Sunday, April 29, 2012

19 April 2011 – Poll position


‘Tis the season when everybody seems to be paying attention to pollsters, so I thought I might take a moment and talk about some polling data that has absolutely nothing to do with anything going on in Canada.  There's a new Brookings paper out that uses polling data to analyze the “climate of public opinion” on climate change in Canada and the US.  I take this as an indication that I’ve been lax in recent weeks in commenting on the state of the climate debate, and that some (or at least one) of you are feeling the lack most acutely.  Mea culpa; I thought I had already sufficiently thrashed this deceased equine, but apparently not.  So be it; I am but a servant.  Ask, and ye shall receive. Time to turn the crank once more.  Boy, my whip!

The paper, entitled “Climate Compared: Public Opinion on Climate Change in the United States and Canada “, was written by 3 political scientists from various universities, and is attached.  It makes for interesting reading.  The timing, too, is interesting, particularly in view of the spate of polling results from around the world showing a general decline in public belief in global warming.(Note D)

From a scientific perspective, the paper has a number of problems.  First, while the pollsters asked all sorts of questions about “belief” in global warming, they didn’t ask any questions that would enable the analysts to differentiate between respondents who believe that global warming is driven by anthropogenic greenhouse gas emissions (which is the key contention of the anthropogenic global warming or AGW hypothesis), and those who believe it to be a natural phenomenon. This is the single most important scientific question of the debate; Kyoto, Copenhagen, carbon taxes, cap and trade, and other emissions control mechanisms are all useless - and potentially destructively useless - unless human carbon emissions are indeed, as the IPCC contends, the single most important factor driving global climate.  All the survey asked was “From what you’ve read and heard, is there solid evidence that the average temperature on Earth has been getting warmer over the past four decades?”  Even if this is true - and I’m not saying it’s not true; if the temperature records are accurate, then the Earth has been warming, although for a lot longer than the 150 years of human industrialization – even if it’s true, from a policy perspective what’s important is not whether but rather why the warming has occurred.  That’s a big oversight and, as the purpose of the study is to guide policymaking, the authors’ failure to investigate the issue of causality undermines the whole exercise.  As one observer puts it, ”the fact of warming tells us nothing about the cause.”

The failure to differentiate between the existence of a warming trend and its origin is not surprising, because the second problem is that the authors confuse “determinant” with “correlation.” On the basis of their results, for example, they argue that political affiliation is a “determinant” of belief in global warming.  The problem is that the word “determinant” implies a causal relationship.  Correlation is not causation.  One can certainly note a correlation between political affiliation and belief in global warming based on these data, but the data do not support the alleged causal link, because the questions weren’t designed to elicit which came first, the political affiliation or the belief.  In fact, the data equally support the argument that belief in global warming is a determinant of political affiliation.  This is not all that different, ironically, from how the IPCC, based on an observed, linear increase in atmospheric carbon dioxide concentrations and an observed (though nonlinear) increase in average global temperatures during the 20th Century, has posited a causal relationship between the two phenomena.  Very little empirical evidence presented to date justifies inferring a causal relationship between temperature and CO2 concentrations, and that which does - the Vostok and other Antarctic ice cores - actually demonstrates that over the past million years or so, temperature increases have preceded increases in CO2 concentration by hundreds of years - or in other words, the arrow of causality appears to point in the direction opposite to that posited by the AGW thesis.(Note E)  This example illustrates the importance of not interpreting correlation to mean causation without data to support your interpretation.

This is not a small point.  For example, based on the authors’ definition of “determinant” and the data they present in Table 4, there’s a stronger case for arguing that being a woman is a “determinant” of believing in global warming (63% correlation) than there is for arguing that being a Republican is a “determinant” of non-belief in global warming (43% correlation). As a matter of fact, after Democrats (69% believers), there is a far higher correlation between folks on the distaff side and belief in global warming than there is between political affiliation and belief. According to a 2009 Rasmussen poll, women in the US also significantly lead men in believing that Jesus Christ rose from the dead (87% women/70% men) and that he was the son of God (89% women/74% men).(Note A)  This leads us to some odd conclusions.  On the basis of the results cited in the Brookings article, for example, one is not justified in concluding that political affiliation determines belief in global warming; but on the basis of the results presented in the Brookings article and this second poll, one IS entitled to surmise that, by a statistically relevant margin (10% or greater) women are more likely than men to believe in a proposition that, shall we say, lacks empirical support.

This observation is neither sexist nor critical of the theologically-inclined; I am merely making a value-free deduction from statistical evidence presented by one of the Western world’s premiere polling firms.  Don’t shoot the messenger!

