[Statlist] Spring Seminar of the SSS-ER, 14.05.04, Bern

[Riccardo Gatto]

==================================================================

  Spring Seminar of the Section Education & Research of the Swiss

                     Statistical Society

==================================================================

                     Friday May 14 2004

       University of Bern Insitute of Exact Sciences 
		Sidlerstrasse 5, 3012 Bern
	
                 Room B7 (2nd underground)

More details concerning the location can be found under:
www.bau.unibe.ch/fakplaene/phil_nat/statistik.html.

                    Beginning at 14:00

14:00   Speaker: Prof. J. Rice, University of California Berkley
        Title: Statistical Methods for Detecting Stellar Occultations by
        Kuiper Belt Objects: the Taiwanese-American Occultation Survey

14:50	Short break

15:00   Speaker: Dr. Tim Hesterberg, Insightful USA
        Title: Bootstrap for Statistics Teaching and Practice

15:50   Break (Drinks and snacks will be offered)

16:20   Speaker: Dr. Tim Hesterberg
        Title: Practical Issues in Resampling

17:00	End

All members of the Swiss Statistical Society and other interested
persons are kindly invited.

Riccardo Gatto (Representative of the Section Education & Research)
University of Bern
Department of Mathematical Statistics and Actuarial Science

==============================================================

Abstracts

Prof. J. Rice, University of California Berkley
Statistical Methods for Detecting Stellar Occultations by Kuiper Belt
Objects: the Taiwanese-American Occultation Survey

The Taiwanese-American  Occultation Survey (TAOS) will detect objects
in the Kuiper Belt,  a region of the solar system beyond the orbit of 
Neptune, by measuring
the rate of occultations of stars by
these objects, using an array of four 50cm wide-field robotic
telescopes located in the mountains of Taiwan. Thousands of stars will be 
monitored, resulting in
hundreds of millions of photometric measurements per night. To optimize
the success of TAOS, we have investigated various methods of gathering
and processing the data and have developed statistical methods for
detecting
occultations.  The resulting
estimated detection efficiencies will be used to guide the choice of
various operational parameters determining the mode of actual
observation when the telescopes come on line and begin routine
observations. In particular we show how real-time detection algorithms
may be constructed, taking advantage of having multiple telescopes.  We
also discuss a retrospective method for estimating the rate at which
occultations occur.


Dr. Tim Hesterberg, Insightful USA
Bootstrap for Statistics Teaching and Practice

There are two audiences for this talk:
(1) Educators and statistical consultants
who want to help students and clients understand statistical concepts.
Bootstrapping and permutation tests provide output we may graph in
familiar ways (like histograms) to help students and clients
understand sampling variability, standard errors, bias, p-values, and
the Central Limit Theorem (CLT)-not just in the abstract, but for the
data set and statistic at hand.

(2) Statistical practitioners,
who want to use resampling for
* freedom from assumptions like normality,
* checking whether normal approximations may be used
  (experienced statisticians may be surprised how inaccurate
  Normal-based methods are in the presence of skewness),
* higher accuracy than classical procedures,
* ease of handling different statistics,
* ease of handling different sampling designs.
Examples include telecommunications (where t-tests have Type I error
four times too large), and portfolio optimization (showing the high
variability and severe bias of standard portfolio optimization).

I'll demonstrate easy-to-use menu-driven software, in S-PLUS.


Dr. Tim Hesterberg, Insightful USA
Practical Issues in Resampling

I'll focus on some practical issues in resampling.
The first is that resampling is really only practical using
software that makes it easy.  I'll demonstrate S-PLUS software,
suitable for use by introductory statistics students on up.

A second practical issue is the number of bootstrap samples needed;
in particular you need a lot for BCa confidence intervals, 37 times fewer
for bootstrap tilting intervals.

A third practical issue is that resampling should mimic how the
data were collected.  This is not always straightforward.
For example, the naive way of doing finite-population bootstrapping
can result in greater variance than sampling without replacement.

The final issue is that bootstrapping underestimates standard
errors, substantially in the case of stratified sampling with small
strata.  There are four remedies, all relatively simple.