It's hard to affect policymakers with climate science information

Policymakers and climate negotiators tend to use scientific information in a very conservative way, hardly allowing it to dent their prior beliefs, according to an experiment conducted on a sample of 217 policymakers attending the Paris COP21 conference, more than half of them acting negotiators, including eight heads of delegations. Some presentation formats, though, seem to be more effective than others, depending on peculiar characteristics of the target audience.

"We tested how our sample update their beliefs on year 2100 global mean temperature increases in response to a statistical summary of climate models' forecasts," Bosetti, a professor at Bocconi University's Department of Economics and a fellow of Fondazione ENI Enrico Mattei, says. The same information was provided both to the sample of policymakers and negotiators and to a group of 140 European MBA students trained to play a country role in a climate negotiation simulation. While the prior beliefs of the two groups were similar, their estimates after the exposure to the scientific forecasts are considerably different, with policymakers and climate negotiators much more closer to their unconditional priors than the MBA students. Be it for a strong confidence in their priors or for a reluctance to report conditional probabilities that differ from their country's negotiation position, "the policymakers' reported conditional probabilities failed to fully incorporate the scientific information they received," the authors write.

The gap between policymakers' initial beliefs and scientific evidence, though, can be partially reduced by using an adequate presentation format. The scholars provided the information in three different formats, with different abundance of details, and while the format didn't affect MBA students, providing policymakers with the richest format, which includes individual model estimates in addition to the statistical range, increases the likelihood of reporting conditional probabilities closer to the scientific information.

"Our results," the authors conclude, "point to the importance of testing behavioral effects targeting the population of interest and suggest a more effective, and relatively easy to implement, format to visually communicate scientific information to policymakers."