statistical society of australia incorporated newsletter
February 2005 Number 110
Registered by Australia Post Publication No. NBH3503 issn 0314-6820 Symposium and dinner in honour of Joe Gani’s 80th birthday
On Tuesday and Wednesday 14-15 December 2004 the Centre for Mathema- tics and its Applications (CMA) of the Australian National University (ANU) held a National Research Symposium entitled “A Celebration of Modelling and Applied Probability”. The venue was the Shine Dome of the Australian Academy of Science (AAS) in Gordon Street, Acton, Canberra. The Symposium honoured Professor Joe Gani, currently a member of the CMA, who turned 80 on Wednesday 15 December. The event was sponsored by the CMA, SSAI and ANZIAM (Australian and New Zealand Industrial and Applied Mathematics), a Valerie Isham, Joe Gani and Zari Rachev. Division of the Australian Mathematical Society (AMS). 1964 he launched the Journal of Applied In 1974 Joe returned to Australia, Professor Chris Heyde (CMA) opened Probability – the first of four journals as Chief of the CSIRO Division of the symposium with a brief biography currently published by the Applied Mathematics and Statistics (originally of Joe, who has been a close colleague of Probability Trust. Joe found the statistical the Division of Mathematical Statistics), his for over 40 years. During that time scene in Australia in the 1960s somewhat with headquarters in Canberra. There he became instrumental in broadening Joe had contributed immensely to many restrictive; and moved first to the USA organisations – such as ANU, SSAI (of the spectrum of activities of the Division. at Michigan State University, and then which he has been an Honorary Life For example, in collaboration with the in 1965 to Britain at Sheffield University, Member since 1983), CSIRO, AAS (of Australian Universities, he instituted a which he has been a Fellow since 1976) where he founded the Department of regular international program of visiting and AMS (of which he had been President Probability and Statistics. In 1967, he mathematicians and statisticians from during 1978-80). organized the joint Manchester-Sheffield overseas. However, following a Divisional Joe was born in Cairo, Egypt, in School of Probability and Statistics, which Review which proved to be unfavourable 1924. After completing a degree in continues to exist to this day. to some of these initiatives, he resigned Mathematics at Imperial College, London, he emigrated to Australia in 1948. After two and a half years as a lecturer at the University of Melbourne, and a variety of labouring and clerical jobs, he was awarded a Commomwealth ISI 2005 is fast approaching! Postgraduate Scholarship at ANU in November 1953. There he began to study Have you registered? for a PhD under Professor Pat Moran. Don’t miss this great opportunity to attend an ISI Session in Sydney and take Having completed his doctorate at the advantage of this cost-effective and enjoyable way to stay in touch with the latest end of 1955, Joe spent a year as Nuffield statistical developments. Discounted rates are available for SSAI members! Fellow at the University of Manchester in Details on the scientific program, satellite meetings and social program are given 1956-7 with Professor Maurice Bartlett, on pages 10 and 11 in the Newsletter. Further details can be found at the website and several years at the University of www.tourhosts.com.au/isi2005 Western Australia under Professor Larry 2005 Session of the International Statistical Institute (ISI) Blakers, before returning to ANU as Sydney, Australia – 5-12 April 2005 a Senior Fellow at the end of 1960. In . . . in honour of Joe Gani’s 80th birthday from his position and returned to the London (“Macroparasite population The symposium dinner was held on USA. There he spent four and a half models: Persistence, population scale and Tuesday evening 14 December in the years at the University of Kentucky, cross-species interactions”); Niels Becker Great Hall of University House at ANU. followed by a decade at the University of ANU (“Control of transmission with Following an introduction by Professor of California, Santa Barbara, where he two types of infection”); Kostya Borovkov Alan Welsh (CMA), talks in honour of Joe set up and developed the Department of of Melbourne University (“On the asymp- were given by Professor Neil Trudinger Statistics and Applied Probability. Joe’s totic behaviour of a simple growing (CMA) and Dr Neville Bartlett (President wife Ruth became seriously ill with breast point process model”); Jeff Hunter of of SSAI). Neil spoke of Joe’s great cancer in 1992 (she died of it in 1997), and Massey University (“Mixing times and generosity and sense of humour, and he retired in 1994, returning to Canberra Neville noted that Joe had accomplished their application to perturbed Markov where he has continued to publish and all three of the things which, according to chains”); John Blake of Birmingham remain active at many levels. a Jewish proverb, define a successful life: University (“Vigorous non-linear, non- He had fathered a child, written a book Following Chris Heyde’s introduction, spherical bubble dynamics: Applications and planted a tree. Finally, Joe responded Professor Svetlozar (Zari) Rachev of the to biology, medicine, chemistry, physics University of Karlsruhe gave the first by saying how absolutely overwhelmed and engineering”); Linda Stals of ANU talk of the Symposium, entitled “A he was by the warmth of the assembly. (“A plantation-nursery system”); Glenn generalized heteroscedastic asset price He also spoke of the writer Isaac Bashevis Fulford of QUT (“Spatial modelling of process: Properties, parameter estimation Singer – one of a rare few who wrote in infectious diseases”); Frank De Hoog of and pricing”. The other talks were by: Yiddish – and related three jokes. One CSIRO (“Predicting winding stresses for Dr Mark Westcott of the CSIRO (“How of these was about a young Jewish man many buses?”); Dr Tony Pettitt of the wound coils with large deformations”); who was convinced by his father to see a Queensland University of Technology Belinda Barnes of ANU (“An ecological psychiatrist regarding a possible neurotic (QUT) (“Statistical modelling for framework linking scales based on condition which might be preventing him nosocomial infections: Estimating trans- self-thinning”); and Geoff Aldis of the from marrying. Later the son told his mission rates for infection control and Australian Defence Force Academy (“An parents that he had been diagnosed with developing surveillance schemes”); Yoshi integral equation model of the control of an Oedipus Complex. To this his mother Ito of Aichi-Gakuin University (“Further a smallpox outbreak”). The Symposium replied: “Oedipus Shmedipus! What geometric methods for the distribution was followed on Wednesday afternoon does it matter? Everything is all right so of the sample correlation coefficient”); by the ANZIAM Canberra Branch Annual long as you love your mother.” Valerie Isham of University College, General Meeting. Borek Puza Report on BioInfoSummer 2004
