Statistics

Reading Spencer’s Radiation Paper; Polar Bears Flap: Scientist Under Investigation

Thanks to all the readers who have altered me to Roy Spencer and William Braswell’s new paper, “On the Misdiagnosis of Surface Temperature Feedbacks from Variations in Earth’s Radiant Energy Balance,” published in Remote Sensing. I have just downloaded a copy of the paper and will read it later today.

It’s been picked up widely in the press, thanks to Jim Taylor writing about it in Forbes. Taylor even managed to sneak in the word “alarmism;” not once, but many times.

I also specifically note that my pal Gav has not yet written about Spencer and Braswell’s new work. Well, best to be cautious. I will follow Real Climate’s example and take my time to analyze the statistical model used in the paper.

When I’m finished, I’ll initiate a new post called “Misdiagnosis of Surface Temperature Feedbacks”, and I’ll put a link to it here. It may be tomorrow, it may be next week. I have to travel again early next week: I wish I had multiple grants from NASA and Greenpeace like some so that I could stay in one place! (Rest assured that I will adopt a full-body scientific attitude when reading the paper.)

Meanwhile…

You’ll have heard that Charles Monnett, with the U.S. Bureau of Ocean Energy Management, Regulation and Enforcement, was iced by his bosses over “integrity issues.”

Monnett was one of those scientists who provided the beat to the tune It’s Worse Than We Thought when he and a colleague managed to slip a peer-reviewed paper into Polar Biology which intimated that polar ice was disappearing so fast that polar bears were drowning at a shocking rate, because they had to swim increasing distances in the newly created open water.

His evidence was in the form of observations of a couple of bears that drowned after a particularly powerful storm (winds passing 50 km/h). He said that these observatios “suggest that drowning-related deaths of polar bears may increase in the future if the observed trend of regression of pack ice and/or longer open water periods continues.”

The activist group Public Employees for Environmental Responsibility has taken up Monnett’s cause and claim that the man is being “hounded” over the paper.

One point of contention is where Monnett said, “These extrapolations suggest that survival rate of bears swimming in open water during this was period was low (9/36)=25%).” That’s the statistic that was picked up by Al “Buy My Carbon Offsets” Gore, and which Gore used to prove that most polar bears would soon be dead.

In a hilarious exchange with investigator Eric May, Monnett tries to deny that the 25% statistic was a statistic:

ERIC MAY: Did they comment at all about any of the stats or —

CHARLES MONNETT: Uh, there’s no stats in [the paper].

ERIC MAY: Well, calculations, for, for example, the 25 percent survival rate.

CHARLES MONNETT: Oh, well, that’s just a mindless thing. That’s in the discussion. Um, that is not a statistic. Um, that’s a ratio estimator. It’s a it’s a fifth grade procedure. Do you have kids?

ERIC MAY: No.

CHARLES MONNETT: Okay, well, if you had kids, you would know that in about fifth grade, they start doing a thing called cross multiplication. “X” is to “Y” as, you know, “N” is to “M.” And you can — there’s, there’s a little procedure you use to compare the proportions. And so that’s a, um, simply a calculation. It’s not a statistic.

ERIC MAY: Okay.

CHARLES MONNETT: And, uh, we were very careful, um, in how we presented that, to first make it clear that we had – we didn’t have sufficient sample size, although a, a, a peer statistician type would probably argue we did. But we felt we didn’t have a sufficient sample size to do statistics and, you know, and to estimate, to do any estimators or confidence intervals or anything like that on. And we put caveats throughout that section, saying that, uh, “it’s possible.” And we felt that, um, we didn’t want to leave the reader thinking that, “”Okay, they went out, and they surveyed it, and there were four dead bears.” Because this is a survey, and it only looks — it only covers a small part of the habitat.

Well, Chuck, old boy, that was a statistic. A bad, too; and wildly misleading.

It’s all made worse when Monnett admits that he actually only saw three dead bears—and not four, as he claimed many times and in many places. Monnett says that this goof was “not scientific misconduct anyway. If anything, it’s sloppy.”

It is obvious that Monnett’s work was badly off. But if this is all they have against him, best guess is that Monnett, being a federal employee, will not be fired. He could argue that, after all, his paper was reviewed and accepted by his peers, that no paper is considered final, and so forth. And how often are government employees fired?

