Estimation of the distribution function of a standardized statistic

• 成果类型: Article
• 署名作者: Lee, SMS; Young, GA
• 署名单位: University of Cambridge; University of Hong Kong
• 刊物名称: JOURNAL OF THE ROYAL STATISTICAL SOCIETY SERIES B-STATISTICAL METHODOLOGY
• ISSN/ISSBN: 1369-7412
• DOI: 10.1111/1467-9868.00074
• 发表日期: 1997
• 页码: 383-400
• 关键词: edgeworth expansion; bootstrap; approximations; simulation; ORDER

摘要

For estimating the distribution of a standardized statistic, the bootstrap estimate is known to be local asymptotic minimax. Various computational techniques have been developed to improve on the simulation efficiency of uniform resampling, the standard Monte Carlo approach to approximating the bootstrap estimate. Two new approaches are proposed which give accurate yet simple approximations to the bootstrap estimate. The second of the approaches even improves the convergence rate of the simulation error. A simulation study examines the performance of these two approaches in comparison with other modified bootstrap estimates.