The NAS NRC panel on temperature reconstructions has released its report.

The press release states

There is sufficient evidence from tree rings, boreholes, retreating glaciers, and other “proxies” of past surface temperatures to say with a high level of confidence that the last few decades of the 20th century were warmer than any comparable period in the last 400 years, according to a new report from the National Research Council. Less confidence can be placed in proxy-based reconstructions of surface temperatures for A.D. 900 to 1600, said the committee that wrote the report, although the available proxy evidence does indicate that many locations were warmer during the past 25 years than during any other 25-year period since 900. Very little confidence can be placed in statements about average global surface temperatures prior to A.D. 900 because the proxy data for that time frame are sparse, the committee added. …


The Research Council committee found the Mann team’s conclusion that warming in the last few decades of the 20th century was unprecedented over the last thousand years to be plausible, but it had less confidence that the warming was unprecedented prior to 1600; fewer proxies — in fewer locations — provide temperatures for periods before then. Because of larger uncertainties in temperature reconstructions for decades and individual years, and because not all proxies record temperatures for such short timescales, even less confidence can be placed in the Mann team’s conclusions about the 1990s, and 1998 in particular.

The committee noted that scientists’ reconstructions of Northern Hemisphere surface temperatures for the past thousand years are generally consistent. The reconstructions show relatively warm conditions centered around the year 1000, and a relatively cold period, or “Little Ice Age,” from roughly 1500 to 1850. The exact timing of warm episodes in the medieval period may have varied by region, and the magnitude and geographical extent of the warmth is uncertain, the committee said. None of the reconstructions indicates that temperatures were warmer during medieval times than during the past few decades, the committee added.

Realclimate on the report:

However, it is the big picture conclusions that have the most relevance for the lay public and policymakers, and it is re-assuring (and unsurprising) to see that the panel has found reason to support the key mainstream findings of past research, including points that we have highlighted previously:

At Climateaudit on the other hand:

I’d characterize it more as schizophrenic. It’s got two completely distinct personalities. On the one hand, they pretty much concede that every criticism of MBH is correct. They disown MBH claims to statistical skill for individual decades and especially individual years.

However, they nevertheless conclude that it is “plausible” – whatever that means – that the “Northern Hemisphere was warmer during the last few decades of the 20th century than during any comparable period over the preceding millennium”. Here, the devil is in the details, as the other studies relied on for this conclusion themselves suffer from the methodological and data problems conceded by the panel. The panel recommendations on methodology are very important; when applied to MBH and the other studies (as they will be in short order), it is my view that they will have major impact and little will be lefting standing from the cited multiproxy studies.

Update

David Appell:

This should finally put an official end to the silliness that’s gone on for the last few years. I don’t doubt that McIntyre will continue to bloviate, but journalists especially now have no reason to give him any traction.

Roger Pielke Jr.:

My reading of the summary of the report and parts of the text is that the NAS has rendered a near-complete vindication for the work of Mann et al.

Coby Beck:

Well, the report is out and it seems to be a fairly strong vindication of Mann et al. There is some more fuzzy language that will surely be seized apon by some but there is certainly nothing to support the allegations of errors, omissions and frauds that had been thrown around. The main conclusion is that many other studies support these same findings and that this is not a central issue in the present and future of climate change.

Comments

  1. #1 Davis
    June 26, 2006

    …after a propagandistic blog by a large group of politicized scientists.

    If that’s how you view #1, why should we consider it a virtue that you’re #2?

  2. #2 Tim Lambert
    June 26, 2006

    Wow Lubos, your blog is much less authoritive than RealClimate. I thought you were smarter than them? How come you are getting thrashed in the authoritative stakes by William Connolley?

  3. #3 jp
    June 26, 2006

    Per

    Well done you have actually made a statement and phrased a question, you are one up on Lubos. You have written: “How do you show that there is a sensitivity to temperature ? If you have a completely random population with no response to temperature, some tree stands will show a rise, some a decrease, and some no change over a time period. This is basic statistics. If you simply select those trees with a rise (as 20th century temperature), it doesn’t follow that you are selecting temperature sensitive trees.”

    First a comment on language, in dendrochronology the term ‘Sensitivity’ has a specific meaning, it is a measure of the variance in a tree ring series. When a individual tree (or chronology) shows a high amount of variance it is called Sensitive, when a series shows little annual variability it is referred to as Complacent. Of course the word sensitive can also be used in a variety of contexts. If the objective of the research is to extract a climate signal from a tree ring series it would be of no value to sample trees that were not sensitive to climate. There are many abiotic and biotic variables that influence ring growth. For example, in many stands of trees, growth is not primarily limited by a combination of climate variables but rather by competition, access to nutrients, and similar variables. In my earlier comment on dendroclimatology I was referring specifically to “site” selection. The majority of sites for dendroclimatology are selected because they represent locations where the species in question respond in a measurable way to changes in some climate variable. This is why these sites are often found near altitudinal and latitudinal treeline (potentially sensitive to temperature), in semi-arid environments (potentially sensitive to moisture) or in other exposed locations where climate plays an important role in ring growth. When you go out and core trees this becomes apparent immediately. In a complacent tree (site) there is very little variance in the ring growth while in a sensitive tree (site) there is considerable variance in ring widths. So how “do you show that there is a sensitivity to temperature”, you measure it by constructing a chronology from a group of trees at a site and examining the correlation between the ring series and a range of climate variables. At a good site for dendroclimatology there will individual trees with a high sensitivity, a strong correlation between the trees growth patterns within a stand and a chronology that has a significant correlation with a climate variable(s). There is nothing random about site selection. If I am sampling trees to construct a chronology of snow avalanches or flooding I go to those environments. If I am attempting to produce a reconstruction of a climate variable I will select locations where the trees are sensitive to those variables.

    This is of course basic dendroclimatology, I would start by reading Fritts classic text on Tree-Rings and Climate, and then buy a corer and get out in the field. After you have cored a few hundred trees across a variety of locations and examined their ring patterns you will begin to appreciate how site selection is an important part of this process. After a few thousand trees you may actually learn to like it.

    And to Lubos I am certain that your APT committee will be very impressed by your Technorati rating.

  4. #4 Eli Rabett
    June 26, 2006

    Well, z, climate data is one of those areas where you can’t control the system, so you do need sophisticated statistics, string theory, on the other hand, has no data. Most physical sciences are somewhere in the middle.

    When I was in grad school, we had a killer comp question: Explain the universe, give two examples, little did I think that my fellow students would take that seriously enough to come up with 10^somegodawfulnumber universes. If we look hard enough we may even find the one with Ringworld in it, although it is a mighty strange part of the landscape.

  5. #5 per
    June 26, 2006

    “Just curious, per, but could you provide more information showing that there is a completely random population with no response to temperature? ”

    Dear Davis

    I am not making that case; I am simply setting out the difficulties in showing that you have a tree group that has a linear response to temperature.

    If you set out with a wholly random group, some will go up, some will go down, some will stay the same. If you select only the trees which are going up, and then argue that they are temperature sensitive, there is a circularity in your argument.

    It is well known that some trees in some positions are not responsive to temperature. What i didn’t appreciate is that when some researchers look for “temperature-sensitive” tree sites, they don’t bother archiving the data for the non-temperature sensitive sites.

    While it is a nice and intuitive idea that some tree sites are temperature sensitive, it is kind of important to test this idea. The NRC report has extensive discussion of “the divergence” problem, whereby many tree sites fail to show the increased temperature seen in thermometer records post-1980. This calls into question the use of trees as a temperature record.

    cheers,
    per

  6. #6 Hans Erren
    June 26, 2006

    Too much credit for Ken Ring to give Larry Niven’s creation to him.

  7. #7 per
    June 26, 2006

    “At a good site for dendroclimatology there will individual trees with a high sensitivity, a strong correlation between the trees growth patterns within a stand and a chronology that has a significant correlation with a climate variable(s). ”

    dear jp

    it seems you don’t understand what I am saying. You are talking about selection post-hoc.

    It sounds to me like that are stands where there is poor correlation between trees for growth; where there isn’t a significant correlation with a climate variable. It sounds in other words like a series of samplings of random data, where you discard the “bad ones”, and end up “finding” one with a good response to temperature.

    cheers

    per

  8. #8 jp
    June 26, 2006

    Per

    In order to archive the data you first need to travel to the site, sample (core) the trees, store and transport the cores back to the lab, mount and surface the cores, measure the ring widths, identify missing rings, cross-date different cores, reject and accept cores for inclusion in the chronology (quality control), construct the chronology and then begin the statistical analyses. This process is time consuming and expensive. When you are in the field and you are sampling trees if you see the ring series are complacent, you save your money and time and you move on to another site.