The authors of the Brookings paper should probably also have further investigated the remarkable split among Democrats on the question of belief in global warming - the 53% divergence between the two sides (69% believers/16% non-believers) is simply extraordinary, and statistically far more interesting than the statistically irrelevant 2% divergence between Republicans (43/41). It’s also interesting, although the authors pretty much ignore it, that uncertainty about whether there is sound evidence for global warming correlates positively with age; the older you are, the more likely you are to be on the fence.  In any case, according to another 2009 Rasmussen poll, the proportion of Americans who are unsure whether there is evidence for global warming (16%) is lower than the number who believe in ghosts (23%).(Note B)  Not sure if that proves anything, but I would suggest that it says something about our ability/willingness to believe - to use the Rasmussen article’s words - “as-yet-unprovable things.”  Interestingly, that latter survey shows that the percentage of US women who believe in ghosts is virtually the same as men, which calls into question the point I made above.  Why would men be proportionally more likely to be sceptical about global warming and matters of faith than about ghosts?  Similar studies show that US liberals, despite being on average slightly less religiously observant than US conservatives, are statistically more likely to report having seen a spectre.(Note C)  All of this is fascinating and bears on the question of why we believe (or disbelieve) the things we do. 

To put it another way, we are justified concluding that “humans are unpredictable and illogical.”  Although, while true, such a conclusion doesn’t lead us anywhere…and in any case, it’s hardly original.

Finally, from a methodological perspective, I have problems with using unscientific, emotionally-loaded words like “believe.”  Of course, using scientifically acceptable terminology would push the polling effort toward questions that would require respondents to have at least a passing familiarity with oodles of complicated research and would require laymen to evaluate the strength of evidence (“Do you think that the Urban Heat Island effect, poor station siting, station dropout, manual manipulation of temperature data and excessive interpolative smoothing undermine the US ground-based temperature record?”) - at which point you might start getting a lot more “don’t knows”, or responses that an institution like Brookings could potentially find uncomfortable.  I feel bad for Brookings, by the way; the last time I was there, they had a very earnest Gordon Gecko “greed is good” chappie give a presentation on how carbon markets were the way of the future.  It must be more difficult to find such presenters now that the Chicago Climate Exchange has gone belly-up and assumed room temperature.  There’s also the issue of whether questions of science should be decided by polls instead of through validation or falsification of an hypothesis by observed data; if that was the way science worked, we’d all be breathing phlogiston and having our children’s cranial contours checked by credentialed phrenologists to determine whether they were likely to become axe-murderers.  Finally, I would note that while the Brookings article talks about “climate change”, the poll questions were all about “global warming”.  This suggests to me that the authors think that the two are the same, although the IPCC and most proponents of the AGW thesis have argued for several years now that they are not, and that increased carbon dioxide concentrations may lead to cooling as well as to warming (because if they can’t, then the last 15 years of temperature data tend to falsify the AGW thesis).  This terminological confusion suggests either that the authors didn’t get the memo, or that the poll was conducted while “global warming” was still the officially-sanctioned term, and had not yet been replaced by “climate change”.  If they repeat the exercise, they might wish to note that the new term is “climate disruption” (or maybe “man-caused disasters” - I’ve lost track of the euphemisms) so as to forestall accusations of lax lexicography.

Why should we care about all this?  No reason.  No reason at all.  It’s an interesting paper about how climate change is perceived in the US (and Canada).  But the horse is starting to stiffen up and stink, so I think I’ll lay off it for awhile, and let it continue to decompose in peace.

D) []

E) The Vostok Ice Core data and their interpretation is a fascination subject.  The data themselves are available on-line at [].  For interpretations, see J.R. Petit, et al., “Climate and atmospheric history of the past 420,000 years from the Vostok ice core, Antarctica”, Nature 399 (1999), 429-436. Another author notes that “A review of the recent refereed literature fails to confirm quantitatively that carbon dioxide (CO2) radiative forcing was the prime mover in the changes in temperature, ice-sheet volume, and related climatic variables in the glacial and interglacial periods of the past 650,000 years”.  Willie Soon, “Quantitative implications of the secondary role of carbon dioxide climate forcing in the past glacial-interglacial cycles for the likely future climatic impacts of anthropogenic greenhouse-gas forcings”, Physical Geography, 4 July 2007 [].  Also, “Over the full 420 ka of the Vostok [ice core] record, CO2 variations lag behind atmospheric temperature changes in the Southern Hemisphere by 1.3 +/- 1.0 ka [thousand years].”  Manfred Mudelsee, “The phase relations among atmospheric CO2 content, temperature and global ice volume over the past 420 ka”, Quaternary Science Reviews 20 (2001), 583 [  The_phase_relations_among_atmospheric_CO2_content_temperature_and_global_ice_volume_over_the_past_420_ka.pdf].  The same conclusions have been reached by many other researchers; see, for example, H. Fischer, et al., “Ice core records of atmospheric CO2 around the last three glacial terminations”, Science 283 (1999), 1712-1714; A. Indermuhle, et al., “Atmospheric CO2 concentration from 60 to 20 kyr BP from the Taylor Dome ice core, Antarctica”, Geophysical Research Letters 27 (2000), 735-738; E. Monnin, et al., “Atmospheric CO2 concentrations over the last glacial termination”, Science 291 (2001), 112-114; P.U. Clark and A.C. Mix, “Ice Sheets by Volume”, Nature 406 (2000), 689-690; and N. Caillon et al., “Timing of atmospheric CO2 and Antarctic temperature changes across Termination III”, Science 299 (2003), 1728-1731.