Following the success of last year’s to do the Educational Computer Lab and How to Read It: A Geneticist’s View”. BioInfoSummer event, the Australian session while others attended Contributed NIB prizes were awarded to Ian Wood of National University’s Centre for talks. The Graduate Course Award was QUT (best poster), Brett Easton of ANU Bioinformation Science hosted yet another very popular, with nineteen students (best student talk) and Alicia Oshlack of extremely successful International Centre enrolled and sixteen on the waiting list. WEHI (best talk by a researcher within of Excellence for Education in Mathema- There are plans to expand the class size ten years of receipt of PhD). tics (ICE-EM) Summer Symposium in during next year’s BioInfoSummer to Next year’s BioInfoSummer will Bioinformatics, with the theme “Genome cope with the demand. focus on the theme of “Open Problems to Phenome Modelling”, 6-10 December The list of speakers this year expanded in Bioinformatics”. It is proposed that 2004. Sponsors also included Cray, Ceanet, to include international as well as it will be held during the week of 28 the Australian Partnership for Advanced local experts. The keynote and invited November – 2 December 2005. Details Computing and ANU’s National Institutes speakers were Warren Kaplan (Garvan will be confirmed by March 2005, and for Bioscience (NIB), Health and Human Institute, Sydney), Gavin Huttley (ANU), posted on the events page of ANU’s Science (NIHHS) and Engineering and Matthew Wakefield (ANU), Bill Pearson Mathematical Sciences Institute: http:// Information Sciences (NIEIS). There were (University of Virginia, USA), Michael wwwmaths.anu.edu.au/events/ approximately 150 registrants. Wise (University of Western Australia, The days were organised into Themes, Perth), Ziheng Yang (University College allowing attendees who were unable to London, UK), Allen Rodrigo (University attend the whole week to select those of Auckland, NZ), Geoff McLachlan areas of particular interest. The Themes (University of Queensland, Brisbane), were: Introduction to Molecular Biology; Conrad Burden (ANU), Hilary Booth Sequence to Structure; Comparative (ANU), Ian Dodd and Keith Shearwin Genomics; Analysis of Gene Expressions (University of Adelaide) and Terry and Regulatory Networks. Each day Speed (Walter and Eliza Hall Institute, started with Educational Lectures Melbourne). A poster session was held followed by a mix of Educational, Keynote on Tuesday evening, and student scholar- and Contributed talks. Most days held ships were awarded to 36 successful parallel sessions in the afternoon, allowing applicants. The ceremony was followed Prizewinners: (l to r) Alicia Oshlack, those students who had registered for the by a well-attended public lecture by Rob Conrad Burden (CBiS who presented Graduate Course Award in Bioinformatics Saint (ANU) on “Biological Information the prizes), Ian Wood and Brett Easton.
2 SSAI’s Canberra office has moved – The new postal address is: PO Box 5111, Braddon ACT 2612 SSAI Newsletter – February 2005 The telephone remains the same (02) 6249 8266 – Our new Fax (02) 6249 6558 Australian Bureau of Statistics Centenary Still counting after all this time – but not by hand
By Emma Macdonald In 100 years, the Australian Bureau of Statistics has come a long way – and for that chief statistician Dennis Trewin is eternally grateful. Gone are the days when the first census figures of 1911 were tabulated by hand – all four million cards of them – and took up to six years to finalise. At the 2001 census, it took just 10 months to process 8.6 million forms and return the first results to the community. This speedy compilation was aided by 920 computers processing 59.3 million sheets of paper, 15 scanners scanning in as much as one million pages a day, and 1.5 km of optic fibre cabling with 52.5 km of communications cable linking the technology together. While the technology driving the modern ABS had changed immensely, some things remained the same as one century ago when the ABS was founded, according to Mr Trewin. “Our purpose of providing information to support a democracy remains constant,” he said. “As does the need for integrity and professionalism of the organisation, and the protection of confidentiality. Don’t think statistics are boring, Mr Trewin said. Think of them as a mirror to who and what we are and how Australia is functioning as a society and economy. Not only do the financial markets depend on the ABS, so do governments. ABS data on the ageing population, falling fertility rates and a potential labour market shortage has infiltrated our collective conscience and made the Howard Government implement various policies in order to counter the trend. “The debate surrounding those trends NEW,OLD: Australian statistician Dennis Trewin with a comptometer, used to has actually resulted in a change of count statistics up until the 1960s, and a new slimline laptop used by statisticians attitude and now we are seeing fertility today. Picture: Vikky Wilkes. creep back up again. should be doing. school programs, so maths teachers One of the ABS productions Mr Trewin can incorporate more statistics into is most proud of is the groundbreaking The subsequent series of measures has their curriculum. Measuring Australia’s Progress series, been picked up in the United States, and which began in 2002. copied by Ireland. Apart from statistical literacy, the aim is that before Australia’s next census He came up with the idea of using It also won Mr Trewin and his ABS takes place in 2006, students across the a series of social, economic and team the “best contribution to society” country can conduct their own statistical environmental indicators to measure award by the Bulletin magazine. research as part of the “Census At School” progress and quality of life, rather than As the ABS’s 3000 staff – 1732 of whom simply using Gross Domestic Product, or, program. are based in Canberra – celebrate its 100th as other suggested, alternative measures This article is reproduced with birthday, Mr Trewin would like to make such as the Genuine Progress Indicator permission of The Canberra Times following statistics a greater part of all of our lives. developed by American academics. its publication on 22 January 2005. “But one number and a whole lot of “I think statistical literacy is a skill all * An article from the ABS about the value judgements didn’t seem to me Australians need to have.” Census At School program will appear in to be what a national statistical agency The ABS is currently working on the May issue of SSAI Newsletter.
