In power analysis, why do we set alpha at 0.05 and beta at 0.8? Is there any instance where we would want to increase or decrease those numbers?

Question

In power analysis, why do we set alpha at 0.05 and beta at 0.8? Is there any instance where we would want to increase or decrease those numbers?

In: 4

Answer 1

Yes to both and there can be industry standards but these numbers are good cause they don’t require a ton of testing be poduce good results.

Let’s take alpha. If we have a pcalue less than alpha, we treat that as essentially there’s a 95% or more chance that we found the null hypothesis is false.

If you were a casino, and you wanted to test dice go see if they were fair and random, you may want a smaller alpha to make sure that there’s no bias in the dice. It’s cheap to throw away dice that don’t pass, but if you test them you may want to be even more sure that the results are truly random