Homework Help: Bernoulli confidence intervals

1. Aug 1, 2010

Gekko

Confidence intervals

1. The problem statement, all variables and given/known data[/

Use CLT to construct approximate symmetric 100(1-alpha)% confidence interval [L,R] for p then show that [L,1] is then an approximate 100(1-alpha/2)% confidence interval for p

3. The attempt at a solution

When [L,1] then we have a one sided confidence interval.
What we effectively need to show is that the area under the normal curve from -inf to alpha/2 is equal to the area under the curve from -inf to alpha/4 + the area under the curve from alpha/4 to inf
I was going to look at the error function as a way to solve this. I havent managed it though. Any thoughts?

Last edited: Aug 2, 2010
2. Aug 2, 2010

Gekko

This isn't related to Bernoulli. Thanks to CLT we assume normality so the question is how to formally prove the above in a mathematical way.
Can we just use the CDF?