How is causality possibly demonstrated in cases like this? It seems theoretically impossible to demonstrate causality here, there are a huge number of unknown variables in the system under study. So they must have just found a correlation in two data sets and best-effort ruled out known alternatives?<p>It bothers me how what actually happened (“correlation found”) gets translated into a story read by the public (“x causes y”) when the average person likely has no sense of the relative strength of different types of findings which are all messaged as “x causes y”.<p>This seems to happen a lot in climate science, cosmology and social science due to the physical or ethical impossibility of running controlled experiments. I really don’t think the general public understands the scientific method well enough that this distinction doesn’t need to be pointed out when results are published in the media.