Queen Mary and Westfield College
School of Mathematical Sciences


Summer 1999

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,
Mathematics Building, Mile End Road, Queen Mary and Westfield College.

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
-------------------------------------------------------------------------

6 May 1999 E.J. Godolphin Connected Row-Column Designs.
              Royal Holloway U.of London


13 May 1999 D. A. Preece More dancing on Ramsgate sands:
              University of Kent New constructions for Rees
                                           neighbor designs with k = v.
                          

27 May 1999 C. Lewis Which Neighbour Design is Best?
              Queen Mary and
              Westfield College, London


1 July 1999 P. Druilhet Optimality of neighbour balanced
              Ecole Nationale de la designs.
              Statistique et de l'Analyse
              de l'Information,
              Rennes
_________________________________________________________________________

For more information ask:

Barbara Bogacka
School of Mathematical Sciences
Queen Mary and Westfield College
Mile End Road
London E1 4NS

Tel: 0171 975 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
-----------------

E.J. Godolphin
Connected Row-Column Designs

If a row-column design is disconnected, either through a poor
choice of design or because some observations go missing through some
mischance, then many treatment contrasts are inestimable. The problem of
determining the causes for a row-column design's disconnectivity has been
studied for several years. We consider this problem and show that
treatment disconnectivity is due to a partitioning of the treatments
which generalises the well-known but much simpler partitioning that always
occurs with incomplete block designs. A process which identifies these
partitionings is described and the procedure is illustrated by examples.
__________________________________________________________________________


D. A. Preece
More dancing on Ramsgate sands:
New constructions for Rees neighbor designs with k = v.

The simple Lucas/Walecki procedure for constructing neighbor designs
with k = v is well known and has been often rediscovered. There are
however various other simple procedures for constructing such designs.
These alternatives are not nearly so well known; indeed some of them
may not yet be in the literature, and the speaker does not know of a
proof of one of them. The procedures can be used equally to produce
row-complete Latin squares. Some of the procedures can be used to
produce Rees neighbor designs with v = 13 and so provide solutions of
Dudeney's problem involving 13 children who dance in rings on Ramsgate
Sands.
_______________________________________________________________________


C. Lewis
        Which Neighbour Design is Best?

Neighbour designs where the influence of neighbouring treatments
is limited to immediately prior and following plots, such as those
considered by Azais, Bailey and Monod (1993) for block designs,
and by Preece (1994) in circular block designs, are discussed for the
symmetric parameter set; namely that b=t and k=r.

By posing the question: "Which neighbour design is best?" three
distinct possibilities of neighbouring influence arise, whose subsequent
designs can be seen to originate from, or to be equivalent to, those
described in Azais et al.(1993) and Preece (1994). Each
of these types of neighbour design is detailed, and average efficiency
factors under the A-criterion are given where the neighbouring
effects have been accounted for.

Finally, by considering all the information presented, the original question
is answered, in a consistent manner with other areas of statistics.


Azais, J. M., Bailey, R.A. Monod, H., (1993).
A catalogue of Efficient Neighbour-Designs with Border Plots,
Biometrics, 49, 1252-1261.

Preece, D.A., (1994).
Balanced Ouchterlony Neighbour Designs and Quasi Rees Neighbour Designs,
Journal of Combinatorial Mathematics and Combinatorial Computing, 15,
197-219.
________________________________________________________________________






%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%