Many statistical tests assume that data (or residuals) are sampled from a Gaussian
distribution. Normality tests are often used to test for evidence that the distribution
is not Gaussian, but these tests are less helpful than many scientists expect them to
be.