SSAI Newsletter – February 2005 SSAI’s Canberra office has moved – The new postal address is: PO Box 5111, Braddon ACT 2612 3 The telephone remains the same (02) 6249 8266 – Our new Fax (02) 6249 6558 Editorial
The first newsletter of 2005 sees the to print reports of Branch activities, finish of some segments, the start of a whether regular meetings, workshops new one and the continuation of many on special topics or associated with others. The website of the month will particular individuals. We also accept PO Box 5111, not be continuing, neither will the contributions from people who have Braddon ACT 2612 Editors’ competition. In place of the attended statistical events outside Phone (02) 6249 8266 latter there will be a series of posers Australia. Keep those photographs Fax (02) 6249 6558 from Canberra Branch member, Borek coming too, so that there will be a Email: [email protected] Puza. There are prizes attached to permanent record of the wide range of Society Web Page these, and the Editors encourage you activities supported and attended by Society members around the country, http://www.statsoc.org.au to put pen to paper and tackle the problems. and indeed around the world.
As always, the Editors are delighted Best wishes for the new year! Editors
Alice Richardson, School of Mathematics Member News and Statistics, University of Canberra, PO Box 1, Belconnen ACT 2616 At the August 2004 Joint Statistical modelling, as applied to the design Email: [email protected] Meeting of the American Statistical and analysis of sample surveys; for Association, fifty-two new Fellows important contributions to official Michael Adena, Covance Pty Ltd were announced. They included two statistics; and for editorial service. PO Box 5125, Braddon, ACT 2612 members of the Statistical Society Email: [email protected] William T.M. Dunsmuir, Professor Phone: (02) 6274 9704 of Australia, and their citations are reproduced below. Congratulations of Statistics, University of New South Correspondence to Ray and William! Wales. For outstanding contributions in time series, space-time modelling Please direct all editorial correspondence Raymond L. Chambers, Professor and methods for count data; for to Alice Richardson. of Social Statistics, Southampton leadership in consulting in climate Disclaimer Statistical Sciences Research Institute. modelling and quality improvement; The views of contributors to this For outstanding contributions to and for administration and statistical Newsletter should not be attributed to the theory and practice of statistical activities. the Statistical Society of Australia, Inc. Subscriptions APOLOGY The Newsletter of the Statistical Society of Australia is supplied free to all members of the society. Any others The Newsletter Editors apologise to the Australian Bureau of Statistics wishing to subscribe to the Newsletter for incorrectly associating the ABS with the ‘causes of death’ data in the may do so at an annual cost of A$25.00 November competition. This set of figures is a hoax that regularly does the for an issue of four numbers. rounds at Christmas time, and bears no relation to information collected and published by the ABS. The ABS has asked that all figures attributed Advertising to their organisation that are published by SSAI be checked with the Advertising will be carried in the ABS first, which can be done through their Director of Media and Public Newsletter on any matters which the Relations. The ABS enjoys a strong and productive working relationship Editors feel are of interest to the members with the SSAI and the Editors are keen to maintain this relationship by of the Society. For details of advertising rates, etc. contact the Editors at the above exercising due care with any ABS material that appears in the Newsletter addresses. in the future.
Printer National Capital Printing In this issue 22 Pirie Street, Fyshwick ACT 2609
Editorial ...... 4 Three Doors ...... 12 DEADLINE FOR President’s Corner ...... 5 Conferences ...... 15 NEXT ISSUE: Workshops and FASTS ...... 6 AusCan Scholar Programs ...... 16 10 APRIL 2005 Profile of Adrian Baddeley ...... 7 Branch Reports ...... 17 ISI 2005 ...... 10
4 SSAI’s Canberra office has moved – The new postal address is: PO Box 5111, Braddon ACT 2612 SSAI Newsletter – February 2005 The telephone remains the same (02) 6249 8266 – Our new Fax (02) 6249 6558 President’s Corner
Joe Gani’s Eightieth Birthday integration. This will no doubt Joe Gani celebrated his eightieth birthday on the 15th December facilitate last minute changes 2004 and a two day workshop was held in his honour at the and announcements that will Australian Academy of Science in Canberra. The event was inevitably arise. Email addresses arranged by the Centre for Mathematics and its Applications at have changed along with the ANU and congratulations must go to Chris Wetherell and the move so change your address team that carried out the organisation. An impressive collection book entries now. of people gave talks along with comments about the influence Workshops that Joe Gani has had on their work. Besides having an It is great to see so many impressive academic career and being the driving force behind workshops/courses happening the establishment of two respected journals, Joe has influenced around the country and the many statisticians/mathematicians around the world by his registrations are running at very encouraging levels. Our Executive vision and ability to provide constructive suggestions. During his Officer is providing assistance by looking after registrations, fee time as chief of the CSIRO Division of Mathematics and Statistics collection and the issuing of receipts. This service is not free of in the seventies, Joe implemented programs and changes that charge but it frees organizers from having to sort out GST matters brought the best of the profession to Australia either as visitors or and so on. Some have complained that we are charging too much to longer-term appointments. This had the effect of making us all for this service but it is substantially less than that charged by much more aware of being part of a global community with high some universities who are becoming more commercially oriented standards and broadened our horizons. Joe is very much a legend when providing such facilities. in our community. ISI 2005 AusCan Scholar One of the ways that SSAI is assisting the ISI 2005 conference We are seeking applications from Australian researchers is by providing people to review and classify abstracts. A group interested in spending some time in Canada during 2005. Details of people have agreed to assist with this and their help is greatly of the programme are available from the SSAI website and if appreciated. SSAI members will be able to take advantage of anyone knows of a person who would be interested then please substantial discounts when registering so check out the conference encourage them to apply. website for details (www.tourhosts.com.au/isi2005/). ASC/NZSA 2006 Direct Election of Executive Officers Arrangements are well in hand to stage the 2006 conference in At the SSAI AGM in July, Ken Russell moved that we consider conjunction with the NZSA in Auckland. Sky City is an excellent the direct election of Executive officers by all SSAI members venue in the heart of Auckland within easy walking distance of a instead of the current approach where a nominating committee, wide range of accommodation and shopping. We have been most made up of the Executive plus Branch presidents, puts forward impressed with the facilities and the surrounding environment. nominations to Central Council. Doug Shaw has put together a David Scott heads up the organising committee and William paper on the subject and this has been circulated to the Executive Dunsmuir chairs the program committee. Feel free to approach and then will go to the next Central Council meeting early in either of these two people if you have any suggestions. February. I expect that this document will soon be published so that all members can consider the issue and express their views. Neville Bartlett It is interesting to note that we are very much out-of-step with Email: [email protected] comparable organisations.