    That said, if you are looking for tree-ring chronologies that are likely to have less sensitivity there are many chronologies available from the ITRDB that have a poor climatic signal. Alternatively you could search for chronologies that were gathered for other purposes (e.g. dating control of geological events such as floods and mass movements) or sample your own.

    The NRC report has identified what has been known in the paleoclimatic community for a long time, it is a difficult business to extract a climate record from a tree-ring chronology. That is why it is but one piece in the paleoclimate puzzle.

  9. #9 Eli Rabett
    June 26, 2006

    Oh I don’t know Hans, I rather think of Ringworld as an imaginary place for picaresque journeys through a fantasy landscape full of magic, dangers and grotesque creatures. A good fit for our Ken.

  10. #10 Dano
    June 26, 2006

    The NRC report has identified what has been known in the paleoclimatic community for a long time, it is a difficult business to extract a climate record from a tree-ring chronology. That is why it is but one piece in the paleoclimate puzzle.

    Hence multiproxy chronologies.

    It’s a good idea, which is why MBH98 was totemized, in order to kill the good idea. Can’t have a free market of ideas, you know.

    Per knows, jp, what the coring and dating procedure is. He’s just dissembling, obfuscating, and purveying his brand of FUD (nobody here is buying).

    Best,

    D

  11. #11 per
    June 26, 2006

    “When you are in the field and you are sampling trees if you see the ring series are complacent, you save your money and time and you move on to another site.”

    like I said, post-hoc selection of the data that “fits” what you want it to.

    “The NRC report has identified what has been known in the paleoclimatic community for a long time, it is a difficult business to extract a climate record from a tree-ring chronology.”

    This has been discussed in some detail at climateaudit. The NAS report puts further severe constraints on the use of trees as a valid proxy for temperature.

    toodle-pip !

    per

  12. #12 per
    June 26, 2006

    Dano-“It’s a good idea, which is why MBH98 was totemized…”

    Unfortunately for you, it was the IPCC who totemised MBH’98. They got Mann to head up the section which lionised his own work. It was put “front and centre” of the publicity, and the summary for policymakers of the TAR.

    That is why it is so unfortunate that this paper had major errors in its methods and data; it features unusual PCA which the NAS panel censured; they calculated adverse statistical measures, and simply withheld the information; their statistical analysis was censured by the NAS panel; and they made statements about the robustness of their reconstruction to the absence of tree proxies, which do not hold up.

    Yes, Dano, science is a place where it is essential to have free access to methodology; MBH have refused this, and it is easy to see why. And still you support them.

    bemused, per

  13. #13 jp
    June 26, 2006

    Per

    Well I gave you chance but that refusal to learn is just engrained. Here is a demonstration of your logic. A pharmacutical company is funding a drug trial for a disease that affects a small percentage of the population. Doctor A suggests they administer the drug to a ramdomly selected sample of the population to test its efficacy. Doctor B suggests it would be more effective to stratify the sample and actually test the drug on a random sample of the population that is afflicted or at high risk of the disease. What would the smart company do? What would Per do?

    Following the trial, the company announces great success. However, an immediate audit is ordered. Per declares the results are invalid because they have found what they we looking for!

  14. #14 Eli Rabett
    June 26, 2006

    per might want to talk with Dr. Lott.

  15. #15 Dano
    June 26, 2006

    like I said, post-hoc selection of the data that “fits” what you want it to.

    I’d be happy to sign you in, David, to a dendro listserv so you can argue your weak bullsh*t…er…heroic Lomborgian overthrowing of a discipline you haven’t studied++ there. Let me know. I’ll be on vacation next week.

    Best,

    D

    ++ A strike tag is so much more appealing here…

  16. #16 per
    June 26, 2006

    dear jp

    are you really a practicing scientist ? Do you really fail to understand what post-hoc data selection is, and why it is so very problematic ?

    You are selecting only those trees which show a correlation with temperature. You are then claiming that they are temperature-sensitive- but you have already selected them on that basis ! The temperature sensitivity is a circular argument.

    To be quite clear, if any pharma company submitted a study where they did post-hoc data selection, they might well be liable for criminal prosecution.

  17. #17 jp
    June 26, 2006

    Per

    You have demonstrated: (a) you have no knowledge of paleoclimate, (b) you cannot read, and (c) your time has past.

    The world moves on, the audit fades to black….

  18. #18 per
    June 26, 2006

    “(b) you cannot read”

    ? you are advocating post-hoc data selection; that is what you wrote. I explained exactly why that is problematic.

    You have not engaged with that argument at all, nor attempted to explain why it does not apply. All i get is nebulous smears; and they have to be pretty nebulous, because your specific claims have a habit of coming unstuck.

    oh well, it is Doltoid…

    per !

  19. #19 Dano
    June 26, 2006

    Do you really fail to understand what post-hoc data selection is, and why it is so very problematic ?

    I’d be happy to sign you in, lil’ per, to a dendro listserv so you can argue your Lomborgian overthrowing of a discipline you haven’t studied there.

    Let me know. I’ll be on vacation next week.

    Best,

    D

  20. #20 Hans Erren
    June 27, 2006

    Where are you going? I can recommend the Gaspe peninsula in Canada, you can help to find the lost site!

  21. #21 per
    June 27, 2006

    “I’d be happy to sign you in,…”

    see Dano, that’s the thing about science; it is specific and falsifiable. You could have leveraged your (no doubt) phenomenal knowledge in botany, and cited a reference, given a specific example; that would be specific and falsifiable.

    But instead, we get a vapid proclamation from Mount Dano, telling us that you know it all, and there is a listserv where I may humbly learn, before prostrating myself at your feet. Specific ? No. Falsifiable ? No.

    Obfuscatory ? Yes. And who is the lil’ bot who just loves to carp on about “obfuscatory, mendacisationising” posts? Step forward Dano.

    forever in your gratitude

    per

  22. #22 Kevin Donoghue
    June 27, 2006

    Dano, I think that was per’s idea of a polite refusal. I wonder would per’s line of argument work with a traffic cop?

    “I really wasn’t speeding officer.”

    “Radar says you were, sir.”

    “But that’s a speed-sensitive instrument! You’re trying to measure speed with a speed-sensitive instrument! That’s circular reasoning.”

    “Uh, you mean if I just looked at, say, a lamp-post or something…”

    “…You wouldn’t have any evidence that I was moving at all! Exactly! Toodle-pip Officer!”

  23. #23 Nabakov
    June 27, 2006

    “…science-oriented articles”

    Full of truthiness no doubt.

  24. #24 Hans Erren
    June 27, 2006

    Dano’s tricks:

    The answer is:
    “On the internet”
    “somewhere in this 300+ page document”(“linky”)
    “in this listserv”
    “in your local university library”

    Dano never quotes, he always refers.

    …And then he is off.

  25. #25 per
    June 27, 2006

    Hans- spot on !

    Kevin Donoghue:

    it is clear that I can have no argument with any proxy which robustly measures temperature; but what is at issue is whether the proxy does measure temperature.

    So here is a question; how do you demonstrate (prove) that trees measure temperature ? Tell me how you do the experiment.

    If you can look at the environment of specific trees, identify the site as temperature sensitive, and then lo-and-behold ! you have a testable hypothesis; you test the trees you specify, and the hypothesis stands or fails. This is not what is done.

    Instead, you take trees, correlate them with temperature, throw away the results you don’t like (the ones which don’t change, or which go down), and you are left with the ones which go up (are “temperature sensitive”).

    Can you really not see why this is problematic ?

    yours

    per

  26. #26 Chris O'Neill
    June 27, 2006

    “Where are you going? I can recommend the Gaspe peninsula in Canada, you can help to find the lost site!”

    I don’t know why Hans Erren has any interest in trying to suggest that global warming isn’t happening because he thinks we shouldn’t care what happens after we all die.

  27. #27 Dano
    June 27, 2006

    see Dano, that’s the thing about science; it is specific and falsifiable. You could have leveraged your…etc.

    Shorter per:

    tap dance away from being called on my bullsh*t, quick!

    Dano never quotes, he always refers.

    This from the guy who forced me to coin the phrase ‘ploppedness’. Gimme a f’n break.

    The fact is, folks, that the bots are energized by some FUD phrases that somebody made for them.

    They’ll lose their energy soon and all can go on. Let us note that their attempt to hijack the public dialogue is not working. Look at the headlines, lil’ quibbling bots.

    Best,

    D

  28. #28 Ian Gould
    June 27, 2006

    “Too much credit for Ken Ring to give Larry Niven’s creation to him.”

    Inventive as Dr. Niven is, I doubt he could ever have come up with a planet wth an orbital velocity greater than the speed of light. (Ring-Earth, after all, is one light year from the sun but still manages one revolution per year).

