Climate scientists' views on climate change: a survey

Hans von Storch and Dennis Bray

In 1996 and 2003 we surveyed the opinions on climate change held by climate scientists. The results of these surveys have been subject to many misuses and erroneous claims. Some have selected individual statements out of context (scroll down to number 5) to bolster their claims, while others have argued that the 2003 part of the survey would be strongly biased by skeptics misusing the online-sampling for multiple submissions.

With respect to the latter – the survey was conducted first in 1996 with a mail-out format, which nobody claimed could be biased and the results were published in the Bulletin of the American Meteorological Society. The second survey from 2003 was conducted on the internet, a procedure that in principle could have been misused by multiple submissions by those skeptical or alarmist on climate change who shared the password. However, the 2003 results are internally consistent with the 1996 results. In 2003 scientists expressed increased satisfaction and agreement with the IPCC and increased confidence in the tools of the science. In comparison to 1996, no anomalies were found in the response to questions.

On the skeptical side, the survey has often been used to create the impression that most scientists were not in support of anthropogenic causes of ongoing climate change: Specifically, it was noted that “For example more climate scientists ‘strongly disagree’ than ‘strongly agree’ that climate change is mostly the result of anthropogenic causes.” This interpretation is certainly biased.

We had requested responses on a scale from 1-7 to the question “Climate change is mostly the result of anthropogenic causes.” – with 1 representing “strong agreement” and 7 “strong disagreement”. Thus, scales 1-3 signal agreement, 4 an ambivalent position, and 5-7 disagreement. The frequency distribution for the two surveys in 1996 and 2003 are:

Thus, the statement, that more respondents strongly disagree than strongly agree is technically correct (10% vs. 9%), but highly misleading. If we pool the 1-3 positive responses to “agreement”, and 5-7 to disagreement, then the ratio in 1996 was 41:45 in favor of disagreement; in 2003, however, this ratio has become 56:30 in favor of agreement; all scales 1-3 have seen strong increases in frequency, while 5-6, with the notable exception of scale 7, have seen marked reductions.

Furthermore the question refers to “climate change” in general. We intended to ask for responses to the statement "Ongoing climate change is mostly the result of anthropogenic causes", but some respondents may have considered Holocene climate change in general. Thus, “disagreement” with the statement does not necessarily signal doubt about the perspective of a dominantly man-made climate change in the coming decades, but it mostly reveals an assessment of presently emerging climate change. The problem is that some commentators interpret our numbers as responses to "Future climate change is mostly the result of anthropogenic causes". "

One should also consider the possibility that scientists, who are naturally “plagued” or honored by doubts, may have a hard time choosing maximum confidence (i.e., category 1); therefore many may prefer a weaker “2”. (This tendency is obvious when we ask for assessment of the skill of models, and the respondents are hesitant to go for maximum confidence).

After careful quality checks, which lead to the identification of a few errors related to incorrect coding of answers, the responses of the 2003 survey have now recently been published.

We are presently preparing a new survey; this time, we will implement a more efficient barrier for manipulating response rates, even if we do not believe that this was really an issue in our 2003 survey. As the purpose of the survey is intended to be of service to the scientific community, we would openly welcome questions posed by the community to be added to the upcoming survey. However, please keep in mind that the survey poses general rather than specific questions and that the length of the survey is a major consideration.

Hans von Storch and Dennis Bray
Institute for Coastal Research
GKSS Research Center
Max Planck Str. 1
21502 Geesthacht