Bray and von Storch just published the results of their latest survey of climate scientists. It contains lots of interesting and very detailed information, though some questions are a little biased in my opinion. Still, they find a strong consensus on human causation of climate change: 87.4% of respondents are to some extent convinced that most of recent or near future climate change is, or will be, the result of anthropogenic causes (question v007). Responses were given on a scale from 1 (not at all) to 7 (very much). In line with Bray (2010) a response between 5 and 7 is considered agreement with anthropogenic causation. In their 2008 survey the level of agreement based on the same question was 83.5% and in 2013 it was 80.9%.
Question v013 asked a somewhat similar question as we did in our 2012 climate survey, namely the percentage of global warming that is attributable to human activities:
84.2% of respondents picked one of the two answer options that correspond to the canonical “more than half” or “most” of global warming that according to the IPCC is human caused. However, the corresponding IPCC statement is with regards to warming since the 1950’s, about which there is a lot more confidence, whereas this question specifies the warming since 1850.
But wait a moment, hasn’t the earth warmed a lot more than 0.5-0.7 degrees C since 1850? Yes, it definitely has; we’ve recently breached the 1 degree mark relative to the 1850-1880 average, so the range given in their question is quite outdated. A defensible choice at the time of drafting the survey would have been to quote the latest IPCC number of 0.85 (0.65 to 1.06) degrees warming over the time period 1880-2012, even if current temperatures have gone up sharply since then.
Global average surface temperature relative to the 1850-1880 mean. Last annual average shown is 2015; if the first few months of 2016 are a guide, the vertical scale might have to be adapted for 2016. Figure by Jos Hagelaars.
Moreover, the answer options for v013 do not cover the full range of possibilities. Natural factors could have caused warming or cooling. Imagine that natural factors would have caused a cooling of 0.1 degrees C since pre-industrial times (which is not at all implausible), then to achieve closure with the observed warming of 1.0 degrees, anthropogenic factors should have contributed 1.1 degrees, or 110% of the observed warming. We discussed this argument in detail in our ES&T paper emanating from the climate science survey we conducted in 2012.
I emailed Dennis Bray about these and other issues after having responded to their survey back in 2015. He defended their choice of lowballing the observed warming as being consistent with their previous surveys and not being much different from more recent, and also likely contested, estimates. Strangely, he disagreed with the possibility of a factor being responsible for more than 100% of the observed warming, even in the hypothetical example above.
Cook et al (myself included) recently wrote an article in which we reviewed the existing ‘consensus’ estimates. This latest Bray and von Storch survey finds a level consensus on attribution that is consistent with other studies, though towards the lower end of the range. From their description I don’t think there is a bias in their sample of scientists, though there is always the possibility of self-selection, where people might be more likely to respond to a survey if it originates from a source who they perceive to be credible. Repeatedly, surveys have found that the level of consensus goes up as you zoom in to a sample of scientists with more relevant expertise. The Bray and von Storch results, as are ours, are mostly representative of a broad group of climate related scientists.
A detail of particular interest to me is that the survey questions included the response option “no answer”. That explains the different sample size for different questions (“Number of obs”). It’s probably no coincidence that question v013 (asking for a specific range of percent contribution) has a smaller sample size (n=587), and by inference more “no answer” responses, than the other, but simpler, attribution question v007 (n=640). This is consistent with what we found in our 2012 climate science survey: fewer respondents picked a specific percentage range of attribution compared to providing a qualitative judgment thereof. Though admittedly “no answer” (in the Bray and von Storch survey) is less ambiguous in this context than “I don’t know”, “unknown” or “other” (in our survey).
Amidst the questions on science and society I perceived some questions to have an “anti-consensus” (v069) or “anti-alarmist” (v067) tone to it, but there were no questions asking for mirror image perceptions. Doomsday stories need to be investigated before they get out of hand (v067): of course. But no question was asked whether stories downplaying a scientifically established risk should be investigated. I would have likewise responded: of course. To his credit, Dennis Bray acknowledged in his email that this was an oversight on their part.