  29. #29 Kevin Donoghue
    June 27, 2006

    per: Instead, you take trees, correlate them with temperature, throw away the results you don’t like (the ones which don’t change, or which go down), and you are left with the ones which go up (are “temperature sensitive”).

    No, that’s not how it is done, at least not according to jp’s account above. The aim is to identify trees which show high variance. That makes sense whether the planet is getting warmer or colder. If as you say researchers simply “throw away the results they don’t like” that’s just fraud. Sure scientists cheat from time to time but if you really think it is common practice than you may as well dismiss all research that you haven’t done yourself.

  30. #30 jp
    June 27, 2006

    Kevin

    As Per has clearly demonstrated, he has no interest in learning only in disinformation. The assessment by Dano (and others) of Pers (David Bells) character is accurate. I suggest we starve the troll.

  31. #31 per
    June 27, 2006

    kevin, jp, this is a direct cite from jp.

    “So how “do you show that there is a sensitivity to temperature”, you measure it by constructing a chronology from a group of trees at a site and examining the correlation between the ring series and a range of climate variables. At a good site for dendroclimatology there will individual trees with a high sensitivity, a strong correlation between the trees growth patterns within a stand and a chronology that has a significant correlation with a climate variable(s). ”

    as far as I can see, it doesn’t say that the only cut is pre-data, and selection for high variance. It specifically defines a good site as one with a significant correlation with a climate variable; this is post-hoc analysis. By definition, sites without that correlation aren’t good.

    If you think that bad scientific practice is uncommon, or isn’t engaged in throughout the scientific enterprise, you are wrong. Scientists make mistakes all the time; the beauty of the scientific enterprise is in making your procedures open, free and honest discussion of what you do, and replication.

    Just for instance:

    http://www.cs.princeton.edu/~ken/Langmuir/langmuir.htm

    It is amusing to see who is running from free and open discussion.

    per !

  32. #32 per
    June 27, 2006

    “As Per has clearly demonstrated, he has no interest in learning…”

    well, jp, I am prepared to learn. If your analysis is not post-hoc, explain to me why.

    But perhaps we can agree; that what you have described is post-hoc analysis.

    yours

    per

  33. #33 Kevin Donoghue
    June 27, 2006

    per: It specifically defines a good site as one with a significant correlation with a climate variable; this is post-hoc analysis.

    No it isn’t. There is normally a strong correlation between the speed of a car and the speedometer reading. If there isn’t, it certainly isn’t a good speedometer.

  34. #34 Hans Erren
    June 27, 2006

    See Dano that’s your problem, showing your credentials will reveal your identity.

    Keep on swearing, I am not impressed,

  35. #35 JJ
    June 27, 2006

    Having perused a number of AGW sites, pro and con, I am struck by the tremendous difference in quality of the argumentation presented by each side.

    The skeptics are, by and large, wont to focus on the science, and specific critique thereof. The proponents on the other hand, most frequently offer little more than ad hominem in its various and asundry forms. This thread is no exception.

    Do they no longer teach fundamentals of formal logic in schools? Has no one heard that ‘Because a climate scientist says so’ is not a valid argument? That ‘who are you to ask that question’ is not an answer to that question?

  36. #36 per
    June 27, 2006

    Dear Kevin

    you just said “no it isn’t”. Here is the direct cite from jp.

    “At a good site for dendroclimatology there will individual trees with a high sensitivity, a strong correlation between the trees growth patterns within a stand and a chronology that has a significant correlation with a climate variable(s). ”

    I have simply paraphrased jp; if you can explain where I have misunderstood, I will be grateful. I have gone by what jp said; if you can explain that the reality is entirely different, that is fair enough.

    But this is post-hoc; it is only once you have excluded the sites with a bad correlation, that you are left with the sites with a good correlation to temperature. Entirely post-hoc.

    what am I missing ?

    yours, per

  37. #37 Don Baccus
    June 27, 2006

    The skeptics are, by and large, wont to focus on the science, and specific critique thereof. The proponents on the other hand, most frequently offer little more than ad hominem in its various and asundry forms. This thread is no exception.

    This is hilarious given that skeptics continuously accuse the community of climatologists of being guilty of scientific fraud. They say this in the press, to Congressional committees, etc etc.

    If you want science, go read Real Climate’s excellent summary papers on what’s known about global warming. These are written by some of the leading climate experts in the world.

    However, my guess is that you’re just trolling and could care less about the state of the science…

  38. #38 JJ
    June 27, 2006

    Don Baccus,

    “This is hilarious given that skeptics continuously accuse the community of climatologists of being guilty of scientific fraud.”

    That is an example of ad hominem tu quoque.

    “If you want science, go read Real Climate’s excellent summary papers on what’s known about global warming. These are written by some of the leading climate experts in the world.”

    This is an example of a variation of ad hominem called ad verecundiam.

    Why the apparently irresistable urge to include fallacious reference to ‘leading climate experts’? Who cares if they consider themselves ‘leading’ whatever? Leaders are often dead wrong. And frankly, in a field as broad as ‘climate science’, nobody is a ‘leader’ in every aspect of his own work, much less in the field as a whole. Appeal to authority is fallacious. Are their methods and conclusions correct? That is what matters.

    “However, my guess is that you’re just trolling and could care less about the state of the science…”

    And that is just pure ‘ol ad hominem.

    Thanks for demonstrating exactly what I’m talking about.

    JJ

  39. #39 Carl Christensen
    June 27, 2006

    you can determine leaders; the ones with the peer-reviewed publications, which whittles out a lot of the skeptics

  40. #40 JJ
    June 27, 2006

    So, I post an observation that the majority of the pro-AGW arguments are little more than ad hominem, and the first two ‘contrary’ responses are 100% ad hominem.

    Who will make it three? Step right up and throw another irrelevant fallacy on the pile! Dont miss out on your opportunity to be among the first … i.e. a ‘leader’.

    LOL.

  41. #41 MarkR
    June 27, 2006

    From Von Storch, Zorita and Gonzalez-Raucen

    4) With respect to methods, the committee is showing reservations concerning the methodology of Mann et al.. The committee notes explicitly on pages 91 and 111 that the method has no validation (CE) skill significantly different from zero. In the past, however, it has always been claimed that the method has a significant nonzero validation skill. Methods without a validation skill are usually considered USELESS.

  42. #42 Carl Christensen
    June 28, 2006

    It’s an “ad hominem” attack to point out that in science, the papers in peer-reviewed journals shows the validity of the research? Also, you are whining that everything is an ad hom attack, even someone pointing you to papers on realclimate.org. How silly.

    PS — I agree with the von Storch analysis, too bad the Mc&Mc are just grasping, reedy twits (yes — that’s an ad hom :-)

  43. #43 Chris O'Neill
    June 28, 2006

    According to Lubos:

    “Sorry to say, but the answer is .. my higher intelligence and physics background than 2480 of these 2500 people.”

    Would that be the same higher intelligence that enabled you to not realize that “Predict your Climate” relies on any artificially chosen mean value for the forecast temperature whereas the hockeystick reconstruction doesn’t? What an enabling higher intelligence you have.

    “and former mineral consultants (who was the #1 mathematician in Canada, by the way)”

    Poor Canada.

  44. #44 Ken Miles
    June 28, 2006

    Having perused a number of AGW sites, pro and con, I am struck by the tremendous difference in quality of the argumentation presented by each side.

    The skeptics are, by and large, wont to focus on the science, and specific critique thereof. The proponents on the other hand, most frequently offer little more than ad hominem in its various and asundry forms. This thread is no exception.

    If that’s your perception, your either ignorant, blind, stupid or stoned.

    The best place to go for the science is the scientific literature. This strongly supports the global warmers.

    The next best place to go for a discussion of the science is the IPCC reports. These strongly support the global warmers.

  45. #45 Chris O'Neill
    June 28, 2006

    According to Lubos:

    “the statistical significance of all results before 1600 is zero …. the panel has announced that all dendro conclusions before 1100 are statistically worthless”

    Please try to be consistent within the one paragraph Lubos. It may affect your credibility which is very impotant to all of us.

  46. #46 JJ
    June 28, 2006

    CC,

    “It’s an ‘ad hominem’ attack to point out that in science, the papers in peer-reviewed journals shows the validity of the research?”

    A. That is not what you did. Your comment was nothing more than an ad hominem against opposing researchers, not a scientific rebuttal of any research. And yes, attacking the researcher is ad hominem.

    B. Yes, swinging ‘peer review’ around like a trump card is also fallacious. That research appears in a peer reviewed journal does not ‘validate’ it. Note that one of the fundamental functions of peer reviewed journal articles is to refute previously published peer reviewed journal articles. Similarly, that research does not appear in a peer reviewed journal does not invalidate it.

