Abstract:

     We present a method of constructing simultaneous confidence intervals with marginal equal tail probability for a mean vector. Estimation of sample mean value is obtained by resamplling. We further construct simultaneous confidence intervals of a mean vector with percentile interpolation and the estimation. We compare the method with Bonferroni, Efron and Normal Exact methods dependent on normal approximation. Simulation results show that the method is simple and feasible.

Key words: simultaneous confidence intervals, percentile, bootstrap