Thomas Lumley • 2/9/2019

What have I got against the Shapiro-Wilk test?

The article critiques the common use of the Shapiro-Wilk test for normality. It argues that testing for a Normal distribution is rarely the scientifically interesting question, unlike testing for power-law or Poisson distributions. The author explains that due to the Central Limit Theorem, approximate normality is common, making the test unnecessary for large samples and underpowered for small ones.

0 comments

#Statistical Testing #Normal Distribution #Shapiro Wilk Test