    Peer review, performed properly (and often it is not), is a useful convenience, not a magical truth filter. When properly performed, the peer review process has inevitable, demonstrable, and unquantified Type I and Type II error. Often, it is also susceptible to bias. It cannot be legitmately used to summarily dismiss a postion, much less a person (or class of persons, as you attempted).

    “Also, you are whining that everything is an ad hom attack,”

    No, I am not. Nice hyperbole, though. Complements the straw man you create next:

    “… even someone pointing you to papers on realclimate.org.”

    Note that I identified the ad hominem in that portion of the post as the inclusion of the irrelevant ‘leading climate experts’ crap, not the offered reference to the RC website. Intentional misstatement of the other’s position is yet another fine example of the generally low quality of the argumentation dished up by the AGW proponents. Thanks.

    Earlier on this thread, you made several ‘appeal to authority’ fallacious comments. Then you chastised Lubos for doing it. Then you did it some more. Then you chastised M&M for *not* commenting outside their field of knowledge. And then you called *other* people ‘pseudo-intellectual twits’. Interesting chain of events, that.

    JJ

  47. #47 Kevin Donoghue
    June 28, 2006

    per,

    You assert that you are merely paraphrasing jp’s comment, but jp clearly doesn’t agree. Don’t you find it little awkward to have the original author rejecting your paraphrase? Any reader who is interested can compare his account with yours and confirm that they just don’t match. Nowhere does jp say anything to underpin this (your) description:

    Instead, you take trees, correlate them with temperature, throw away the results you don’t like (the ones which don’t change, or which go down), and you are left with the ones which go up (are “temperature sensitive”).

    jp defines sensitivity differently from you:

    …in dendrochronology the term ‘Sensitivity’ has a specific meaning, it is a measure of the variance in a tree ring series. When a individual tree (or chronology) shows a high amount of variance it is called Sensitive, when a series shows little annual variability it is referred to as Complacent. [my emphasis]

    jp describes a procedure which, diligently followed, gives the data a fair chance to show a falling trend in temperatures. You describe a procedure which can give only one result.

    You’re not paraphrasing.

    Now, you ask what it is you are missing. That’s never an easy question to answer, even when one knows a great deal about the person who is asking. But I think you provide a clue when you say: “it is only once you have excluded the sites with a bad correlation, that you are left with the sites with a good correlation to temperature.” My best guess is that you are inferring that a site with a good correlation to temperature won’t provide evidence of falling temperatures. That’s not a valid inference. Another possibility is that you have read about the pitfalls of extrapolation and you believe that researchers have indeed tumbled. Maybe they have, but jp’s account doesn’t suggest that. Unless you explain your theory in more detail I don’t think there is much point in my trying to offer a critique.

  48. #48 per
    June 28, 2006

    kevin

    I accept that sensitivity/ complacency refers to the variability in ring width/whatever; tell me what phrase you wish to use to convey that there is a correlation between tree growth and temperature if you find “temperature sensitivity” ambiguous, and I can use it. I have quoted directly from jp, and I have made clear this is the bit which I am paraphrasing;

    “So how “do you show that there is a sensitivity to temperature”, you measure it by constructing a chronology from a group of trees at a site and examining the correlation between the ring series and a range of climate variables. At a good site for dendroclimatology there will individual trees with a high sensitivity, a strong correlation between the trees growth patterns within a stand and a chronology that has a significant correlation with a climate variable(s). ”

    These are jp’s words. At good sites, there is a significant correlation with temperature. So what sort of site gives a poor correlation with temperature ? Does this description not imply the presence of some sites where there is a lack of correlation with a climate variable, or poor correlation between tree growth patterns ?

    jp has described what appears to be post-hoc analysis. If his description is wrong, or if there is a different process involved, I am sure jp can tell us. Likewise, if JP has a tantrum, and throws his toys out of the pram simply because there is a legitimate criticism of what he has said, then that does not say much for his ability to explain his research.

    yours

    per

  49. #49 Carl Christensen
    June 28, 2006

    err, JJ, science by peer-review in journals seems to be the best bet. It was good enough for Einstein, it’s good enough for everyone. Jeez, can one imagine Einstein trying to get through the noise of a blog? HAHA. And having Steve McIntyre & his right-wing lackeys put him down ad infinitum for a simple mistake he may have made. Sorry, but blogs are good for entertainment; leave the science to peer-reviews not reviews by “nanny_govt_sucks” & that ilk.

  50. #50 Hans Erren
    June 28, 2006

    Einstein wasn’t peer reviewed. (‘Annalen der Physik’ had only editors in the early years)

  51. #51 per
    June 28, 2006

    well said, JJ:

    “Earlier on this thread, you made several ‘appeal to authority’ fallacious comments. Then you chastised Lubos for doing it. Then you did it some more. Then you chastised M&M for not commenting outside their field of knowledge. And then you called other people ‘pseudo-intellectual twits’. Interesting chain of events, that.”

    CC has managed “twits”, “right-wing lackeys”, “liar”, so he gets an A+ for abuse; but it is amazing how short he is on science and logic.

    Then he comes onto a blog to make comments, and then starts complaining that you can only discuss science in the peer-reviewed literature. Why then is he on a blog discussing a scientific point ?

    amusing

    per

  52. #52 Jeff Harvey
    June 28, 2006

    I’ve been away a couple of days doing research. In that time I have had a major rethink about Per’s comments, and his uncanny ability to show how much of an ability he has to discuss scientific issues across a range of disciplines with immense depth and clarity. His support for Bjorn Lomborg has impressed me too, and its finally dawned on me that university study is not necessary to attain a complete spectrum of wisdom in a suite of fields far removed from your own professional area. Per, I yield to you. Lubos should also realize that he could have saved about 10 years or more of university time and a lot of money when studying string theory; why waste this time if all you need is to be infinitely intelligent (like Per) and thus dispense with the characters of all of those with whom you disagree?

    Per, you are therefore correct in everything you say. I defer to you. During my BSc programme at Liverpool University and thereafter in three post-docs and in my current position as a senior scientist I have studied and researched population ecology, but now I know that Per is an expert in this broad discpline. He can discuss, in detail, any of the following: Nicholson-Bailey, Thompson and Lotka-Volterra models; functional responses types I-III; k-factor analysis; factors affecting the evolution of semel and iteroparous reproductive strategies, including the r- and K selection continuum; factors promoting species co-existence within defined feeding guilds; the Terborgh-Soule models of exponential decay in predicting extinction rates; a suite of hypotheses, including (in the invasive species literature), the enemy release, evolution of increased competitive ability and novel weapons hyopotheses; in community ecology the enemy-free space hypothesis and the diversity-stability hypothesis; the role of ecophysiological constraints in studying multitrophic interactions and community structure and function; neutral models in determining niche breadth the outcome of interspecific competition; linking above- and belowground community level processes via primary and secondary plant metabolites; constitutive and induced defences in plants and animals; the importance of biotic and abiotic forces in successional gradients et al.

    I expect that Bjorn Lomborg is smilarly qualified, based on the fact that he studied political science at Aarhus University. Although its likely that his study covered very few or none of the areas I mentioned above, Per’s theory is still probably correct: being smart means you can master everything with no need for persky university study. So my advice to Tim, Carl, Dano, Eli, and others is this: Per’s widsom knows no boundries. I can’t wait to discuss the invasive ecology literature with him, and then swith to an in-depth discussion of endocrinological factors regulating ontogeny in parasitic organisms, perhaps finishing up on top-down and bottom-up regulation in ecological communities, and the role of structural heterogeneity in generating stability. I am sure that he has read all of the relevant literature by Elton, Gleason, Hutchinson, Hairston and others. On top of that, I’d be simply delighted to know what he thinks about the Lack optimal clutch size model.

  53. #53 z
    June 28, 2006

    “I post an observation that the majority of the pro-AGW arguments are little more than ad hominem”

    Perhaps that would appear more scientific and less ad hominem if you accompanied it with a statistical analysis of the pro and anti AGW sites.

  54. #54 z
    June 28, 2006

    “and former mineral consultants (who was the #1 mathematician in Canada, by the way)”

    I myself am the best statistician in the world; however no other statistician is sufficiently sophisticated to understand the methodology which brought me to this conclusion.

  55. #55 JJ
    June 28, 2006

    CC,

    “err,”

    An unusually appropriate word with which to begin your response. I commend your choice. :)

    “JJ, science by peer-review in journals seems to be the best bet.”

    Science by betting is a contradiction in terms. Peer review *is* a bet, and one with unknown odds. It is thus a useful convenience, but it is not definitive. And more to the point, it is certainly not an excuse for summarily dismissing the positions of anyone.

    “It was good enough for Einstein, it’s good enough for everyone.”

