Organisers: Theresa Nunan, Fiona Beer, Stephen Bush, Richard Hutchinson, Derchieh Hung, Jason Rabbitt, Rian Caccianiga, Damien Kennedy, Kevin Wang

A Young Statistician is someone who is studying statistics or has graduated in the last 5 years and is working or studying in a statistics related area (or wants to be!) You don't have to be young to be a Young Statistician, just new to the profession.

The primary goal of these sessions at ASC2008 is to assist and early career statisticians in building their statistical networks while investigating the different roles that statisticians fill in various industries. A mix of experienced and emerging statisticians will give a perspective of their role in their organisations, and how statistical methods are central to the decision making process. Although aimed at assisting Young Statisticians, all are welcome to attend!

Invited Speakers

A series of speakers have been invited to present in the Young Statisticians sessions from a variety of statistical professions. the aim of each of these presentations is to give attendees a taste of working in each of these industries, the wort of work that is done, the people who the statistician works with on a day-to-day basis, and even some pointers for those who are interested in pursuing a career in the industry.

Confirmed speakers:

Networking Session

This session will enable participants to engage in discussion with other statisticians on topics relevant to early career statisticians and the Society. This session will also give the attendees the opportunity to meet the Young Statistician representatives for each state, and the National Section Chairs.

One of the goals of this session will be to investigate methods of engaging early career statisticians in Society activities, and to look into ways the Society can assist early career statisticians in gaining a foothold in the profession.

GStat Presentation

An important function of the Society is the accreditation of its members. Accreditation enables those in the wider community to recognise the qualifications and experience of statisticians, and be sure of the quality of the work performed. Within the statistical community, it enables the practitioners to be recognised for their experience and training. The GStat accreditation is appropriate for Young Statisticians who have completed their studies and have started their career in the statistical community. There will be a presentation of awards to those attendees who have successfully applied for accreditation prior to the conference.

Note that applications received after April 2008 are not guaranteed to be approved prior to the conference.

Poster Session

Young and Early Career Statisticians are encouraged to take advantage of the chance of presenting their work in the ASC poster sessions. Data Analysis Australia is kindly sponsoring prizes for the best posters submitted by a young or early career statistician. The posters will be judged by the invited speakers, and the prizes awarded at the end of the last YS session on Monday.

Young Statisticians Dinner

A Young Statisticians dinner will be held on Monday 30 June (venue to be confirmed), providing an opportunity for young statisticians to network and socialise in an informal setting. The cost of the dinner will be partly subsidised through generous sponsorship by Data Analysis Australia and the Australian Pharmaceutical Biostatistics Group. If you are a young or early career statistician and are interested in attending, please RSVP by COB 9th June to derchieh.hung@csl.com.au. If you have any questions, please contact Derchieh on 9389 1772.

For further information on these sessions please contact Theresa Nunan (theresa.nunan@abs.gov.au), Fiona Beer (beerf@amgen.com) or Stephen Bush (stephen.bush@uts.edu.au).

STOCHASTIC MODELLING AND STATISTICS IN THE SOCIAL SCIENCES

Organisers: Ross Maller, Alan Herning,

Veronica Rodriguez, Geoff Lee

Keynote speaker: Fiona Steele

Organisation: Centre for Multilevel Modelling University of Bristol

Web: www.cmm.bristol.ac.uk/team/fiona.shtml

The Stochastic Modelling and Statistics in the Social Sciences theme at the ASC2008 will focus on the linking and analysis of Longitudinal data. Data linkage and longitude surveys are becoming more popular in Australia and New Zealand as analysts are seeking to monitor and analyse a wider range of characteristics of the population over longer periods of time.

The keynote speaker for this theme, Dr Fiona Steele from the University of Bristol, UK, will present the 2008 E.K. Foreman Lecture. The Foreman Lecture is an annual event honouring the founder of the Methodology Section of the Australian Bureau of Statistics (ABS). Dr Fiona Steele has undertaken extensive work in the modelling of longitudinal data and has taught a number of short courses in multilevel modelling and event history analysis in the UK and overseas.

Invited speakers for this theme will include Prof. Steve Zubrick (Institute for Child Health Research), who will be presenting a talk on his use of Western Australian data linkage capacities in extending and value adding to cross-sectional population household surveys of family and child development. Vince Galvin from Statistics New Zealand will be discussing his investigations into the New Zealand Linked Employer-Employee Dataset. Glenys Bishop from the Australian Bureau of Statistics will discuss her experiences with longitudinal data assembly and analysis as part of the Census Data Enhancement Project.

For further information on these sessions please contact Alan Herning (mailto:alan.herning@abs.gov.au).

