Instead of taking the square of the difference, the first order correlation can be used directly to give: