When Scientists Disagree: How We Frame Uncertainty Influences Public Tust of Science

Climate change science is inherently uncertain: Will rainfall increase or decrease in a particular region? By how many inches will sea level rise? Scientists, policy-makers and educators face a conundrum when communicating climate change predictions: How to accurately present possible future outcomes while also acknowledging the uncertainty inherent in such estimates. For example, to communicate the extent to which sea levels may rise as a result of climate change, one could a) present different scientists’ opinions (some scientists think it will rise by 4 inches, some think it will rise by 8 inches, etc.), b) give an average estimate of scientists’ predictions (sea level will rise by about 10 inches), or c) give a range estimate of scientists’ predictions (sea level will rise by between 4 and 16 inches).  We explored how framing the message in these ways affects public trust of climate science. We hypothesized that those presented with scientists’ differing predictions would trust the science and scientists less than those presented with an average or range of their predictions. Participants read a short paragraph stating that 98% of the world’s scientists agreed that sea level will rise over the next 50 years as a result of climate change. Participants were then randomly assigned to receive information about scientists’ estimates of the magnitude of sea level change (identical in all conditions) in one of three forms—focusing on the scientists’ differing estimates, on the average of their estimates, or on the range of their estimates. We then measured participants’ level of trust in these scientists and their belief in climate change. We found that participants’ trust in the science and their belief in climate change were about the same for the average and range conditions, but significantly lower in the differing estimates condition.  However, these effects were moderated by political orientation: Liberals showed little difference in their reported trust across conditions, whereas conservatives reported significantly lower trust in the differing estimates condition. The way that differing predictions among scientists is presented appears to influence how much the public trusts the source of that information.