EPSRC CASE Studentship in Statistics

               Queen Mary, University of London

       Horiculture Research International, Wellesbourne

         Improving the Efficiency of Crop Experiments 
       by Using Variance Estimates from Historical Data


Applications are sought for this PhD studentship. You should have, or
expect to obtain in 2001, a good first degree (first or upper second
class) in statistics or mathematics, or an MSc in statistics or a
closely related subject. 

The successful applicant will join one of Europe's most prominent
research groups in the design of experiments, and a large and active
group of research students, in the School of Mathematical Sciences at
Queen Mary, University of London. You will work in collaboration with,
and spend some time in, the Biometrics Department at Horticulture
Research International, Wellesbourne, Warwick.

To apply for this studentship, contact:

Dr S. G. Gilmour
School of Mathematical Sciences
Queen Mary, University of London
Mile End Road
London E1 4NS

E-mail: [log in to unmask]
Phone: 020 7882 7833
Fax:   020 8981 9587

You should apply as soon as possible. 


Studentship:

This studentship is available to UK residents (and on a fees-only
basis to EU residents) to support study for a PhD on the project
described below. It covers, for three years of study, tuition fees
and, for UK residents, living expenses at the standard EPSRC rate for
London (£8500 in 2000/01) plus a contribution from the cooperating
body (£3000 in 2000/01).


Supervisors:

Dr S. G. Gilmour, School of Mathematical Sciences, Queen Mary,
University of London (www.maths.qmw.ac.uk/~sgg) 

R. N. Edmondson, Biometrics Department, Horticulture Research
International, Wellesbourne (www.hri.ac.uk/site2/research/fres.htm) 


Project:

The classical theory of designed experiments requires that every
experiment should provide its own estimate of variance. This
requirement was based on the assumption that experimental variability
may depend on a number of factors and that some factors such as
weather, soil type and management practice may vary between
experiments. However, practical experience has shown that the
variability of responses is often similar in similar experiments on
the same crop. The constraint that an experiment must provide a
reliable estimate of its own error variability was never practical
where facilities such as growth chambers, mushroom sheds, glasshouses
or shelf-life rooms were limited and economic constraints now mean
that field trials are also often too limited to meet this
requirement. 

You will make a retrospective study of the variability of a range of
field vegetable crop experiments, field flower experiments, glasshouse
vegetable experiments, mushroom experiments and growth chamber
experiments using the unique and extensive resources of archived crop
data available at Horticulture Research International. You will assess
and quantify the stability of variance estimates from similar crops
across a range of experimental conditions including seasons, locations
and management practices. For some crops, individual plant records are
available and for these crops you will study the effect of plot and
block size on crop variability. You will investigate a range of
statistical techniques for measuring stability of variances across
experiments and for combining variance estimates from historical data
to provide reliable variance estimates for current experiments based
on a substantial number of error degrees of freedom. Appropriate
combined estimates of sample variances across experiments will be
constructed and these combined estimates will be combined with the
estimates for current experiments to give reliable variance estimates
for current experiments. Appropriate distribution theory will be
developed for the combined variance estimates to provide reliable
statistical inference on treatment effects. The utility of historical
variance information for the weighted combination of treatment
estimates in experiments with stratified treatment information will be
investigated. 

This project is particularly appropriate for research training because
there are clear-cut practical problems for which the available methods
are no longer adequate. In working on this project, you will acquire a
good practical understanding of crop and field experiments. On
successful completion of the PhD, you will be ideally placed to work
in either industry or in academic research.

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

Dr Steven Gilmour

School of Mathematical Sciences
Queen Mary, University of London
Mile End Road
London E1 4NS
United Kingdom

Tel: +44 (0)20 7882 7833
Fax: +44 (0)20 8981 9587 (School fax, not private)
E-mail: [log in to unmask]

Web page: http://www.maths.qmul.ac.uk/~sgg