The regular precentile confidence interval for bootstrap result is easy, we just pick alpha/2 and 1 - alpha/2 quantile of estrimated theta_hat* sequence.
And another kind of confidence interval is called pivotal interval, which could makes our confidence interval centered at theta_hat_n, which is estimated by the total observations.
Further to see the derivation below.
Here we illustrate two different kinds of confidence interval, blue is precentile CI, and red one is so called pivotal CI.
Source: All of Statistics--a concise in statistical inference , P110
Commentaires