    More fallacious reasoning. Appeal to authority. Appeal to tradition. Both illogical – not scientific.

    “Jeez, can one imagine Einstein trying to get through the noise of a blog?”

    Yeah. I can imagine what guys like you would have said about him, when he came up with General Relativity:

    *This guy is a friggin postal clerk. He thinks we should throw out the esteemed work of ISAAC NEWTON, whose theories of physics have been supported by hundreds of derivative works in peer reviewed journals? That bozo-haired clown should stick to selling stamps and let the REAL Physicists tell us what is what. HAHA.*

    But now, you invoke his name on the other side of that particualr fallacy. Haha, I guess.

    “And having Steve McIntyre & his right-wing lackeys put him down ad infinitum for a simple mistake he may have made.”

    The mistakes Mann made were not simple. They were fairly involved, such that it took someone with specific knowledge and experience in the area of statistics to point it out. Mistakes that made it into peer reviewed journals more than once. And the political diatribes you persist with are more ad hominem.

    “Sorry, but blogs are good for entertainment;”

    Other blogs have much higher science content. This one is mostly political polemics, pretending to be science. Belies the promise of the URL. If only this were http://www.political-hacks-pretending-to-be-scientific-blogs.com

    “… leave the science to peer-reviews …”

    The content of some other blogs is at least as reliable as most peer review. The content of most AGW proponents’ posts is less so.

    “…not reviews by ‘nannygovtsucks’ & that ilk.”

    and Carl Christensen and his? We leave them to what?

    JJ

  56. #56 Don Baccus
    June 28, 2006

    JJ: “The content of some other blogs is at least as reliable as most peer review.”

    This boy’s going to single-handedly revolutionize science…

  57. #57 Kevin Donoghue
    June 28, 2006

    per: tell me what phrase you wish to use to convey that there is a correlation between tree growth and temperature

    I see nothing wrong with referring to it as the correlation between tree growth and temperature. If that’s too long-winded I guess you could call it “the correlation” or Corr(gro, temp) if it’s unclear which variables you have in mind.

    jp has described what appears to be post-hoc analysis.

    It doesn’t appear so to me. Sure there are sites where the correlation is poor. They will provide no useful information, just as the traffic cop will get no useful information about the speed of my car by studying the lamp-post. That’s why he uses a radar speed-gun, with a very high Corr(speed, reading). Consider what might be done if there is doubt about the reliability of a batch of radar systems. They could be tested by comparing their readings with a sample of cars whose speed is calculated using a stop-watch and two landmarks a suitable distance apart. Any radar guns showing a low Corr(speed, reading) would be returned to the manufacturer and any data gathered using them would be deemed unfit for use as evidence. The procedure described by jp is analogous.

    Selecting appropriate proxies for temperature by studying the data is not post-hoc analysis. It is what any sensible researcher would do.

  58. #58 per
    June 28, 2006

    “His support for Bjorn Lomborg has impressed me…”

    Dear Jeff, read again. Did I support BL, or did I merely point out that a particular selective quotation was extremely misleading, and that he was exonerated by the DCSD ? The latter is a fact, and I can scarcely imagine that you would wish to support deception about BL ?

    “During my BSc programme at Liverpool University and thereafter in three post-docs and in my current position as a senior scientist… ”
    well, in all that time, it is a shame you didn’t learn how to argue coherently. Blowing your own trumpet doesn’t advance your case. Insulting other people doesn’t advance your case.

    As you know, I have made no specialist claim to knowledge. I welcome better knowledge, and strive to learn. I am prepared to ask questions.

    If you cannot answer simple and relevant questions, don’t take it out on me.

    toodle-pip !
    per

  59. #59 Dano
    June 28, 2006

    The content of some other blogs is at least as reliable as most peer review. The content of most AGW proponents’ posts is less so.

    Suuuuuuuure.

    Sure pal.

    Best to commence ignorage.++

    Best,

    D

    ++ Let me guess: that’s ad hom or not scientific or some fallacy, huh?

  60. #60 per
    June 28, 2006

    dear kevin

    the trees with high correlation are defined as such only after you have done the analysis; this is post-hoc by definition.

    “Any radar guns showing a low Corr(speed, reading) would be returned to the manufacturer and any data gathered using them would be deemed unfit for use as evidence. The procedure described by jp is analogous.
    Selecting appropriate proxies for temperature by studying the data is not post-hoc analysis. It is what any sensible researcher would do.”

    Ok, here is a question for you to see if you are a reasonable researcher. You have got four stands of trees from the same area, and which have been suggested by prior considerations to show correlation with temperature, and you core samples for each of the four. If three stands show no (or an inverse) correlation with the temperature record, and one shows a reasonable correlation, can you simply use the one stand as showing a correlation with temperature, and discard the other three ? Have you then shown that this stand of these trees are sensitive to temperature, and record the temperature ?

    I look forward to your reply.

    In the meantime, I note that thermometers, and speed guns, can be repetitively tested, and calibrated. There is only one temperature record for a given stand of trees.

    There is a reasonable belief that if you assemble the bits required for a speed camera/ thermometer, then given appropriate quality control criteria (such as everything screwed in, etc), that a speed camera or thermometer will perform as expected, and that a very high percentage (90+%) will function as a speed camera or thermometer. With trees, there is apparently no way of identifying those stands which will have a temperature response; the % that will show correlation with temperature is unknown, and almost certainly biased by the tree stands that are simply not archived or analysed (strangely, complacency seems an appropriate choice of word here).

    I do not think your analogy to thermometers/ speed cameras is particularly insightful.

    yours, per

  61. #61 Dano
    June 28, 2006

    Ok, here is a question for you to see if you are a reasonable researcher…[h]ave you then shown that this stand of these trees are sensitive to temperature, and record the temperature ?

    I’d be happy to sign you in, David, to a dendro listserv so you can try to argue your weak FUD there.

    Let me know. I’ll be on vacation next week.

    Best,

    D

  62. #62 per
    June 28, 2006

    “Perhaps you had better tell us again how wonderful you are ?”

    just in case you hadn’t noticed, jeff, I posted this twice before, and you still told us what a senior scientist you were.

    Such a very senior scientist, and yet; you got the words wrong from the NAS report. Such a very senior scientist; yet you are not prepared to argue your contention that MBH is a sparkling example of peer-review.

    per

  63. #63 per
    June 28, 2006

    Dano- “I’d be happy to …”

    do anything but answer the question ? A simple yes or no would suffice :-)

    “The answer is: “On the internet” “somewhere in this 300+ page document”(“linky”) “in this listserv” “in your local university library””

    Didn’t Hans just get it right ?

    yours, per

  64. #64 Dano
    June 28, 2006

    Shorter per:

    Hand-wave away from being called on my bullsh*t, quick!

    Best,

    D

  65. #65 JJ
    June 28, 2006

    Dano,

    “Let me guess: that’s ad hom or not scientific or some fallacy, huh?”

    No, that is not a fallacy. It is just chicken $#!^. Running away from a dispute instead of engaging on the merits. Very similar to how you are dealing with Per’s questions, and descriptive of your entire participation on this thread.

    Happy to sign him up for a Dendro listserv? If his ‘FUD’ is as weak as you claim (over your shoulder, while beating a hasty retreat), then why can you not effectively (read: scientifically) respond to it here? Is there something about the rational defense of your position on tree ring/temperature methodology that only works when you have people shaking pom-poms in the background?

    Or do you just not know enough about the subject to justify what you believe about the subject? Its OK to admit that. Nobody knows everything.

    JJ

  66. #66 Chris O'Neill
    June 28, 2006

    Lubos advised Jeff Harvey:

    “besides reading my postings, it may also be a good idea for you to read the report.”

    It may also be a good idea for Lubos to take his own advice.

  67. #67 Kevin Donoghue
    June 28, 2006

    per,

    Your question suggests to me a possible source of confusion. What we need to keep in mind here is the purpose of the research. Your objection seems to be predicated on the assumption that what we are discussing here is hypothesis testing as taught in introductory statistics courses. If for example a researcher is testing the hypothesis that a particular species of tree is sensitive to temperature, or that its sensitivity depends on a particular set of variables, then of course it would be inexcusable to make lots of observations and throw out the ones that don’t fit.

    That’s not the case being discussed here. I refer you to jp’s explanation:

    If the objective of the research is to extract a climate signal from a tree ring series it would be of no value to sample trees that were not sensitive to climate.

    What we are talking about here is the construction of a proxy variable – a common enough task in science. In that context jp’s approach seems sensible to me. Obviously you disagree but you haven’t explained why.

    …the % [of trees] that will show correlation with temperature is unknown, and almost certainly biased by the tree stands that are simply not archived or analysed….

