockey stick is even a hockey stick. And then he concluded that part with the observation that what you do on an “expert panel” is “kind of [wing] it.” Incredible. He winged his entire analysis.

    On the question of whether the 20th century was the warmest in 1000 years, he gave a chuckling account of the committee’s choice of the word “plausible” and left us with this, both on his slide and stated: “[Using] 30-year averages, warmest in 1000 yrs – (Plausible, -reasonable, impossible to bring a convincing argument against, no numbers).

    Of course, his comments about not knowing the errors leaves us with no convincing argument for, either, but that side of the coin was left unexplored. I’m left wondering whether the retreat of the northern treeline since the MWP would be considered “a convincing argument against.”

    During the Q&A, he laughed about a conversation he had with a reporter from the Dallas Morning News, who told him Rep. Barton had said, during a speech, that the HS was statistically discredited. North went on to laugh disparagingly about Barton’s dismissal and talk about being cynical concerning politicians and their hearings. No matter that North himself said not 20 minutes earlier that no one knows what the error bars are on the proxy reconstructions. It’s as though his mind contained two non-overlapping magisteria (to use a peculiarly apt SJ Gouldian logism).

    At the very end, pace Lee, North said that the real basis for putting the “A” in AGW is the correspondence between the GCMs (“we know all the forcings,” and, “the lynch-pin is the physics“) and global temperatures. He dismissed the uncertainties in climate physics with passing notice.

    That puts paid on it for me. There is no scientific case for AGW. I swear, there isn’t another branch of physical science that would get the pass the climate science gets for shoddy thinking and suspect methodologies.

  13. Posted Sep 7, 2006 at 2:01 AM | Permalink

    #12. Thanks for that Pat. Even without statistical training the average person can see this is not the way one conducts oneself if this were a serious project. I think most people can compare this patronizing and flippant attitude so prevalent in the field with virtually anything else, from rocket science to doing one’s tax, and see there is something very wrong here. Barton made a serious attempt at due diligence in one small area. They were found wanting. Next.

  14. Posted Sep 7, 2006 at 2:14 AM | Permalink


    Shade the line itself.

    Yes I think there should be no mean line at all, just boundaries representing the limits of the null hypothesis, derived from monte-carlo simulations say, which is itself hockey-stick shaped. Then, if in a reconstruction the line goes outside the boundary, then that would suggest, in that part of the reconstruction, a significant divergence from expected. Done this way, the only real signal in these data would probably be the LIA.

%d bloggers like this: