This talk introduces Efron's original nonparametric bootstrap and the subsampling bootstrap (Politis and Romano). It does not require knowledge of theoretical statistics. Topics covered include
* bootstrap t confidence intervals,
* bootstrap percentile confidence intervals,
* second order correct procedures,
* why naive bootstrap procedures do not work for hypothesis tests and what to do about it (invert a bootstrap confidence interval),
* why naive bootstrap procedures do not work for regression and what to do about it (bootstrap residuals), and
* the subsampling bootstrap for time series and other applications for which Efron's original nonparametric bootstrap fails.
O. Behnke, L, Brenner, L. Lyons, N. Wardle, S. Algeri