    Here you are saying that a variable is both unknown and probably biased. That makes no sense at all. Perhaps you mean that the true correlation is unknown – as true correlations invariably are – while the estimates are biased. For all I know that may be true, but you shouldn’t complain about jp losing patience with you if you won’t take the trouble to present a coherent case. In any event it is only trees which actually do show a correlation with temperature which are of interest. Whether they are common or not isn’t especially relevant.

  68. #68 JJ
    June 28, 2006

    Kevin,

    “Obviously you disagree but you haven’t explained why.”

    He has, but you havent grasped it yet. It is a bit subtle.

    “In any event it is only trees which actually do show a correlation with temperature which are of interest.”

    True, but incomplete. It is not trees which show a correlation to temp that are of interest. It is the trees that show a response to temp that are of interest. The latter being a subset of the former.

    Correlation is not sufficient. There must be causation. Per’s question boils down to: How does this apparent post-hoc selection method (which certainly will find correlation, as it preferentially tosses everything else) identify causation?

    WRT the counterexamples that you provided (speed guns, etc) the answer is that such things are tested. One might test trees for response in the same manner, but that does not appear to have been done. Or has it? That is the question.

    JJ

  69. #69 per
    June 28, 2006

    dear kevin

    my congratulations on maintaining a civil dialogue.

    You are of course correct when saying that I cannot say that something is both unknown and biased. I will stick with unknown, but it pains me to see so many people flagrantly burying/ not analysing perfectly good data, and that is a bias.

    JJ’s point is good. The point of comparing growth records with temperature is not just to demonstrate a correlation, but then to suggest that this is causation, and that therefore, the whole of the growth record for the rest of the trees shows similar causation, and hence temperature. Given that you have tossed the non-responders, how do you know that you have a temperature response ?

    “If the objective of the research is to extract a climate signal from a tree ring series it would be of no value to sample trees that were not sensitive to climate.”

    That may be jp’s quote, but it presupposes that there is a climate signal there that can be extracted. If there is no such climate (or specifically, temperature) signal there, then excluding the non-responding stands simply gives you a biased sample.

    Let me be absolutely clear that I have minimal knowledge of botany, and even statistics. It may well be (as JJ says) that these questions have been answered in a significant way, and I would be delighted to find out about it; whatever the answer is.

    yours

    per

  70. #70 Tim Curtin
    June 28, 2006

    JJ and per: the following citation is one of a series of tree ring studies in Utah Arizona and Wyoming, that have some interest. Most of the sites studied (including a later study on Bighorn) show no or minimal correlation between tree growth and temperature, but very strong correlations between growth and drought/floods over the last 1,000 years (NAS please note). Unfortunately while there are times and places when larger precipitation and warmer temperatures coincide, at other times and sites they do not (eg Bighorn). Using the radar speed gun analogy, what we have here is a radar gun that does not match the MBH gun’s readings. Which one is defective? We have more reason to reject the MBH batch which post-NAS is valid only back to 1600 than to reject these with their impressive continuity back to 1000. The Jackson studies seem worthy of attention by Jeff and CC.
    NSF-DEB-9815500- FINAL REPORT, COLLABORATIVE RESEARCH, “LATE
    HOLOCENE EXPANSION OF UTAH JUNIPER IN WYOMING: A MODELING
    SYSTEM FOR STUDYING ECOLOGY OF NATURAL INVASIONS”,
    Steve Jackson (University of Wyoming) and Julio Betancourt (University of
    Arizona).

  71. #71 Kevin Donoghue
    June 29, 2006

    per / JJ,

    Certainly it is well to remember that correlation does not reveal causation. That warning is drummed into the head of every stats student and with good reason. But it is also important to be aware that it doesn’t always matter. For example, let’s imagine we find a very strong correlation between obesity and mortality. Is this because obesity is bad for the health or because those who are on their last legs don’t exercise? For a doctor the direction of causation is important; but for a life assurance company it doesn’t matter in the least. The correlation tells us that obesity is a good proxy for mortality risk and obesity is observable whereas (ex-ante) mortality risk is not.

    I think I now have a fair idea of what per’s concern is. At all events I will give it one last shot, since it relates to statistics not botany (of which I know nowt). Suppose we have a large table of numbers, consisting of hundreds or maybe thousands of columns. The first column represents temperature and contains 100 numbers corresponding to the years 1901-2000. Each of the other columns represents a tree and contains 200 numbers which may or may not track temperature closely for the period 1801-2000. The optimists assert that some unknown proportion of the columns are a good proxy for temperature, but the only way to identify them is by statistical methods. The table is all we have. (In practice, as Tim Curtin’s comment indicates, some can be ruled out on botanical grounds; but never mind that.) The pessimists assert that picking columns by calculating correlations will give spurious results. How are we to know who is right?

    Statisticians have a strong predisposition to regard data as worthless until there is strong evidence to the contrary. One simple approach would be to calculate all correlations with temperature for 1901-2000 and identify the best performers, then repeat the process for the two sub-periods 1901-1950 and 1951-2000. If the same trees mostly emerge as winners in both sub-periods that’s not likely to be a fluke. If there isn’t a goodly overlap between the two sets of winners then the verdict goes to the pessimists. Other tests can be used to compare the “temperatures” reported by the winning trees for the period 1801-1900. If their high correlations with 20th century temperatures are spurious it is likely that the “histories” they report for the 19th century will be wildly divergent.

    There are many tests which can be carried out in cases like this, some simple and some horribly complicated. If you want to know more then I can do no better than emulate Dano and suggest that you look in a good library. I quite liked The Analysis of Time Series by Christopher Chatfield, but “liked” is a relative term in this context.

    This thread has gone on long enough. Thanks for your comments.

    Kevin

  72. #72 per
    June 29, 2006

    Kevin

    I am less than clear that it is such a rigorous test to find a correlation over 1900-2000, and then test this by looking at the correlation over 1950-2000.

    Thermometers, and speed cameras, have a large signal to noise; a car will be moving at 60, then slow to zero, and the readout has to match that. For trees and temperature, we are talking about ~0.6C over a century. The biological effect which we are looking to measure must be small, and that must give even more difficulty when establishing a correlation with temperature.

    I note that it is common for tree records to give a high correlation for the test temperature period (e.g. 1900-2000), and a lower correlation for the verification period (e.g. 1850-1900). This is a situation which is consistent with your description of a “spurious” relationship with temperature.

    Finally, I fear that even the simplest aspect of this procedure is biased. Take for instance, the fact that temperatures have risen for the last century, and so we are wanting a change in temperature; so we discard the samples with little variance (complacent) and only analyse the samples with high variance. Just imagine if temperatures had been stable for the last century; no doubt, we would be discarding the samples with high variance (sensitive), and the complacent samples would be giving us a great correlation with the temperature record. This is nothing less than throwing away samples which give the “wrong” result.

    “In any event it is only trees which actually do show a correlation with temperature which are of interest. Whether they are common or not isn’t especially relevant.”

    I made the point clearly that we can expect from our knowledge of a thermometer/ speed camera, that given quality control, almost all of these devices will measure temperature or speed. So we know how the system works, and that explains why we can predict how these devices will respond.

    With trees, it appears that we cannot predict which trees/ stand will respond in advance. It looks like only a (very) small proportion of trees show a correlation with temperature, and it would follow that the vast majority show no such correlation. It looks like we do not know how these devices respond, and the idea that temperature correlation in a small % of trees proves temperature sensitivity looks rather speculative to me.

    toodle-pip !

    per

  73. #73 Chris O'Neill
    June 29, 2006

    Tim Curtin opines:

    “the MBH batch which post-NAS is valid only back to 1600″

    I see Tim has fallen for the misinterpretation of the press release tautology. The press release said “Less Confidence in Temperature Reconstructions Prior to 1600″. Anyone who has spent 5 minutes studying MBH99 would know that it shows a sudden increase in confidence limits before 1600. Repeating this in the press release sucks ignorant people into believing that it is news and gives denialists like Tim Curtin the opportunity to make blatantly wrong exaggerations.

    By the way Tim, how are you going with your high school chemistry? Have you got up to the section on molecular weight yet?

  74. #74 per
    June 29, 2006

    dear chris

    you are correct that mbh99 artificially inflates the 95% confidence intervals pre-1600 by an inadequately documented procedure. Nonetheless, the NAS panel has no confidence that this is an adequate procedure. They specifically state that the uncertainties prior to 1600 cannot even be quantified.

    That is a long way away from mbh 98/99, and it is a significant difference to what they say.

    cheers

    per

  75. #75 Kevin Donoghue
    June 29, 2006

    per,

    I wasn’t planning to respond further in this thread, but I really can’t let this pass. You wrote:

    Finally, I fear that even the simplest aspect of this procedure is biased. Take for instance, the fact that temperatures have risen for the last century, and so we are wanting a change in temperature; so we discard the samples with little variance (complacent) and only analyse the samples with high variance. Just imagine if temperatures had been stable for the last century; no doubt, we would be discarding the samples with high variance (sensitive), and the complacent samples would be giving us a great correlation with the temperature record.

    When you were typing that did you give even a moment’s thought to what the terms variance and correlation actually mean? I’m pretty sure you didn’t. To get in touch with these concepts, try putting the following values of three variables into a spreadsheet. Calculate the variances of T, U and V and the correlations Corr(T, U) and Corr(T, V).

    First, lets see what happens when the trend is upwards:

    T: 21, 24, 24, 26, 27, 29, 33
    U: 22, 24, 23, 26, 28, 31, 32
    V: 22, 20, 26, 27, 27, 32, 31

    Does U have greater variance than V or vice versa? Which is more highly cocorrelated with T?

    Now let’s rearrange the observations so that the trend is downwards:

    T: 33, 29, 27, 26, 24, 24, 21
    U: 32, 31, 28, 26, 24, 23, 22
    V: 31, 32, 27, 27, 20, 26, 22

    Is there any change in the situation as regards variance and correlation?

    Let’s try shuffling the observations so that they show no particular trend:

    T: 24, 33, 26, 27, 29, 24, 21
    U: 24, 32, 26, 28, 31, 23, 22
    V: 20, 31, 27, 27, 32, 26, 22

    Try using other numbers and ask yourself, has per been posting utterly nonsensical comments?

  76. #76 per
    June 29, 2006

    Dear Kevin

    on reflection you are correct. It is certainly possible to create a sample with rising trend and lesser variance than a sample with zero trend, and greater variance.

    Nonetheless, when it comes to tree ring samples, I suspect that this isn’t the case. I suspect that the complacent samples show very similar ring widths throughout the chronology, and very little variation. I suspect that the “sensitive” trees have by definition greater variation, and that some will trend up, and some trend down, as you would expect from any sample of random numbers. I suspect that the flat and declining tree ring series show poor correlation with temperature, and those that are rising show a good correlation with temperature.

    But let us accept that you are correct, and it is possible that the complacent series could have a rising or falling trend. Would you like to explain to me why jp discards such series when they could have a perfect correlation with temperature ?

    yours

    per

  77. #77 Dano
    June 29, 2006

    per mendacicizes for the nth time:

    It looks like only a (very) small proportion of trees show a correlation with temperature, and it would follow that the vast majority show no such correlation.

    I’d be happy to sign you in, David, to a dendro listserv so you can argue this weak bullsh*t there. Let me know. I’ll be on vacation next week.

    Best,

    D

  78. #78 MarkR
    June 29, 2006

    I think you are missing the point.

    The Bristlecone/California Basin proxies have very little correlation to temperature, and they (naturally) have very little correlation to any other temperature proxy.

    Take the Bristlecones out and you can’t get a hockey stick.

    Simple as that really.

  79. #79 per
    June 30, 2006

    Dano
    > per mendacicizes for the nth time:

    translation: I am going to call you names, because I have nothing sensible to contribute.

    > I’d be happy to sign you in, David, to a dendro listserv
    > so you can argue this weak bullsh*t there.

    Translation: I don’t have a clue, but it is really important that I pretend I know what I am talking about. I have nothing sensible to contribute.

    thanks for your contribution, D !

    per

  80. #80 Kevin Donoghue
    June 30, 2006

    But let us accept that you are correct, and it is possible that the complacent series could have a rising or falling trend. Would you like to explain to me why jp discards such series when they could have a perfect correlation with temperature ?

    per,

    I can’t speak for jp but I can invent a story about trees if you want one, based on my rambles around the local park. The trees in my story don’t much care about temperature. Herbivores improve their prospects by eating competitor plants and dumping lots of dung which fertilises them. So their rings exhibit a gradual rising trend, with little variance. On jp’s definition they are complacent. It’s true that they might have a strong correlation with temperature although it certainly could not be perfect unless temperature moves at a slower pace than a drugged snail. But such a correlation would be spurious. If observations on these trees were appled to a period where temperatures were high, but the helpful herbivores were absent, they would give the wrong answers. The fact that jp’s procedure excludes these observations is therefore a point in its favour.

    However there is little point in our discussing botany since we have both acknowledged that we know nothing about it. From a statistical point of view I see nothing wrong with the approach jp describes.

    Now I have a question for you. In your comment you used the words “I suspect” to introduce a number of claims. Can you point to any data which justify your suspicions?

  81. #81 per
    June 30, 2006

    >It’s true that they might have a strong correlation with
    >temperature although it certainly could not be perfect
    >unless temperature moves at a slower pace than a drugged
    >snail. But such a correlation would be spurious.

    Hmmm. Apparently, it is quite okay to announce that some statistical correlations are “spurious”, but some correlations are not. Given that we have god-like knowledge of the cause of an increased ring size (your premise), then clearly we can announce spurious correlations. In the real world, we do not have god-like knowledge of the causes of tree growth; and in fact jp excludes these tree sites where temperature is not likely to be limiting.

    So what it comes down to is that jp discards samples from sites which he believes to be candidates for temperature sensitivity, on the grounds of complacency. you have not suggested any coherent reason why he should discard those samples. Indeed, as you point out, the discarding of perfectly good samples comes very close to scientific misconduct.

    yours
    per

  82. #82 jp
    June 30, 2006

    Kevin

    You are very patient man. Per is a troll. Per states “jp discards samples from sites which he believes to be candidates for temperature sensitivity, on the grounds of complacency”. That is a lie, I did not write that. I wrote about the criteria that influence site selection, not cherry picking data from individual trees to fit some temperature trend, but Per knows this. He simply keeps this thread rolling until he gets the last word with his child in man body catch phrase: tootle pip.

  83. #83 per
    June 30, 2006

    >Tree rings sites are selected because they represent
    >locations where there is a high sensitivity in the ring
    >growth to climatic conditions.
    so sites are selected because they are candidates for temperature sensitivity

    >When you are in the field and you are sampling trees if
    >you see the ring series are complacent, you save your
    >money and time and you move on to another site.
    And you discard samples which are complacent.

    I think it obvious that I was summarising what you said, rather than quoting you. My quotes show that what I said was true.

    jp discards samples from sites which he believes to be candidates for temperature sensitivity, on the grounds of complacency

    What is more, I cannot see why jp starts throwing allegations of “lie” about. I have directly cited what he said, and my paraphrase is meant to fairly reflect just what he said. If there is ambiguity, I should be clear that I understand jp to be saying that he discards all the samples from a site which is complacent; and that my phrasing was consistent with this.

    It is a great shame that jp cannot bring his greatly superior knowledge to a civilised discussion without throwing insults about.

    yours

    per

  84. #84 Dano
    June 30, 2006

    jp:

    per has successfully spammed this comment thread.

    But any time a Hockey Stick post is created on the Internets, the FUD astroturf bots come out to spam the comment thread, so there is really nothing new here.

    Best,

    D

  85. #85 jp
    June 30, 2006

    Per

    Respect is earned. Let me explain it in simpler words: when it is clear that the samples at a site are complacent you give up on that site and (now stay with me because this is the tricky part) you open the garbage bag and you discard the complacent samples that you have collected. Actually I’ve been known to simply leave them on the forest floor.

    That was difficult wasn’t it. Done any reading yet?

  86. #86 Kevin Donoghue
    June 30, 2006

    You are very patient man.

    Thanks jp. A man shirking his work if the truth be told. Rest assured that I read your comments upthread carefully. No attentive reader will be taken in by per’s proclivity for “paraphrasing” and “translating” comments. He does it to me too. I deal with a simple question about trends, variances and the pitfalls of correlation by means of an equally simple story:

    … I can invent a story about trees if you want one, based on my rambles around the local park.

    In per’s hands my story then becomes an indefensible approach to scientific methodology:

    Given that we have god-like knowledge of the cause of an increased ring size (your premise), then clearly we can announce spurious correlations.

    My premise, mark you. This despite my making it quite clear that I am not a botanist.

    I wasn’t really expecting an answer to my question:

    In your comment you used the words “I suspect” to introduce a number of claims. Can you point to any data which justify your suspicions?

    No surprise on that score.

  87. #87 per
    June 30, 2006

    >when it is clear that the samples at a site are complacent you give up on that site and (now stay with me because this is the tricky part) you open the garbage bag and you discard the complacent samples that you have collected.

    That is what I said.

    >Respect is earned.