SSAI Office and Website Looking for a job? For some years, SSAI has rented some office space at Covance For a listing of current statistical vacancies in Australia (formerly INSTAT) in Canberra in a rather cramped set of offices and New Zealand visit: at the Ainslie shopping centre. Covance has moved to a more spacious and modern site much closer to the centre of Canberra http://www.statsci.org/jobs and SSAI has moved with them. Our telephone number is the 10 positions already listed in 2005! (as at 19/01/05) same but the postal address has changed. Do you have a position to Coincidentally, SSAI has renovated its website and moved to a commercial server. We have been using university web sites advertise on the website? for quite some time and are very grateful for the assistance from ANU. Special thanks go to Michael Martin for all of his assistance Email a job description to [email protected]. over the last few years. It is time for us to stand on our own feet Listing is free! and not rely on the generous hospitality of others. One major This service is proudly brought to you by the advantage of the new configuration is that it provides us with greater flexibility of operation in that branches and sections can now directly maintain their own section of the SSAI site quite independently of everyone else while maintaining overall
SSAI Newsletter – February 2005 SSAI’s Canberra office has moved – The new postal address is: PO Box 5111, Braddon ACT 2612 5 The telephone remains the same (02) 6249 8266 – Our new Fax (02) 6249 6558 Professional Opportunities
Post-ISI Workshops The AGM was a rare opportunity for Brought to you by the societies themselves to influence FASTS NSW Branch A post-ISI workshop in Canberra direction. In reports to Council it is clear organised by the society with the theme that FASTS is facing both a goverance of methods in official statistics is planned. Meta-analysis impass and a structural deficit in its The workshop coincides with visits Workshop to Canberra of overseas ISI delegates funding, as its member base is declining. associated with the recently completed It has achieved visibility in national The NSW Branch is proud to present politics, and appears to have handled a DACSEIS project, funded by the European a course on the use of meta-analysis succession at both executive officer and Union. The workshop is free to members during the drug development program. chair levels well. The incoming chair or ISI delegates but attendees are asked Professor Stephen Senn, one of the most to register with the society ahead of time was a contested position, allowing for well known statisticians world-wide to assist organisers. The workshop will some articulation by candidates as to be held at ABS House in Belconnen, why they were seeking this office. In the contributing to the drug development on Thursday 14th April. Program and event the meeting elected Tom Spurling industry and author of the popular books registration details will be available on to succeed Snow Barlow. Council debate Cross-over Trials in Clinical Research, the SSAI website, and announced during centred on essential issues: what is unit Statistical Issues in Drug Development ISI. Contacts for the workshop are Paul of the organisation, how to respond to and Dicing with Death will be visiting Sutcliffe [[email protected]] and body of unrepresented scientists, what Australia in late March, early April as part Stephen Horn [[email protected]. do societies want from an umbrella body of the ISI conference. During his visit he au]. like FASTS. Until now it has influenced has agreed to present a 2-day workshop the government policy agenda without The Department of Family and on meta-analysis on March 30-31. Community Services is hosting a follow- committing itself to ‘public good’ on seminar with the same speakers on lobbying – maybe it should? If it is to be Meta-analysis has now achieved the 15th April at Tuggeranong Office a successful lobby group it will need the a century of use and controversy. In Park, Greenway that will address policy power and licence to respond quickly. this workshop Stephen will cover the implications of panel survey data quality. This might be difficult to reconcile with practicalities of carrying out meta- This seminar will be open to members and ‘member responsiveness’. An overhaul analysis concentrating not only on delegates, but they are asked to register of its constitution was set in motion in different models for analysis but also interest ahead of time with Helen Boden light of inconsistencies in its structure on contentious matters such as when [[email protected]]. Details will and governance. Those curious as to what and when not to pool. Some parallels be made available on the SSAI website. this organisation might be doing on our to the analysis of multi-centre trials behalf should consult the website: www. Liaison with other FASTS.org. will be made and attention will also be drawn to some differences in practice Stephen Horn organisations – FASTS between ‘public’ organisations such as the The Federation of Australian Scientific SAS Applications Cochrane Collaboration and commercial and Technological Societies (FASTS) to organisations such as pharmaceutical which the Society is affiliated through Workshop sponsors. Common pitfalls will be its membership of the Australian Forthcoming Professional Development covered. Case studies will include the Mathematics Council held its annual opportunities brought to you by the NSW controversy over breast cancer screening general meeting 24 November in Canberra. Branch of SSAI. FASTS acts as an umbrella group for its and that of rofecoxib and heart disease. member societies, and as such reflects in A series of one-day workshops designed Rofecoxib was the drug for arthritis and amplified fashion the same pressures the to provide members with skills in pain management that was recalled by statistics society has been responding to: applying SAS is being planned. These international drug company Merck on declining membership levels in traditional workshops will be presented by SAS, are September 2004 after a study showed an professional associations even as condensed versions of the corresponding increased risk of heart attack and stroke. professional activity burgeons; splintering SAS Institute courses and will be offered of disciplinary loyalties; remoteness of at very competitive prices. So, if you are interested in meta- analyses and would like to be entertained society activity from membership. FASTS • Mid to late February 2005 runs an office out of Canberra. Outside – SAS Programming Fundamentals with some discussions of the controversial of a political lobby its main activity is the – Introduction to Statistics using nature of these methods please join us annual ‘science meets parliament’ event. SAS at the workshop. Further details on the In 2005 it is planning a Governance of • March 2005 speaker, the venue and registration can be Science Conference whose intent is: “to found at http://www.maths.unsw.edu. [discuss] global changes in the funding – Introduction to Data Mining using au/~inge/ssai_nsw/ and structure of science and research . . . SAS and focus of role and prospects of science Contact: Caro Badcock – Caro Badcock societies . . .”. [email protected] [email protected]
