or a "two-hand" regression. (take a pure noise scatter plot and put one hand over the upper-left and the other over the lower-right to help your audience "ignor

or a "two-hand" regression. (take a pure noise scatter plot and put one hand over the upper-left and the other over the lower-right to help your audience "ignore the outliers", and voila, a perfect linear correlation!)