Categories: Statistics

26 replies »

  1. You are right to be a little wary of the Spencer and Braswell’s statistics… source data is available here if you want to do it properly:

    http://ceres-tool.larc.nasa.gov/ord-tool/jsp/SSF1degSelection.jsp

    As for Monnett, you have obviously not read the paper in question, and so I’d suggest you do so before further making things up:

    http://www.peer.org/docs/doi/7_28_11_Polar_Bear_paper.pdf

    The mortality calculation can obviously be done much better than the simple ratio suggested in their paper. A good example for showcasing the ‘new stats’ in fact. Perhaps you could focus on that instead of wishcasting a situation based on the politics.

  2. Gav,

    Sure, I read the polar bear paper. Most recently as this morning. The statistic-that-was-not-one was, as Monnett himself suggested, a sloppy extrapolation.

    I’m also glad you agree that he fudged the “3” dead bears to “4.”

    I particularly liked where he said, in the paper, and without any observational data to back him up, “Other bears may have suffered sublethal effects and later succumbed due to exhaustion or inspiration of sea water as a result of swimming long distances in rough seas.” Just as still other bears might have benefited from having a fat meal the day before and holding up during the storm.

    Plus, thanks for the link to the data! As always, you brighten my day.

  3. “thanks for the link to the data!”

    one lives in hope that you might use your powers for good.

  4. I’m also quite weary of Spencer’s paper. So I’ll look forward to both Briggs’ and Gavin’s criticisms of it.

  5. Gavin – could you site what you see as “made up” in Briggs’ comments?

    Thanks

  6. I’m still trying to figure out how the bureaucrats manage, regulate and enforce ocean energy. If they can do that, why doesn’t Hansen call them and tell them it’s too hot, reduce the ocean energy AKA temperature?

  7. The mortality calculation can obviously be done much better than the simple ratio suggested in their paper.

    /snark on/

    Gavin… Curious. Did you point that out when the paper was first published? Did you criticize this paper for being published with such a weak ratio being presented as thorough scientific validation, or as you so like to say “Robust”?

    /snark off/

  8. How can we…or more importantly, why should we be “wary” of the statistics used in & conclusions derived from–in a “PEER REVIEWED” paper???

    After all, so many of the warming advocates, alarmists, and/or whatever one might label them have held that “peer review” is the gold standard of credibility. If that no longer holds true–which Gavin is implying here now–opens to question & debate the credibility of ALL related peer reviewed papers….

    I wonder if this means that someday there will be an expose’ of e-mails & voice-recordings of reputable climate scientists undertaking to blacklist the journal Remote Sensing and certain associated reviewers…

    FRANKLY, this is getting funny [in a not so funny way]. Cloud-related data is very poor (generally nonexistant relative to what is desired) and any resulting calculations are inherently ambiguous. See NASA: http://climate.nasa.gov/uncertainties/ — where solar issues and cloud issues are identified as major sources of uncertainty.

    Still.

    CERN’s CLOUD exerimental results will contribute hugely to filling & quantifying uncertainties…but…some [so-called] experts have effectively dismissed [on record] such data long ago in the absence of any other data…just selectively selecting different data sets from the same hyper-limited/deficient data. So of course new studies & research will be dismissed per precedent. To do otherwise would be to admit mistakes were made.

    This whole global warming alarmism has been so poorly–extraordinarily so–communicated to the public by the self-proclaimed experts they have no one to blame but themselves for the ongoing debate, in all its forms. This was entirely avoidable.

    What seems very apparent is certain researchers have made the “correlation must reflect causation” error (maybe blundering into a real cause-effect relationship here & there by pure luck) they’ve got no choice but to keep up the same story — the analytical analogy to painting oneself in the proverbial corner. Eventually, probably relatively soon, they’ll be forced to leave their tracks in the wet paint. Tracks leading from their well-documented position to their self-inflicted credibilty downfall. That must be terrifying.

  9. “How can we…or more importantly, why should we be “wary” of the statistics used in & conclusions derived from–in a “PEER REVIEWED” paper???”

    Never … EVER … trust a pro-AGW “pal-review” paper.

  10. one lives in hope that you might use your powers for good

    That’s Gavin, as usual the best person at making an argument against…Gavin, suggesting statistical and in general scientific analysis should be subordinate to Greater Things such as the “good”, whatever that is.

    As if a sloppy paper made for a Good Cause would be preferable to a great paper uninterested in morality and politics.

