so I've been looking into the Anderson Darling and Kologorov-Smirnov test because I have datasets with a large number of observations (~6000) and would like to fit distributions to them. They are continuous features and follow non-normal distributions.

But I don't know the "rules of thumb" for sample sizes here. I have a stats friend who say that KS-test will always fail for large sample sizes like this, but will still give a relatively correct answer when comparing different distributions.

However, my research online turned up the complete opposite answer in this site where they say that AD is good for sample sizes up to 1000 and after that KS-test is the way to go:


Which test is better at different sample sizes? Does anyone know any relevant literature which addresses this question?

Thank you in advance.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.