4 ms·

Are there actually two accepted ways of defining type I error rates? (Genuinely curious, I am not a statistician)