标准误和可信区间的Bootstrap法及其计算机实现

Bootstrap Methods for Standard Error, Confidence Intervals and Their Computer Practices

本文介绍由B.Efron所创立的Bootstrap法,包括Bootstrap基本概念,Monte Carlo算法的标准误和Bootstrap的百分位数法和纠偏百分位数法所计算的可信区间。通过传统统计学的标准区间和Bootstrap区间比较,说明Bootstrap法的准确性和优点。作者用自编的由BASIC语言和SPSS语言相互连接的计算机程序来实现Bootstrap的过程。

This article introduced bootstrap method created by B. Efron, including bootstrap basic idea, bootstrap standard error for Monte Carlo algorithm and bootstrap confidence intervals for percentile method and bias-corrected percentile method. By comparihg standard interval of traditional statistics with bootstrap intervals, it illustrated statistical accuracy and advantage of bootstrap method. Author practised bootstrap procedure with compute programmes written by himself in which BASIC language connected with SPSS language.