Title: Bootstrapping for Teaching Statistics

Speaker: Tim Hesterberg, Insightful Corporation

Abstract: 
This talk is intended for those who want to take advantage
of statistical computing to improve statistics teaching.
Statistical concepts such as sampling distributions, standard errors,
and P-values are difficult for many students. 
Bootstrapping and permutation tests give students hands-on
experience illustrating those abstract concepts.
I'll demonstrate, using examples from the new "Bootstrap
Methods and Permutation Tests" chapter for Moore & McCabe,
Introduction to the Practice of Statistics.  

A suitable computing environment is needed to make resampling
convenient.  For introductory statistics, a graphical user interface is
essential.  I'll demonstrate the S+Resample library, and its GUI.
The library works with S-PLUS, including the free student version.

The examples also demonstrate a major weakness in current statistical
practice, the use of Gaussian-based methods without doing diagnostics
to check the accuracy of those methods.  In one of our examples the
P-values produced by standard t tests are off by a factor of four!
The old pre-computer rule of using Gaussian-based methods if n > 30
should be replaced by resampling diagnostics.