On the misuse of regression in earth science