I have a population of $1000$ elements. I need to take a sample $n$, and compare metrics before and after policy impact. What is the minimum sample size for $95%$ interval confidence?
Based on Cochran formula it looks to be $385$. But this looks a lot. Even if I adjust the $p$ coefficient the sample goes to $200$. But still looks a lot. Am I looking at this from the right angle?
What is the right way of calculating Minimum sample size for metrics difference?