A detailed critique of the Cook et al. article.
1. The Oxford English Dictionary (2014) defines consensus as “Agreement in opinion; the collective unanimous opinion of a number of persons.” A consensus among scientists does not require unanimity, but it does imply at least a strong majority. Cook et al. redefined consensus to mean not merely to have an opinion, but to “express an opinion” and in writing. It is obvious that many more scientists accept a theory than may happen to directly express their acceptance in their articles (see 5). Thus the Cook et al. method is bound to underestimate the consensus of acceptance.
2. As shown in the chart, Cook et al. threw out two-thirds of the published literature, 7,970* of 11,944 articles, because these articles did not express a position on AGW. They thus ruled some 21,240 publishing climate scientists out of the consensus (7970 articles X 2.665 authors/article, the 1991-2011 average.)
But the consensus as defined and commonly understood is what the majority agree to. It is logically impossible to rule out a two-thirds majority of climate authors and still measure the consensus on AGW.
3. Of the 11,944 articles in the Cook et al. database, only 64, or 0.5%, “endorsed” the position that humans are the principal cause of global warming, the standard definition of AGW. This leaves a conundrum: If direct endorsement is alleged to be the criterion of consensus, yet 99.5% of publishing scientists do not meet that criterion, then either there is no consensus or direct endorsement is not the proper measure or consensus. This led Legates et al. (2013) to write, “[Cook et al.’s] claim of 97.1% consensus, therefore, is arguably one of the greatest items of misinformation that has been circulated on either side of the climate debate.”
4. Cook et al. ruled 7,930 articles out of the consensus because they did not “address or mention the cause of global warming,” labeling them as taking “no position.” But as just noted, if two-thirds of authors truly have no position, then ipso facto AGW cannot be the consensus position of climate scientists. Conversely, if AGW is the consensus position, then Cook et al. have demonstrated that a majority of authors accept the theory but do not “express an opinion” on it in their articles, thus falsifying the Cook et al. method.
5. The articles that Cook et al. reviewed are about global warming—that is why they came up in their search. Obviously, the authors of the “no position” articles do have an opinion on global warming—they are writing about it. I argue that virtually all of them accept AGW and are part of the consensus.
6. I make that argument because publishing scientists almost never directly endorse the ruling paradigm of their discipline. Not one author of 500 recent articles on plate tectonics did so, nor did the authors of 261 articles on evolution or 100 articles on lunar craters. Of 283 articles on global warming in Environmental Research Letters in 2013 and 2014, none endorsed AGW, including Cook et al. (see Endorsement). Applying the Cook et al. method to plate tectonics, evolution, meteorite impact, and the ERL articles leads to dividing zero by zero.
Before we can know whether a publishing biologist accepts Darwinian evolution, to take one example, does the biologist really have to say so explicitly and in writing? Of course not. The literature of science itself falsifies the use of explicit endorsement, the sine qua non of the Cook et al. method, as the criterion of consensus.
7. The premise of their method is that only authors who directly attribute global warming to human activities can count in the consensus (“address or mention the cause of global warming”). But authors who write about some aspect of global warming other than its cause have no particular reason to make such an attributive statement. The Cook et al. method thus has do not with the scientific consensus on AGW, but with language and whether the subject of an article lends itself to a statement on attribution.
8. Cook et al. ruled out of the consensus many articles by distinguished climate scientists whose acceptance of AGW is not in question, placing them in the "no position" category. These included (with number of omitted articles) R. Bradley (3), K. Briffa (2), E. Cook (5), J. Hansen (6), M. Hughes (2), P. Jones (3), T. Karl (5), M. Mann (2), M. Oppenheimer (3), B. Santer (2), G. Schmidt (3), the late S. Schneider (3), S. Solomon (5), K. Trenberth (7), and T. Wigley (3). (From the Cook et al. Supplemental Materials.) Can a method that omits from the consensus articles by such eminent climate scientists possibly be right?
9. Most of the above authors in the “no position” category, and many others, also had articles in one or more of the three Cook et al. endorsing categories. James Hansen, for example, has 4 articles in Cook et al. category 1 (Explicit Endorsement with Quantification), 6 in category 2 (Explicit Endorsement without Quantification), and 6 in category 3 (Implicit Endorsement), as well as the 6 in category 4 (no position), as noted. That articles by a single author can appear in several different Cook et al. categories is critical to understanding exactly what Cook et al. measured and did not measure. Hansen, for example, has articles in Cook et al. category 1, establishing that he fully accepts that humans are the primary cause of observed global warming. This is what we want to know: whether James Hansen is part of the consensus on AGW, and he is. That he also has articles in other Cook et al. categories has nothing to do with whether he accepts AGW. Rather it has to do with the subject of those articles and whether they lent themselves to the explicit endorsement that Cook et al. required. See here for an analysis of two of the Hansen et al. articles.
10. Cook et al. emailed 8,547 authors in their database inviting them to rate their own papers and received a response rate of 14%. They could have asked these authors whether they accepted AGW, but went out of their way to avoid doing so, saying, “We are not asking about your personal opinion but whether each specific paper endorses or rejects (whether explicitly or implicitly) that humans cause global warming.” Thus, even without the low response rate, this querying of authors suffers from the same fatal flaws as the overall rating methodology of Cook et al. It is about the subject and language of articles, not about whether their authors accept AGW, which as noted, Cook et al. ruled out as a legitimate question.
For all these reasons, it is clear that Cook et al. classified articles: they did not measure the scientific consensus on anthropogenic global warming. What did they measure? Here is my take on that question.
* Cook et al. found 7,930 that took no position and 40 that were "uncertain on AGW," for a total of 7,970.