6 SSAI’s Canberra office has moved – The new postal address is: PO Box 5111, Braddon ACT 2612 SSAI Newsletter – February 2005 The telephone remains the same (02) 6249 8266 – Our new Fax (02) 6249 6558 2004 Pitman Medal
Profile of Adrian Baddeley – 2004 Pitman Medal Winner
At last year’s Australian Statistics developed methods for object recognition Conference in Cairns, Professor in images (for instance, recognizing cells Adrian Baddeley was presented with in a microscope image) using stochastic the 2004 Pitman Medal in recognition models, and worked on summary of outstanding achievement in, and statistics for spatial point patterns. He contribution to, the discipline of Statistics. also applied survival analysis techniques Those of you seeking an official citation to spatial data, and did research on should look towards future issues of the stereological estimation theory. However, Australian and New Zealand Journal amidst Adrian’s research achievements of Statistics. If, on the other hand, you during this period and his enjoyment of would like something a little lighter with Amsterdam living there was a sad note a human interest slant, then read on. when he divorced his long term partner Adrian Baddeley was born and brought up Jane in 1991. in Melbourne, where he attended Eltham Adrian moved to Perth in 1994 to take High School. Adrian became interested up a lectureship at the strong Statistics up his current position as Professor of in statistics while at school, inspired by Department at the University of Bath. Statistics at the University of Western a combination of good teachers and M.J. He admits to fond memories of the West Australia. He was head of the Department Moroney’s book “Facts from Figures” Country, but whether these involve the of Mathematics and Statistics from 1998- which offers the following advice: “If you fiendishly potent scrumpy cider remains 2000, allegedly because the Dean informed are young, then I say: Learn something a mystery. Whatever the source of his him that it was, “a job that Baddeley about statistics as soon as you can.” inspiration, Adrian produced ground needs doing.” While at UWA Adrian has Suitably motivated, Adrian headed off breaking work in stereology leading to focused his research on practical methods to the Australian National University a paper on the estimation of surface area for spatial statistics. In collaboration with from which he graduated BA Hons in Rolf Turner he has written an R software from vertical sections which has been Statistics and Pure Mathematics in 1976. package, ‘spatstat’, for the analysis of cited more than 500 times. His lecturers during this degree included point pattern data, and has developed Eugene Seneta (for branching processes), Adrian returned to Australia in 1985 practical estimation algorithms. as a research scientist, and later senior Richard Tweedie (for Markov chains) and Despite Adrian’s ongoing record of research scientist, at CSIRO Division of Chris Heyde (for martingales). Adrian academic achievement, he finds time to Mathematics and Statistics in Sydney. was introduced to geometrical probability indulge in his favourite past-times of t’ai He worked on statistical image analysis, by Roger Miles at this time, a topic that chi and scuba diving. He is particularly spatial data analysis and stereology, and has proved dear to his heart. fanatical about the latter and spends most learned a huge amount about applied By the time Adrian left ANU the predictors weekends with his girlfriend Melissa on statistics and computing from his of a distinguished career in statistics were this or that diving trip. He has clocked up colleagues. Adrian developed techniques already firmly in place. He had won for measuring the distance between two more than 600 dives, and even underwent various awards as an undergraduate images (including the “Baddeley metric”) a voluntary heart operation in order to (University Medal, SSA prize, Hannah with application to image cleaning continue this activity. There are hundreds Neumann medal); he had published three and optimal image reconstruction. His of underwater photos on Adrian’s web scientific papers; and he had long hair and time at CSIRO came to an abrupt end pages; see http://www.maths.uwa.edu. a preference for hanging around lefties in 1987 when, following what Adrian au/~adrian/scubaphoto.html. Many (although these last two factors may not politely refers to as a turbulent period are stunning, but beware downloading be statistically significant). The next step of reconstruction at the organization, he without a broadband connection! was a move overseas, so in 1977Adrian resigned in front of 200 people and a It is perhaps fitting to finish with a story set off for Europe via Indonesia. His stay video camera. related by my colleague (and Adrian’s in Paris was eventful – one of the rafters sometime dive buddy), Berwin Turlach, in his hotel room carried the message, In 1988 Adrian moved back to Europe as a research group leader at the Centre for which combines Adrian’s twin passions “Congratulations! You have found the for spatial statistics and diving. Once, cheapest room in Paris! Get Out While Mathematics and Computer Science (CWI) in Amsterdam. He enjoyed Amsterdam at the end of a long dive, which left You Still Can”. Nonetheless, he survived Berwin completely lost and wondering to attend a meeting on geometrical hugely, and learned fluent Dutch and coffee appreciation. CWI itself provided a where they were, the dive boat appeared probability in Paris before embarking on miraculously above them. Back on the a cycling tour of France. stimulating environment in which Adrian pursued his work in statistical image boat, Berwin complimented Adrian on Adrian’s next move was across the English analysis and spatial statistics. In 1991 he his navigation skills, to which Adrian Channel. He spent 1977-1980 doing a PhD was appointed as adjoint professor of responded, “but don’t you know that in mathematical statistics at the University mathematics at the University of Leiden a random walk is recurrent?”. While of Cambridge, supervised by David (a common type of arrangement for Adrian’s underwater navigation might Kendall. He won the Adams prize for senior Dutch researchers). He supervised be somewhat haphazard, his rise to the this work and was elected as a Research PhD students Marie-Colette van Lieshout top of the statistical tree owes nothing to Fellow at Trinity College for 1979-1982. and Annoesjka Cabo, and worked in chance. At the end of this period Adrian took collaboration with Richard Gill. Adrian Martin Hazelton
