I'm curious how power analysis can be conducted when the scientist is stratifying on one or more variables?
I came across this approach but it went over my head. Ideally, the answer would not be "do X" but motivate for a scientist (not statistician) why we might formulate this question mathematically in a certain way.
I've seen that others are asking about bootstrapping as an option for analysis and essentially counting how many times the times the null would be rejected. This is appealing as simulations are familiar to me.
