Anderson-Darling test

normality test with weight on the tails

A goodness-of-fit test that measures how far the empirical distribution is from a fitted distribution, giving extra weight to the tails. A p-value below 0.05 means the normal model is questionable, which matters for capability estimates.