SSAI Newsletter – February 2005 SSAI’s Canberra office has moved – The new postal address is: PO Box 5111, Braddon ACT 2612 7 The telephone remains the same (02) 6249 8266 – Our new Fax (02) 6249 6558 Sri Lankan Statistical Conference International Sri Lankan Statistical Conference and Tsunami Came Face to Face Kandy, Sri Lanka: December 28-30, 2004 http://www.st.rmit.edu.au/~desilva/conference/slstat.htm
The original planning for this conference his totally dedicated group began when the two Co-Chairs, Professors of local committees that Basil M. de Silva and Nitis Mukhopadhyay, included a large number of met in India in December 2002. The faculty members, students conference venue was the Postgraduate and staff at the PGIS. Institute of Science (PGIS) in the middle The conference began of the beautiful lush green campus of the as scheduled in the early University of Peradeniya. The facility was morning of December 28 wonderful and modern. by lighting the traditional The conference was sponsored by a oil lamp under Sri Lankan number of international organizations drumbeat and the rendition including the Department of Statistics of her national anthem in the at the University of Connecticut (Storrs, presence of more than 170 U.S.A.), the School of Mathematical and delegates during the opening Geospatial Sciences at RMIT University ceremony. (Melbourne, Australia), the Department The situation was a little of Statistics and Computer Science at the awkward because of the University of Peradeniya (Sri Lanka), the overpowering emotions from PGIS (Sri Lanka), the Indian Association freshly inflicted grief, but at for Productivity Quality and Reliability the same time, the human (Calcutta), and the Statistical Society of spirit inside the auditorium Canada. wanted to soar high too. The foreign and Sri Lankan delegates This was supposed to be Sri Lankan traditions evident at the opening started arriving on location around the moment of glory, the ceremony. December 25, 2004. Some arrived earlier culmination of two years’ of for vacationing in the coastal areas and extreme hard work on the part of so many (University of Kiel, Germany) gave islands. At the time, the members of colleagues. Everyone realized that there the second plenary presentation in the the international and local organizing was no turning back. After observing a interface of mathematics, statistics and committees were steadfast busy finalizing minute of solemn silence to commemorate finance, a fast growing field in its own some of the last minute details. No one the lives lost to the sea, all the delegates right. was going to leave anything to chance! quickly moved the program forward. The participants were represented The air became more festive by the The chief guest, Hon. Professor Tissa from countries including Australia, hour. The evening of December 25 was Vitharana, Minister of Science and Austria, Canada, Finland, Germany, gorgeous and the spirits were high as Technology of Sri Lanka, kindly declared India, Malaysia, New Zealand, Singapore, everyone exchanged greetings and hellos the conference open. He delivered the South Africa, Sri Lanka, Switzerland, with the arriving colleagues. After all, no keynote address challenging statisticians Syria, Thailand, United Kingdom and international statistical conference of this to help very poor countries like Sri Lanka the United States of America. Larger magnitude ever took place on Sri Lankan to advance quickly in science, education, contingents came from Australia, Canada, soil as far as anyone could remember. health and information technology. His India, New Zealand, United Kingdom Unfortunately, the unimaginable presence was a miracle of sorts and a and the United States of America. The tsunami hit the Sri Lankan coast and other morale booster especially in view of his largest contingent consisted of colleagues countries in the region in the wee hours of assignment (delegated by the President) to of Sri Lankan origin, residing in Sri Lanka December 26. It was impossible to fathom head Sri Lanka’s urgent health and human and elsewhere. the destruction and the toll on human services as well as relief efforts in the face There were thirty invited paper sessions lives at first. All over the world, people of this world’s biggest natural catastrophe. on varied topics such as adaptive and felt numbed, and the Sri Lankans were The Chancellor of the University of algorithmic approaches, applied time Peradeniya and other special guests no exception as everyone was quickly series, Bayesian methodology, biological also welcomed the gathering during the engulfed by the horrific sense of disbelief applications, biostatistics, clinical trials, inauguration. and helplessness. computational statistics, corporate The conference organizers, however, Professor Kanti V. Mardia (University decision making, data mining, data exhibited unbelievable courage, of Leeds, U.K.) gave the first plenary warehouse and analysis, estimation, determination and defiance against nature presentation “Past Revolutions and financial econometrics, hypothesis testing, to go ahead with the international forum Future Prospects in Science and Statistics” likelihood based inference, modeling on- as planned. Bravo to all the organizers, which set the tone for this conference’s line auction data, operations statistics, especially to Professor Lakshman theme, namely, ‘visions of futuristic option pricing models, regression Dissanayake, Director of the PGIS, and methodologies’. Professor Albrecht Irle techniques, reliability analysis, sampling
8 SSAI’s Canberra office has moved – The new postal address is: PO Box 5111, Braddon ACT 2612 SSAI Newsletter – February 2005 The telephone remains the same (02) 6249 8266 – Our new Fax (02) 6249 6558 Sri Lankan Statistical Conference
