Hypoththsi testing

[fashepard]

David

How do I calculate the probability of committing a type II error. If the level of significance is 1% that is the probability of a type 1

 

[David Harper CFA FRM]

Hi Frank,

Hmmm...i've never done that, sorry. My understanding is that beta, P[Type II error], is non-trivial. It is typical to start from the "alpha side" (to calibrate significance/confidence where the null is presumed true, then minimize for the other error); e.g., selecting 95% confidence VaR = 1 - 5% significance.

So,
Alpha; i.e., Prob[Type I]) or confidence = 1 - significance of test; and
Beta; i.e., Prob[Type II]) = 1 - power of test

but even given alpha of, e.g., 99%, i think getting the beta (prob of type II) is non trivial. I don't know ...

(I've got a couple of customers who know more about this, i think, maybe they will help)

David