I am reading ISLP chapter 3.2 about multiple linear regressions. F-statistic seems like to be a statistic calculated per sample. But also we know that it is just and estimation for the stat on whole population.
In that case, I wanted to verify this thing about F-distribution.
- Is F-distribution a sampling distribution of something over all possible sampling in given population ?
- Or is F-distribution a sampling distribution of something over all possible populations with?
I am not looking for a proof. Just some clarification.