FINANCIAL STATISTICS

Organiser: Fima Klebaner

Finance applications have given tremendous impetus to research in probability theory and statistics over the last few decades. The hugely successful Black-Scholes model, which relies on an assumption of the log-normality of stock price processes, together with extensions of it to various other stochastic processes, has revolutionised option pricing, and indeed the whole area of asset pricing models. The statistical and econometric methodologies associated with fitting, calibrating and testing these models have, correspondingly, grown enormously in scope and sophistication. Related financial activities such as hedging strategies, portfolio allocation and assessment, efficient market theories, etc., also rely very heavily on statistical methods, many newly developed for the particular requirements of the discipline. Actuarial and risk analysis techniques have likewise undergone revolutionary changes over the period. The Statistics in Finance session will touch on a few of these topics, aiming to give a flavour of some of the current concerns.

STATISTICAL METHODOLOGY

Organisers: Michael Martin, Steve Roberts

Keynote speaker: Alastair Young

Organisation: Department of Mathematics Imperial College, London.

Web: www3.imperial.ac.uk/people/alastair.young

Prior to taking up a Chair in Statistics at Imperial in 2005, Alastair Young was Reader in Methodological Statistics at the Statistical Laboratory, University of Cambridge. His research interests are in contemporary, computer-intensive approaches to frequentist statistics, with particular focus on bootstrap and related approaches to parametric and non-parametric inference. He also is very active in the development of approximation methods, such as saddlepoint procedures, in statistics, and in development of inference procedures for spatial data. Alastair is a Fellow of the Institute of Mathematical Statistics, and former Joint Editor of the Journal of the Royal Statistical Society, Series B.

Alistair is co-author (with R.L. Smith) of the advanced undergraduate text “Essentials of statistical inference” published by Cambridge University Press (2005).

MEDICAL STATISTICS

Organisers: John Carlin, Andrew Forbes (on behalf of the Biostatistics Collaboration of Australia)

Keynote speaker: Simon Thompson

Organisation: Director, MRC Biostatistics Unit Cambridge, UK

Web: www.mrc-bsu.cam.ac.uk/BSUsite/AboutUs/People/simon.xml

When synthesising evidence from multiple sources to address a question in health care, the original studies almost always differ in rigour and in relevance to the target question. Rigour (or internal bias) reflects how well a study estimates its intended parameters, and varies according to use of randomisation, adequacy of allocation concealment, degree of blinding and attrition levels. Relevance (or external bias)
reflects how similar the source study design is to the target setting, with respect to study population, outcomes measured and details of interventions. The usual approach to addressing differences in rigour and relevance is to choose a minimum standard for inclusion in the evidence synthesis.

This is problematic in that potentially useful evidence may be discarded and no allowance is made for the biases present in the included studies. Here we propose simple methods of allowing for internal and external biases in evidence synthesis. We develop our methods in the context of an example in antenatal care, which concerns the effectiveness of anti-D prophylaxis in pregnant women who are rhesus negative. The purpose of anti-D prophylaxis is to prevent sensitisation of the mother and haemolytic disease of the newborn in subsequent pregnancies. An existing systematic review identified 11 relevant studies. Many were historically controlled, only one was a randomised trial, and doses, populations and outcomes varied between studies and differed from our target UK setting.

We construct prior distributions to represent the bias in each study, for specified independent sources of internal and external bias. The magnitude and uncertainty of each bias are based on empirical data where available, and on elicited expert opinion. We adjust for biases in a meta-analysis of the anti-D prophylaxis studies, using a straightforward analytical approach. Differences between the bias-adjusted metaanalysis results, and the unadjusted meta-analysis results are explored. We conclude that a generic bias modelling approach is beneficial in enabling analysis of all relevant data, with bias adjustment possible using computationally simple methods.

Keynote speaker: Miguel Hernan

Organisation: Department of Epidemiology Harvard School of Public Health

Web: www.hsph.harvard.edu/faculty/miguel-hernan/

The title of his keynote talk will be: "Observational studies analysed like randomised experiments: the case of postmenopausal hormone therapy and heart disease".

The discrepancy between the results from randomised and observational studies of hormone therapy and heart disease has become the paradigmatic example of why epidemiologic studies are not to be trusted. A Women's Health Initiative randomised trial found that the incidence of coronary heart disease (CHD) was 24% greater in postmenopausal women assigned to receive estrogen plus progestin compared with those assigned to receive placebo. In contrast, large observational studies had previously found a reduced 30-40% risk of CHD among hormone users compared with non users. Two of the largest observational studies were based on the Nurses' Health Study cohort in the U.S. and the General Practice Research Database in Europe.

