The “97 percent consensus” on human-caused climate change looms large over the climate debate. Activists proclaim that there exists a scientific consensus around the statement “humans caused most of the recent global warming,” saying the science is settled and it is time to move ahead with solutions. Skeptics protest the methods used to measure consensus, challenge its very existence, or muse that it is (counterintuitively) evidence of less certainty about the causes of climate change.

Last week, John Cook and a group of other consensus advocates revisited this issue in a new paper. Cook et al. reviewed 14 studies which estimate the climate consensus and find:

  • Between 90 and 100 percent of experts agree with the proposition that human activities are causing global warming, and
  • Agreement of those surveyed increases with expertise.

As co-author Dana Nuccatelli writes in the Guardian, “including non-experts is the only way to argue for a consensus below 90–100” percent.

Instead of going through all 14 of the studies surveyed by Cook et al., here are a few examples that show the range of methodologies employed to survey scientific opinion.

  • Aderegg et al. (2010) found a 97 percent consensus amongst the 200 top-publishing climatologists who signed public statements or assessment reports about climate change. Sixty-six percent consensus was achieved from all scientists signing such statements.
  • The Pew Research Center (2015) found a 93 percent consensus amongst working Earth Science PhDs who were members of the American Association for the Advancement of Science (AAAS). Eighty-seven percent of all AAAS members agreed with the consensus.
  • Cook et al. (2013) found a 97 percent consensus by surveying the abstracts of papers on global warming published in the peer-reviewed literature from 1991-2011 (more on that paper below).

This new paper from Cook  et al. emphasizes the relationship between expertise and consensus. When sorted by the expertise of the survey participants, surveys show a strong relationship indicating that the more you know, the more you affirm the primacy of the human causes of climate change. That makes sense, given the literature on the issue.

Screen Shot 2016-04-20 at 11.47.09 AM

Source: Cook et al. 2015. Colors indicate confidence

If one accepts that the multiple methods for establishing a consensus are converging on something like 90 and 100 percent of experts agreeing that humans caused recent warming, what does that really say? It says a lot, but less than is sometimes claimed by activists.

The consensus represented in these studies is a fairly limited proposition: the existence (and to some degree, the extent) of recent human influence on the climate. It says nothing about the future response of the climate to anthropogenic greenhouse gas emissions, the potential impacts of climate change, or the need for climate action.

Details like that distinguish the ninety-seven percent consensus on human-caused climate change with the consensus statements that comes out of scientific assessments from the IPCC, the Obama Administration, or the National Academy of Sciences. Too often, activists will attempt to carry the strong consensus over human-caused climate change to the judgments and statements from those reviews.

However, while we should always question our assumptions, the simple story that there is a high-level of consensus that climate change is real and a threat deserving action may hold. This survey of economists who publish in the field, found that most think we are underestimating the costs of climate change in instruments like the Social Cost of Carbon, and that climate change will likely harm economic growth.

The Implications of Consensus

Climate skeptics are not wrong to argue that consensus among experts does not, in itself, settle a debate. Scientific truth is not established by a show of hands. One does not have to be a published expert in a narrow field to assess the strength of scientific evidence from related (or even distant) fields of expertise. Scientists alarmed about climate change are more likely to choose certain career paths and specialties than those who are not. Group-think can be a real—and counterproductive—phenomenon. Editorial bias can influence what gets published and what does not. All true.

The broader conclusion, however, was best forwarded by Michael Shermer recently in Scientific American. The evidence for human-caused climate change has converged from multiple lines of evidence, been vetted by skeptical reviews, and presents a consistent and cohesive view. No critical theory or invocation of “natural variability” can claim the same.

There is always room for the skeptical assessment of existing theory. But this recent paper from Cook et al. makes it very difficult to argue that there isn’t a consensus among experts about what the data is telling us.

A note on the Cook et al. (2013) survey of abstracts

The 2013 study from Cook et al. has received some criticism on the Right for how it classifies different paper abstracts. Given the controversy surrounding that study, it’s worth considering in some detail.

Cook et al. (2013) classified the agreement of 11,944 papers with the consensus position with a scale of 1 – 7 (data here) by rating the text of the abstract. A score of 1 indicates a quantified agreement with the consensus (i.e. human influence {is greater than 50 percent of, dominates, is primarily responsible for} observed warming); 2 indicates explicit, unquantified agreement; 3 equals implicit agreement; 4 equals no position or uncertain; 5 equals implicit rejection; 6 equals explicit, unquantified rejection; and 7 indicates quantified, explicit rejection.

Cook et al. found that 3,896 papers fit classes 1-3, 78 papers fit in 5-7, and 40 were expressly uncertain. Thus, of the papers espousing a position on human-caused climate change, 97.1 percent supported human-caused climate change to some degree.

The consensus gets thin (33 percent) if you include papers that have no expressed position on human caused climate change; a point highlighted by climate economist Richard Tol. The new Cook et al. study, however, argues that it shouldn’t be a surprise that many papers on global warming do not assert human causation in their abstract, because it is already established knowledge. Why waste limited space calling out one’s support for the overwhelming consensus? If similar criteria were applied to geology, Cook and his coauthors argue, one might conclude that most geological papers do not take a firm position on the existence of plate tectonics and that there is no consensus about that matter.

If you restrict agreement to only those papers that quantitatively affirm that human activities drove most of the observed climate change (class 1, 65 papers), the consensus gets very small (< 2 percent), a fact highlighted by economists David Friedman and David Henderson. But that critique does not really challenge the conclusions of the Cook et al. study, only the phrasing or interpretation.

Moreover, it certainly doesn’t imply that 98 percent of the papers (or even a majority of the same) reject that conclusion. That observation is driven home by the dearth of peer-reviewed papers arguing that human activity is responsible for less than half of observed climate change. Only 10 papers in the Cook et al. rating quantitatively argued that human activities were responsible for less than half of warming. None of those are highly-cited or appear in major journals. And none of those ten were authored by the scientists that are frequently called to testify before Congress and cited by opponents to climate action (e.g. Roy Spencer, John Christy, Judith Curry, Richard Lindzen, Pat Michaels). While those scientists’ publications do not rate as strongly supporting the human influence on the climate, most did not explicitly reject that proposition in the Cook et al. 2013 analysis, and none did so quantitatively.

If climate skeptics are confident that human-caused climate change is small and favor internal variability to explain a large part of the observed warming, that is not well-documented in the literature (or it does not appear in many paper abstracts). If there is a strong, self-consistent, and quantitative case for recent warming having been caused by non-human influences on the climate, it is not being proclaimed in the literature.