A couple of posts ago, I stated that I was undecided on the question of anthropocentric global warming, in part because the experts themselves cannot agree. But just how far from consensus are the experts?

According to this article, environmental scientists Dennis Bray and Hans von Storch conducted a survey of 530 of their fellow researchers in 27 countries on the subject of global warming. On the question of whether global warming is primarily the result of anthropocentric causes, on a scale of 1 (strongly agree) to 7 (strongly disagree), the average score was 3.62. (The value of 4 can be considered as ambivalence or impartiality with respect to this question.) So clearly, at least from this survey, it can be said there is no real majority of opinion on the question, let alone anything approaching “consensus”.

A good critical thinker, when presented with a new piece of data or evidence, tries to establish the provenance and pedigree of that data/evidence before allowing it to inform his or her opinions. So rather than blindly accepting this article’s account of the survey at face value, I did a quick bit of googling and found the original full survey report here. It does appear to be a well designed scientific survey (as opposed to the type of unscientific just-for-fun survey you might find posted on cnn.com or foxnews.com), and also includes far more detail and many more questions than what was covered in the original article. The individuals who conducted the survey are highly credentialed — one is a professor at the meteorological institute at the University of Hamburg, Germany, and director of the Institute for Coastal Research; the other is a researcher at the Institute for Coastal Research.

Of particular interest, the report includes the survey results on questions intended to gage the scientific community’s opinions on how well the current climate models perform. For instance, on the question “How well do you think atmospheric climate models can deal with the following processes?”, on a scale of 1 (very adequate) to 7 (very inadequate), the answers ranged from 3.22 to 4.71 for the following processes: Hydrodynamics (4.45), Radiation (4.71), Vapour (3.83), Clouds (3.22), Precipitation 3.29), and Convection (3.48). This certainly doesn’t instill much confidence in the accuracy of the climate models being used to make these predictions. And yet the media routinely reports climate findings based on these models as if they were “truth”, neglecting to mention the degree of uncertainty associated with these findings.

Bottom line: There is no consensus among the experts, and there is significant uncertainty in the models themselves. So I’ll continue to stick with my “I don’t know”.

Addendum:

Somebody has brought to my attention a critique of this survey, posted by Tim Lambert, a computer scientist, on his blog. The critique points out that the URL to the online survey was originally posted in the Bulletin of the American Meteorological Society and other institutional lists within the climate research community. However, somebody later sent out the survey URL to a climate skeptics mail list. Since the climate skeptics mail list has about 200 members and the total number of survey respondents was 557, the survey results could be significantly skewed, claims Lambert. Moreover, since the survey was anonymous, there is no way to guarantee that only climate scientists participated in the survey.

The original full report actually addresses this controversy:

Critics of the survey suggested that sceptics could submit multiple copies of the survey (see: Lambert, Tim, 2005), thereby biasing the results. (However, no criticism was raised suggesting that the other polemic might also act in a similar manner, that is, a biasing of the results by multiple submissions by climate change alarmists.) It is claimed that the 2003 survey was posted on a sceptics mailing list and concern was raised that the sample for the 2003 survey might not be representative and as such the results invalid. In an effort to determine if indeed the sample was biased the Two-sample Kolmogorov-Smirnov test and the Wald-Wolfowitz Test (general tests that detect differences in both the locations and the shapes of distributions) have been employed.

For the uninitiated (myself included), the Kolmogorov-Smirnov and Wald-Wolfowitz tests are statistical tests to determine whether two samples come from the same population — in this case, whether the survey results came from a single population or from multiple populations with different distributions (e.g., a population of climate researchers and a population of non-researchers and/or skeptics). These tests were applied to both the 1996 survey (the URL for which was NOT leaked to climate skeptics) and the 2003 survey, and the tests indicated no significant differences in population — i.e., it does NOT appear that the 2003 survey was somehow “poluted” with non-climate researchers or climate skeptics.