designs, sequential methodology, statistical Methodologies. Basil M. de Silva and Nitis long time. Every participant will surely methods, statistics in health science and Mukhopadhyay, eds. (December 2004), pp. remember how courageous Sri Lankan environment, and statistics in sport. 588. PGIS, University of Peradeniya, Sri colleagues and others have been against There were few cancellations or delayed Lanka and RMIT University, Melbourne, all odds when the chips were down, arrivals of participants. Everyone around Australia (ISBN 0 86459 339 2). especially in the face of unimaginable understood the situation and painfully One of the goals of this international high stakes and casualties of tsunami of sympathized. Those who could not attend conference was to create a bridge of December 26, 2004. the conference were sorely missed by the communication to share the cutting edge The ultimate hope is that the friendship participants. ideas and methodologies with the younger and collegiality that were nurtured among generation in Sri Lanka. The occasion gave A high point included the compilation scientists from Sri Lanka and other of the refereed conference proceedings everyone a wonderful opportunity to meet countries during December 28-30, 2004 before the conference started and a printed and interact with many local graduate will continue to reign. This bond would copy was given to each participant at the students and younger colleagues. Many then grow into better and more elaborate conference. It is a delight to report that the students enthusiastically asked if this refereed proceedings include both plenary kind of international conference could be scientific exchanges in the future. In a papers and forty two selected invited held again next year! They were definitely global environment of our existence, no papers filling nearly six hundred pages. energized and so were the foreign one can really afford to leave anyone Exact details of this copyrighted volume delegates. This exercise was surely a win- behind, because everyone’s legacy surely are furnished below: win proposition for everyone involved. affects everyone else’s! Proceedings of the International Sri Lankan Sri Lankan hospitality will remain in Nitis Mukhopadhyay Statistical Conference: Visions of Futuristic the memories of all participants for a very University of Connecticut, Storrs
ISI2005 will have an interactive panel session for Young Statisticians on Friday 8 April. For more information check out the conference program or email Anna Munday on [email protected] Statistical Society of Australia Inc. Attention: Young Statisticians and Students
Are you new to the profession or a statistical student and a member of SSAI wanting to attend ISI 55 in Sydney in April 2005? The 2005 Session of the International Statistical Institute (ISI) will be held in Sydney from 5-12 April 2005. SSAI has allocated money to contribute to young statisticians and/or students who need some financial assistance to attend the conference. Conference registration for supported SSAI members has been sponsored by the Australian Bureau of Statistics. If you would like to be considered in obtaining some financial assistance to attend the conference please forward 1-2 page document addressing the section criteria.
Selection Criteria Please send your application, addressing the Selection Criteria together with your CV and transcripts of your • What would be the value to your both short term and academic results to: your developing career to attend the conference? • Which aspect of the conference is most interesting to you? Ms Jane Waslin Why? Executive Officer • What is the level of assistance required? SSAI • Provide two Professional Referees PO Box 5111 You will also be required to provide a Report on return for Braddon ACT 2612 the SSAI which may be reproduced in the SSAI Newsletter and/or posted to the SSAI website. Applications close Monday 7 March 2005
Applications Close Monday 7 March 2005
Should you have any enquiries please contact Jane Waslin Ph: (02) 6249 8266 or email: [email protected]
SSAI Newsletter – February 2005 SSAI’s Canberra office has moved – The new postal address is: PO Box 5111, Braddon ACT 2612 9 The telephone remains the same (02) 6249 8266 – Our new Fax (02) 6249 6558 55th ISI Session – Sydney, Australia
Register now! to predict other variables. His discovery is widely used and is commonly known Discount Registration available as “Granger causality”. He is now for SSAI Members Professor Emeritus at University of This year the 2005 Session of the California, San Diego (UCSD). International Statistical Institute (ISI) to review the papers for the Session to Glenn Stevens has been the Deputy will be held in Sydney, Australia, from ensure the Session is a success. Governor of the Reserve Bank of 5 to 12 April. The Scientific Program will be Australia (RBA) since December 2001 In recognition of the contribution the supplemented with tutorials and short and has spent most of his professional SSAI is making to the 2005 ISI Session, courses. Special theme days will cater career in RBA, joining the bank’s the ISI National Organising Committee for those with interests in finance and Research Department in 1980. He is pleased to announce that members of statistics, environmental statistics and holds degrees in Economics from the the Statistical Society of Australia (SSAI) genomics. University of Sydney and the University of Western Ontario, Canada. can register at the ISI member rate of Details on the Scientific Program, $730. This is a great opportunity for including the full list of Invited and In 1990, he was a Visiting Scholar SSAI Members to attend an ISI Session Contributed Paper Meetings, are at the Federal Reserve Bank of San on home soil and take advantage of listed on the Session website at www. Francisco. He was Head of the this cost-effective and enjoyable way to tourhosts.com.au/isi2005. Economic Analysis Department of the stay in touch with the latest statistical Reserve Bank of Australia from August developments. Keynote speakers confirmed 1992 to September 1995 and head of its International Department from If you are intending to participate, Renowned mathematical biologist September 1995 to December 1996. it is highly recommended that you Lord Robert May, econometrician register now. To register please complete Professor Clive Granger and Deputy Since December 1996 he has been the online