Dear All,
Please see details below of an ESRC funded statistics PhD project at the
University of Bristol. Please note the 7th November closing date.
Best wishes,
Bill Browne.
---------------------------------------------
ESRC 'e-STAT' PhD Studentship
Using simulation and parallel computing for faster sample size calculations
in complex random effect models
Social Statistics PhD
Centre for Multilevel Modelling, University of Bristol
This PhD will form part of an ESRC-funded project, 'e-STAT' funded as a new
Node under the National Centre for E-Social Science. The 'e-STAT' project
is developing generic statistical multilevel modelling methodology for
complex modelling across a range of applications in the social sciences.
Estimation algorithms for the models are being developed using MCMC
techniques.
A brief description of the project is as follows:
It is well known that the dependence induced by clustering in social
science datasets means that the sample size requirements for testing
hypotheses need inflating to account for the lack of independence. In a
recently completed ESRC funded project a piece of software, MLPOWSIM has
been developed that will generate both MLwiN macro code and R code to
perform sample size calculations for a selection of multilevel nested and
crossed designs. The software is limited to 2 separate clustering factors
(whether they be
nested or crossed) and it is challenging to consider how to simulate
realistic datasets for specific crossed scenarios.
For example in education a study of the effects of schools and
neighbourhoods on achievement would typically consider sampling groups of
pupils within a selection of schools with the neighbourhood identifiers
being collected but not part of the sampling design. Most sample size
calculations assume a balanced sampling design but it will
clearly not be possible to sample exactly 10 pupils from each school and
neighbourhood combination as some schools will not teach pupils from some
neighbourhoods and so using simulation to generate samples from a two way
table of pupil counts within schools by neighbourhoods is a potential
solution.
This has been done in MLPOWSIM for the case of 2 clustering factors but in
reality datasets exist with many classifications. For example in recent
work Leckie (2008) examined using the National pupil database to look at
accounting for the effects of primary school, secondary school,
neighbourhood on student achievement while also controlling for student
mobility. This database contains nearly half a million pupil records which
makes using the whole database rather unwieldy. There is therefore a need
to come up with simulation-based approaches for generating appropriate
samples from such a large database to establish potential sample sizes and
sampling schemes that will capture the basic structure of the data and have
appropriate power for testing hypotheses both using the current data and
future years of data. This challenge is the motivation for the PhD project.
One also needs to investigate how to come up with parameter values to be
used in the simulated designs and how sensitive the sample size estimates
are to these values. Another factor in increasing the number of
classifications is that the estimation options for the models become more
limited, and in particular Monte Carlo Markov chain (MCMC) estimation
becomes more commonly the method used. MCMC estimation is inherently slow
as it can involve running for many iterations of the algorithm. When
combined with the simulation approach of producing thousands of datasets
the time to construct a sample size calculation becomes very large. Here
parallel computing can very easily cut the time as it is possible to send
different datasets to different processors and link the results together at
the end. It is also possible that parallel computing may well speed up
individual runs of the MCMC algorithm by parallelizing steps of the MCMC
algorithm.
For UK nationals, the studentship will attract an annual stipend of
£12,600 in addition to tuition fees
The Centre for Multilevel Modelling is a multidisciplinary cross-department
research centre based mainly in the Graduate School of Education (GSOE) but
with researchers in both Geography and Clinical Veterinary Science. The
internationally-renowned multilevel modelling research team are involved in
advancing statistical methods for analysing large-scale datasets and
investigate issues such as school effectiveness, the effect of school
resources on pupil attainment and family effects on child development. The
Centre also produces the software 'MLwiN'.
The student would be jointly supervised by Professor William Browne and
Professor Jon
Rasbash. Professor Browne is based in the Department of Clinical Veterinary
Science whilst
Professor Rasbash is the director of the Centre for Multilevel Modelling
and is based in the GSOE. The student will formally be enrolled on the PhD
within the GSOE but will physically be based at the vet school in Lower
Langford with Professor Browne and his researchers.
The team at the Centre for Multilevel modelling includes applied
statisticians, theoretical statisticians, software engineers, computational
statisticians, psychologists, geographers and educationalists. We provide a
lively research environment and can offer support for applicants wishing to
develop statistical methodology skills and applied statistical modelling
skills. The team has considerable expertise in statistical programming and
software engineering so additional support can be given to candidates
wishing to improve skills in these further two areas.
Applicants should have a masters with a high content of applied statistics
including
• methods of research design and data collection
• methods of data analysis
• a critical perspective on how statistical methods are used to address
substantive research issues in the social sciences
For more information of these criteria see section F18 subsection 4.1.2 of
the ESRC Postgraduate Training Guidelines
http://www.esrc.ac.uk/ESRCInfoCentre/Images/Postgraduate_Training_Guideline
s_2005_tcm6-9062.pdf
Bristol is a vibrant city with a large student population and excellent
transport links.
Applicants: It is anticipated that the studentship will begin as soon as
possible and the latest in January 2010.
For further information in advance of submitting an application, email
Professor Bill Browne ([log in to unmask]) or Professor Jon
Rasbash( [log in to unmask]).
To apply for this studentship online go to
http://www.bris.ac.uk/prospectus/postgraduate/2010/intro/8
(Please note your interest in this studentship in the 'Funding' section of
the application form by using the reference NCeSS).
For information about the application process, contact
[log in to unmask], using the reference NCeSS.
The closing date is 6th November 2009.
http://www.cmm.bristol.ac.uk/ http://www.bristol.ac.uk/education/
----------------------
WJ Browne
[log in to unmask]
|