    You have featured two posts with oodles of name-calling, when I have merely accurately paraphrased what you have said. Did you learn this high level of intellectual discourse when you did your PhD, or do you have to be a Professor to pull this off ?

    yours,
    per

  88. #88 per
    June 30, 2006

    >The trees in my story don’t much care about temperature. Herbivores improve their prospects by eating competitor plants and dumping lots of dung which fertilises them. So their rings exhibit a gradual rising trend, with little variance.

    Dear kevin

    It may be that you don’t understand what a “premise” is; it should be clear that your little story above has premised that there is a gradual rising trend because of herbivore dung and herbivores eating competitors. “So” logically connects with the sentence before and implies causation; therefore you have simply stipulated that this is the cause of the increasing ring width, whether it is biologically likely or not, an issue that I did not address as you are not a biologist.

    >Given that we have god-like knowledge of the cause of an increased ring size (your premise), then clearly we can announce spurious correlations.

    well, it turns out that your example, with your god-like knowledge of cause and effect, isn’t tremendously helpful. In the real world, we simply have samples, correlations, and no hotline to god. You appear to be favouring the view that we can simply discard the samples that you don’t like (extracting the climate signal), until you are left with samples that just tell you what you want.

    >From a statistical point of view I see nothing wrong with the approach jp describes.

    I am gobsmacked. Wouldn’t you have to know that removing samples on the basis of variance does not introduce a subsequent bias ? Clearly, if samples with a low variance had a particular type of response (e.g. no increment in ring size), you would be introducing a substantial bias into your samples.

    Almost everything about this smacks of discarding data that you don’t like. I am deeply suspicious that if you had a flat temperature record, it is the complacent samples that would be retained because they showed a “correlation” to temperature.

    toodle-pip !
    per

  89. #89 per
    June 30, 2006

    Dano:
    >per has successfully spammed this comment thread.

    translation: I am going to call him names, because I have nothing sensible to contribute.

    >But any time a Hockey Stick post is created on the Internets, the FUD astroturf bots come out to spam the comment thread, so there is really nothing new here.

    translation: I am going to call him names, because I have nothing sensible to contribute.

    are you not going away sometime soon ?

    toodle-pip !

    per

  90. #90 Robert
    June 30, 2006

    per admitted:

    I am gobsmacked. […] Clearly, if samples with a low variance had a particular type of response (e.g. no increment in ring size), you would be introducing a substantial bias into your samples [by leaving them out].

    Oh dear. Not only is that not clearly true, that’s clearly not true.

  91. #91 Chris O'Neill
    June 30, 2006

    “The Bristlecone/California Basin proxies have very little correlation to temperature,”

    Says who?

    “Take the Bristlecones out and you can’t get a hockey stick.”

    If you take the Bristlecones out after 1450 you get the same old hockeystick. What are you going to blame for the hockeystick there?

  92. #92 per
    June 30, 2006

    Robert:
    >Not only is that not clearly true, that’s clearly not true.

    well, that is a clear assertion. Would you please like to explain ?

    I have suggest that you have a population consisting of (A +B), high and low variance tree rings. If type A has characteristic >X, and type B has characteristic

    >“The Bristlecone/California Basin proxies have very little correlation to temperature,”
    >Says who?

    The NAS panel, e.g. http://www.climateaudit.org/?p=727

    >If you take the Bristlecones out after 1450 you get the same old hockeystick.

    This statement appears to be an error, lifted from an erroneous statement during the NAS press conference. See:
    http://www.climateaudit.org/?p=724
    From the NAS report:
    >For periods prior to the 16th century, the Mann et al. (1999) reconstruction that uses this particular principal component analysis technique is strongly dependent on data from the Great Basin region in the western United States.

    toodle-pip !

    per

  93. #93 per
    June 30, 2006

    Robert:

    >Not only is that not clearly true, that’s clearly not true.

    well, that is a clear assertion. Would you please like to explain ?

    I have suggest that you have a population consisting of (A +B), high and low variance tree rings. If type A has characteristic >X, and type B has characteristic less than X, then a population of type A, or type B trees, will surely have a different X characteristic than trees with a mixture of A and B type trees.

    this seems simple. What am I missing ?

    yours, per

  94. #94 Chris O'Neill
    June 30, 2006

    “mbh99 artificially inflates the 95% confidence intervals pre-1600″

    What is “artificially” referring to?

    “the NAS panel .. specifically state that the uncertainties prior to 1600 cannot even be quantified.”

    The only thing I’ve found so far related to this is:

    “The uncertainties associated with reconstructing hemispheric mean or global mean temperatures from these data increase substantially backward in time through this period and are not yet fully quantified.”

    I can’t see the phrase “CANNOT even be quantified” specifically stated in this. Is this the sort of bias we can normally expect from you?

  95. #95 per
    June 30, 2006

    >”The uncertainties associated with reconstructing hemispheric mean or global mean temperatures from these data increase substantially backward in time through this period and are not yet fully quantified.”

    so if you ask what the uncertainty is, the answer is that it cannot be quantified; I made no statement that the uncertainties will not be quantified in future. Yes, this is a paraphrase, and I am quite happy with it. You will normally be able to tell the difference between paraphrasing, and quotes by the little “” signs :-)

    >The press release said “Less Confidence in Temperature Reconstructions Prior to 1600″. Anyone who has spent 5 minutes studying MBH99 would know that it shows a sudden increase in confidence limits before 1600. Repeating this in the press release sucks ignorant people into believing that it is news…

    It is news; their conclusions are very different to MBH 98/99, and the statistical criticisms of MBH are damning.

    cheers
    per

  96. #96 Robert
    June 30, 2006

    per asked:

    well, that is a clear assertion. Would you please like to explain ?

    per, rather than make pronouncements about what “clearly” introduces bias, why don’t you try an actual numerical example. Suppose you have two trees, x1 and x2. Let x1 be very sensitive to temperature and x2 not at all–in fact, make it exacly like your example with constant growth so that it is completely unrelated to temperature. Predict the temperature based on tree x1, then on both x1 and x2. Whether you include x2 or not, the coefficient on x1 will remain the same. In fact (and this is what you might have learned in a statistics class), x2 needn’t be constant; as long as it is uncorrelated with temperature, it won’t introduce bias.

  97. #97 Chris O'Neill
    June 30, 2006

    “”The Bristlecone/California Basin proxies have very little correlation to temperature,” Says who?

    The NAS panel, e.g. http://www.climateaudit.org/?p=727

    Where do they actually say it?

    “and they (naturally) have very little correlation to any other temperature proxy.”

    What do you mean “other proxy”? A proxy’s primary purpose is to correlate with climate variables, not directly with another proxy.

    “”If you take the Bristlecones out after 1450 you get the same old hockeystick.”

    This statement appears to be an error, lifted from an erroneous statement during the NAS press conference.”

    Didn’t come from anything to do with NAS. An MBH-reconstruction using the proxies from 1450, and without using the Bristlecones, was done by Wahl and Ammann 2006. Read it, you might learn something.

  98. #98 per
    June 30, 2006

    Dear robert

    that is clear, if it isn’t the question I asked. One of the issues which may have been implicit is that you need the whole stand of trees to have a correlation with temperature, and not just any one.

    If the whole stand of trees has a correlation, that allows you to core old, dead tree stumps, and assume that they have the same good correlation with temperature. The biological “underpinning” is that the conditions are such that the temperature is limiting growth.

    So there is a requirement (as jp stated) that the trees in a stand show a similar profile. 99 complacents and 1 “responder” doesn’t cut it. I would understand that this must change your position ?

    cheers

    per

  99. #99 per
    June 30, 2006

    >Didn’t come from anything to do with NAS. An MBH-reconstruction using the proxies from 1450, and without using the Bristlecones, was done by Wahl and Ammann 2006.

    Are you referring to figure 4a or 5c, where exclusion of the bristlecones causes the reconstruction to fail ? Where the 1400s are up to 0.3 c warmer ?

    >Read it, you might learn something.

    quite !

    per

  100. #100 Chris O'Neill
    June 30, 2006

    “so if you ask what the uncertainty is”

    They weren’t talking about the uncertainty. The were talking about the uncertainties.

    “and I am quite happy with it”

    You think I care what you’re happy with?

    “You will normally be able to tell the difference between paraphrasing, and quotes by the little “” signs”

    It has become pretty obvious that your “paraphrases” need to be taken with a grain of salt.

    “”The press release said “Less Confidence in Temperature Reconstructions Prior to 1600″. Anyone who has spent 5 minutes studying MBH99 would know that it shows a sudden increase in confidence limits before 1600. Repeating this in the press release sucks ignorant people into believing that it is news…””

    “It is news; their conclusions are very different to MBH 98/99, and the statistical criticisms of MBH are damning.”

    Yeah if you say so.

The site is currently under maintenance and will be back shortly. New comments have been disabled during this time, please check back soon.