    Thanks Gavin! Well, at least that attitude does explain your whole career, doesn’t it? Perhaps you might want to change your website’s title to “Moral Science”..

  11. Gavin’s reply that our host should not be “…wishcasting a situation based on the politics” is another keeper for the quotation book. How many studies have been posted by RC and their ilk, without a squeak of criticism over the blatant political implications?

  12. How very funny and revealing Gavin’s comments are. He really does see himself as a climatological Luke Skywalker, with Prof Briggs, Steve McIntyre and many others as having made a conscious decision to go over to the dark side. It explains how ridiculously tribal RC is, with the most shocking abuses of science defended to the death and any criticism attacked in a primarily ad hominem way.
    I used to think this was a phenomenon caused by watching too many Westerns, where the goodies and baddies were clearly defined, but the genre became more nuanced so there must be another explanation. Reminds me of a time when some student friends and I sat around smoking a lot of weed and nearly convinced ourselves we had been chosen by a higher power to save the world – perhaps that’s it!
    On second thoughts perhaps it is not so funny. We have just seen in Oslo what can happen when people believe
    a) that it is their role to save civilisation
    b) that the ends justify the means.

  13. I wonder if Gavin would like to comment on other poor statistics in climatology…..oh I don’t know maybe the dire understanding of statistics which lead to Dr Mann’s broken hockey stick?

  14. Some time ago I had my cat ‘altered’ by the vet. He was never quite the same again. 😉

    I am looking forward to your comments regarding the Spencer paper. Hard to know who to trust when it comes to the ‘science’ of climate these days, but I am happy to see you weighing in on this paper.

  15. What Monnett is trying to explain,to someone with apparently zero background, is that only three of the dead bears was seen during the transect portion of the flights, and hence only three were included in their calculations. This is indicated clearly in Monnett&Gleason – so its your criticism thats “badly off” – fortunately you won’t be subjected to a criminal investigation.

  16. Here are some instinctive statistical questions on Section 2.3 of Spencer and Braswell’s paper:

    * How is the lag of 18 (plus and minus) chosen? Why not 12?
    * Standard errors in the lead-lag regression coefficient estimates?
    * No model diagnostics?

    Maybe the researchers in the area know the literature well so they can read the brief Section 2.2. (Coupled Climate Model Data) and figure out what’s been done exactly to derive the 3 least and most sensitive models. I sure can’t.

    Just to be clear, I am not saying Spencer and Braswell’s conclusions are incorrect. Just as the paper discussed in this post where the statistical analysis is clearly wrong, without analyzing the data further, one can’t conclude that a more appropriate analysis won’t yield the same conclusions.

  17. I would like to clearly claim at this point of time that it is my considered opinion (Pr 0.9999999995, Rsqd an even larger order of magnitude number greater than the previous small punny number).
    There you see I am an expert statistician.

  18. Unfortunately somehow my main claim that pigs may fly sometime in the future got cut off.
    Perhaps just as well.

  19. Yes but Gavin has to explain why the dead polar bears were not upside down, so they could not be used in the analysis, but the one that was stood on it’s head was 😉

  20. Ten polar bears standing on a wall
    Ten polar bears standing on a wall
    And if one polar bear should accidentally fall
    There’d be nine polar bears standing on a wall.

    That would seem to be about equally scientific.

  21. Folks,
    Gavin was kind enough to join the conversation here. I think it the least we can do to engage him in a respectable way. He is after all a player in the whole debate, many of us which insist isn’t yet over. I include myself in the skeptical camp but still would enjoy getting his opinions on the developments of the day, and seeing those opinions discussed calmly without resort to snark.

  22. The BBC was covering this new paper about sea ice cover in the Arctic (http://www.sciencemag.org/content/333/6043/747.abstract). The paper suggested through their proxy studies they could plot past sea ice distributions showing that 5,000 years ago the sea ice extent was some 50% smaller than present. The limit of permanent sea ice was estimated to be 1000 km north of current location.

    Clearly “scientist” Monnett will have a hard time convincing reviewers of the validity of his mortality statistics if it holds that sea ice extent varied this much in the recent past. Polar bears are known to be at least 250,000 years old as a species, so presumably open water swims were necessary in the past.

    And concerning Gavin? I posted similar to this in a comment over at RC – and it was “boreholed”. No wonder there is a piling on when he wanders by other web sites.

Leave a Reply

Your email address will not be published. Required fields are marked *