| DATE: | Thursday, October 26, 2000 |
| TIME: | 4:30-5:30 PM |
| PLACE: | LN 2205 |
| SPEAKER: | Miguel Arcones, Binghamton |
| TITLE: | On The Asymptotic Accuracy Of The Bootstrap Under Arbitrary Resampling Size. |
The bootstrap is a recent statistical method consisting in resampling from the data. Usually the taken bootstrap sample size m is equal to the sample size n. Edgeworth expansions for the bootstrap under an arbitrary bootstrap sample size m are presented for the bootstrap of the sample mean and the bootstrap of quantiles. These results show that in the case of the sample mean the boostrap converges with a faster rate when m=n. However, in the case of the bootstrap of quantiles, the higher order of convergence is attained when m=n2/3. The talk will give a general introduction to the bootstrap.