Trust and science

Another point Kling makes in this piece on trust has to do with science.

For information, it is easiest to trust scientific research when the methods seem capable of producing reliable results and the conclusions have little political significance. Many people distrust politically loaded scientific research because they do not like the conclusions, regardless of the quality of that research. That common form of bias makes it difficult even for reliable knowledge to influence people's opinions. Because economics often has political implications, it is common to distrust economic analysis, even when its conclusions are arrived at by careful reasoning and empirical study.

When the conclusions are politically loaded, it makes sense to check the methodology carefully, regardless of whether or not they support your personal views. Careful research deserves consideration, but unreliable research should be given less weight.

Of course, most people lack the background to evaluate scientific research methods. To me, this suggests that scientists ought to try their best to explain their reasoning to an intelligent lay audience. This means explaining which experiments or observations convince a scientist of the validity of a theory, and why this evidence is compelling. It also suggests that it is important that for high school and college students to learn statistics and research methods.

I have difficulty trusting climate models. If a scientist wants me to believe that a particular climate model is reliable, then I would like to read an essay written for an intelligent layman explaining the tests that the model was able to pass, and how these tests serve to rule out alternative hypotheses about the phenomena explained.

I understand that a model will have some uncertainty surrounding its predictions and properties. In fact, I would expect considerable uncertainty. I would be more comfortable if scientists spelled out the uncertainties than if they glossed over them.

My impression is that no single climate model enjoys the confidence of a large number of scientists. Instead, many climate scientists are willing to endorse a "consensus" that takes a range of estimates from some models. I would like to read an essay written for an intelligent layman that explains why this is a persuasive approach. What is the rationale for including some models while excluding others? Do predictions based on the "average" or "consensus" model out-perform the predictions of any individual model, as in a "wisdom of crowds" phenomenon? Or is the purpose of a "consensus" is to strengthen a political coalition, rather than to improve accuracy?