Let Fn denote the empirical distribution of a sample of size n from a distribution function F, and let Hn,t denote the distribution of n(Fn(t)-F(t)). Contrary to the intuition, Hn,t is not approximated uniformly over t by its bootstrapped counterpart. The main problem is at the values of t near the tails of F. Two results exploring this phenomenon are presented here.
All Science Journal Classification (ASJC) codes
- Statistics and Probability
- Statistics, Probability and Uncertainty
- Applied Mathematics