Miguel will argue that much of the discrepancy arose because the randomised and observational studies estimated different parameters. He will describe a strategy to obtain effect estimates of hormone therapy that are directly comparable to those of the randomised trial, except for the fact that hormone therapy was not randomly assigned in the observational studies. Re-analysis of the observational data yields estimates of the effect of hormone therapy on CHD that are closer to those from the randomised trial.

BIOINFORMATICS AND STATISTICAL GENETICS

Organisers: Susan Wilson, Melanie Bahlo, Brian Cullis, Geoff McLachlan

Keynote speaker: Elizabeth Thompson

Organisation: Department of Statistics University of Washington

Web: www.stat.washington.edu/thompson/

The quantities of biological information and data being collected are increasing at a phenomenal rate, and at all levels of biological organisation, thus enabling more realistic modelling of biological systems. Consequently, these information databases are opening up quantitative challenges that are both fundamentally important and huge in scale.

Elizabeth Thompson is Professor of Statistics at the University of Washington (Seattle) as well as Adjunct Professor in Genome Sciences and Biostatistics. Her research interest is in the development of methods for inference from genetic data, and particularly from data observed on large and complex pedigree structures.

Questions of interest range from analyses of long-term gene frequency differentiation in widely dispersed populations, to short-term extinction of genes in the small population of a highly endangered species; from inference of genealogical relationships among individuals to inference of the genetic basis of traits from data observed on members of a known pedigree; and from analyses of patterns of genome sharing in plants to modern methods for human linkage analysis. In recent years, several of these questions have been addressed using Monte Carlo likelihood.

Elizabeth has won many prizes and awards, and given named, prestigious lectures in Europe and USA.

OZCOTS 2008 - AUSTRALIAN CONFERENCE ON TEACHING STATISTICS 2008

Please contact Brian Phillips, mailto:bphillips@swin.edu.au or Helen MacGillivray, h.macgillivray@qut.edu.au. for any OZCOTS 2008 queries.

- Date: Thursday 3 July and Friday 4 July
- Location: Sofitel Melbourne CBD

Statistical education is of vital significance to all statisticians and the statistical profession across the spectrum of educational levels and disciplines. The supply of statisticians is just one aspect of this spectrum, and the education of future consumers, users, producers, developers and researchers of statistics is both challenging and important for a modern information society and hence to the statistical profession.

A 2-day satellite OZCOTS (Australian Conference on Teaching Statistics) will be held at the Sofitel Hotel Melbourne on Thursday 3rd July and Friday 4th July 2008.

Invited and contributed papers on topics across the statistical education spectrum will be of interest to statisticians, statistical educators and the statistical profession.

OZCOTS 2008 combines the interests of the Statistics Education Section of the Statistical Society of Australia Inc with one of the activities of Professor Helen MacGillivray’s Carrick Senior Fellowship program entitled “The teaching and assessment of statistical thinking within and across disciplines”. The Australian Carrick Institute was established in late 2004 and its mission is to promote and advance learning and teaching in Australian higher education. The Carrick Fellowship Scheme supports leading educators to undertake strategic, high profile fellowship activities in areas that support this mission. Senior Fellows undertake a full time program of highly strategic fellowship activities over one year, and Helen’s senior fellowship is one of only four awarded in two years across all disciplines. Helen is also President-elect of IASE (International Association for Statistics Education).

The OZCOTS satellite will include presentations from members of the Fellowship’s international collaborative team from the UK, USA, Canada, New Zealand, South Africa and Australia, as well as from The Senior Fellow. Discussion and brainstorming forums are planned on resources and resource needs of tertiary teachers of statistics, and on the national and international scene in learning, teaching and assessment in developing statistical thinking. Anyone interested in presenting a contributed paper at OZCOTS 2008 should contact Brian Phillips, mailto:bphillips@swin.edu.au or Helen MacGillivray, h.macgillivray@qut.edu.au.

ASC2008 delegates will be able to attend OZCOTS 2008 on Thursday 3 July 2008, at no extra cost, provided that this is indicated on the ASC2008 registration form.

For full details of OZCOTS2008 please visit the OZCOTS2008 website. http://silmaril.math.sci.qut.edu.au/ozcots2008.

SATELLITE WORKSHOP ON COMPUTING WITH R

On Saturday 28th June and Sunday 29th June 2008, there will be a satellite workshop on the use of the computer package R, covering a wide variety of statistical areas. Each day will be made up of four or five sessions, each focussing on a particular topic. The topics will be selected from:

• Graphics – lattice and ggplot
• Classification models and visualisation of results
• Rattle
• GLMs
• Small-sample asymptotics; the hoa package
• Multi-level models
• Spatial statistics
• BRugs for Bayesian analysis
• Time series forecasting
• R package construction.

Melbourne Monday 30 June - Thursday 3 July 2008