If I have a test and I repeated this test say n=2000
, then I will evaluate type I error while I am fixing the nominal level alpha =0.05
.
I found the type I error is different in some scenario, not exactly 0.05. i.e 0.035,0.04,0.045,0.055, or 0.06
? How can we decide when the observed type I error is close enough to 0.05?
Aucun commentaire:
Enregistrer un commentaire