School of Mathematical Sciences
Queen Mary, University of London
Autumn 2001
STATISTICS SEMINAR: DESIGN OF EXPERIMENTS
All are welcome
The talks are held at 16.30, all in the Mathematics Seminar Room (103)
on Level 1 of the Mathematics Building, Queen Mary, University of London.
Tea and coffee are available in the Mathematics Common Room (102)
from 15.00.
The nearest underground station is Stepney Green.
Turn left at the exit and walk 400 yards.
________________________________________________________________________
DATE SPEAKER TITLE
------------------------------------------------------------------------
11 Oct 2001 M.A. Ollis Protection Against Premature
Queen Mary, Termination of Experiments Based
University of London on Williams Squares With Circular
Structure
18 Oct 2001 T.E. O'Brien Robust Experimental Design
Loyola University Chicago, Strategies for Dose Response
Katholieke Universiteit Models
Leuven
15 Nov 2001 S. Lewis and D. Woods Designing Experiments for
University of Southampton Semi-Parametric B-Spline Models
6 Dec 2001 B. Jones and S. Bate Universally Optimal
GlaxoSmithKline Cross-Over Designs when the
Number of Subjects and Periods
is a Multiple of the Number of
Treatments.
-----------------------------------------------------------------------
For more information ask:
Barbara Bogacka
School of Mathematical Sciences
Queen Mary, University of London
Mile End Road
London E1 4NS
Tel: 020 7882 5497
e-mail: [log in to unmask]
---------------------------------------------
The seminar information is kept on:
http://www.maths.qmw.ac.uk/~rab/seminars.html
______________________________________________________________________
A B S T R A C T S
______________________________________________________________________
M.A. Ollis
Protection Against Premature Termination of Experiments Based
on Williams Squares With Circular Structure
It is known that Williams squares with circular structure are E-optimal
in certain circumstances. We introduce a particular type of Williams
square with circular structure called a witch-hat square which may be
preferable to other Williams squares with circular structure if it is
possible that the experiment may be terminated prematurely. We show that
witch-hat squares exist for all n, where n is different from 4k, give
some enumeration results for small $n$ and report that a computer search
has shown that there are no witch-hat squares of order 8 (it
is known that there are no Williams squares with circular structure
of order 4).
-------------------------------------------------------------------------
T.E. O'Brien
Robust Experimental Design Strategies for Dose Response Models
In the context of nonlinear regression models, this paper outlines recent
developments in design strategies when the assumed model function, initial
parameter guesses and/or error structure are not known with complete
certainty. Designs obtained using these strategies are termed robust designs
as they are intended to be resistant to specified departures.
Robust designs are clearly advantageous in many practical settings since
these designs can be used to test for, say, lack of fit of the assumed model
function or for error heteroskedasticity, whereas so-called optimal designs
often cannot.
---------------------------------------------------------------------------
S. Lewis and D. Woods
Designing Experiments for Semi-Parametric B-Spline Models
The work discussed in this talk was motivated by the need to design
factorial experiments in industry for a response which may have several
turning points or exhibit non-smooth behaviour.
Flexible and parsimonious semi-parametric models for such a response will
be defined which are built from B-spline basis functions and monomials.
Prior information on anticipated behaviour of the response can be
incorporated into the model via choice of the location and the degree of
smoothness of the knots. An advantage of the models is that model-fitting
and inferential methods for general linear models can be applied.
An algorithmic approach to finding efficient designs under these models
will be outlined. The key issue of uncertainty in the locations of the
knots at the design stage will be addressed through two extensions of
standard design search criteria. These approaches will be compared and
contrasted over a variety of examples.
----------------------------------------------------------------------------
B. Jones and S. Bate
Universally Optimal Cross-Over Designs When the Number of Subjects
and Periods is a Multiple of the Number of Treatments
Abstract will be given later.
----------------------------------------------------------------------------
|