> If we want to understand how the world works, we need to be able to construct mechanisms that explain why the results we measure in experiments come about.

This, 1000 times over. Unfortunately we seem to teach that statistical analysis is sufficient to infer causation in a lot our undergraduate science programs.