registration form at www. Governor of the Reserve Bank of Assistant Governor (Economic), tourhosts.com.au/isi2005 or return the Australia Glenn Stevens have been responsible for overseeing the economic Registration Form in Bulletin II to the confirmed as key speakers at the 2005 analysis and research of the Bank’s staff Conference Managers. Even if you are Session of the International Statistical and formulating policy advice for the still undecided as to whether to attend Institute (ISI) scheduled for Sydney next Governor and the Board of the Bank. or not you should register your interest April 5-12. Glenn is currently a Member of Advisory to ensure you are kept informed about Boards for the Melbourne Institute Lord May, an Australian by birth, the Session developments. of Economic and Social research and obtained his doctorate in theoretical the Hong Kong Institute for Monetary If you have already registered for physics from the University of Sydney Research.. the Session please continue to check in 1959 at the age of 23. He is a world the Session website www.tourhosts. authority on mathematical biology. Satellite Meetings com.au/isi2005 on a regular basis for In 2000, he was appointed for five program updates and conference news. A number of satellite meetings will be years as President of the Royal Society of held before or after the 2005 ISI Session. London, a position with a rich tradition Key Dates Details and links for each meeting are and one of the most esteemed in the listed below: 4 March 2005 – Speakers to email world of science. That followed a five- their presentation to the Conference year appointment as the Chief Scientific Managers Adviser to the British Government 31 March – 2 April 2005 4 April 2005 – Registration for the and Head of the Office of Science and Issues for Official Statistics for Small Session commences Technology, playing an influential role Countries (especially island nations) 5 April 2005 – Session opens in national scientific affairs. Noumea, New Caledonia (www.stat. 12 April 2005 – Session closes Lord May holds a Royal Society fi/iaos/future_activities.html) Professorship jointly in the Department of Zoology, Oxford University, and Scientific Program 29 March – 1 April 2005 at Imperial College, London, and is a The Scientific Program for the ISI Fellow of Merton College, Oxford. 14th International Workshop on Session will feature leading keynote Matrices and Statistics Auckland, Clive Granger shared the 2003 Nobel speakers from around the world and New Zealand (http://iwms2005. Prize in Economic Sciences with Robert more than 100 scientific Sessions. massey.ac.nz) Engle for their discoveries in the Geoff Lee, Local Program Committee analysis of time series data. The work Chair and Head of ABS Methodology has fundamentally changed the way 4 – 5 April 2005 Division said ‘’… the Invited Program that economists think about financial Statistics Education and the is shaping up really well”. “The and macro-economic data and has led opportunity to attend a Session of the Communication of Statistics to significant breakthroughs in Statistics Sydney, Australia (http://www.stat. ISI in Australia is a once in a lifetime and Macro-economic forecasting. opportunity”. auckland.ac.nz/~iase/conferences. Professor Granger is also noted for php?show=iase2005) Members of the SSAI will be assisting developing a formal statistical notion of the ISI 2005 Local Program Committee causality based on which variables help
10 SSAI’s Canberra office has moved – The new postal address is: PO Box 5111, Braddon ACT 2612 SSAI Newsletter – February 2005 The telephone remains the same (02) 6249 8266 – Our new Fax (02) 6249 6558 General Schedule
MORNING EARLY AFTERNOON LATE AFTERNOON EVENING Monday 4 April Short Courses Registration Registration Short Courses Short Courses Tuesday 5 April Registration Registration Opening Ceremony Welcome Reception Short Courses Short Courses Short Courses Wednesday 6 April Scientific Meetings Scientific Meetings Scientific Meetings Optional Social Event Thursday 7 April Scientific Meetings Scientific Meetings Presidents IPM Meeting Australiana Night Friday 8 April Scientific Meetings Scientific Meetings Scientific Meetings Optional Social Event Saturday 9 April Scientific Meetings Excursions Excursions Optional Social Event Sunday 10 April Excursions Excursions Excursions Optional Social Event Monday 11 April Scientific Meetings Scientific Meetings ISI General Assembly Farewell Party (optional) Tuesday 12 April Scientific Meetings Scientific Meetings Scientific Meetings
13 – 16 April 2005 Fourth International Symposium on Business and Industry Finance, near Cairns, Queensland, Australia (www.action-m.com/isbis4)
14 – 15 April 2005 Measuring Small Populations Wellington, New Zealand, (www. stats.govt.nz/ISIsatellitemeeting)
Social Program The Social Program will be a highlight of the Session and has been designed to provide participants with an opportunity to relax, experience Sydney and maximise networking opportunities. The following events are included in the registration fee for delegates and accompanying persons: Tuesday 5 April 2005 Opening Ceremony “Centenary Celebration – from Dreamtime to the Future ISI 2005 Sydney, Australia 5-12 April 2005 – Come and see Sydney, an exciting Tuesday 5 April 2005 and cosmopolitan city located on one of the largest and most beautiful harbours in Welcome Reception “Faces of the world. Australia” Thursday 7 April 2005 * Optional events are not included in Have you seen Australiana Evening – “The Way of the registration fee. Life Down Under” Information See the website www.tourhosts.com. The following optional events* will be au/isi2005/social.asp or Bulletin II for Bulletin II? offered to delegates and accompanying more details. persons. Information Bulletin II provides Wednesday 6 April 2005 Contact Details the latest details on the arrangements for the 2005 ISI Session and the Discover the Historic Pubs of the ISI 2005 Conference Managers final registration form. A copy is Rocks GPO Box 128 available on the ISI Website www. Friday 8 or Saturday 9 April 2005 SYDNEY NSW 2001 tourhosts.com.au/isi2005. If you Telephone: +61 02 9248 0800 A night at the Sydney Opera House would prefer a hard copy email Fax: +61 2 9248 0800 Sunday 10 April 2005 [email protected] and a Email: [email protected] copy will be sent to you. Australian Wildlife by Night Website: www.tourhosts.com.au/ Monday 11 April 2005 isi2005 Farewell Party “A Floating Affair” We look forward to seeing you in Sydney!
SSAI Newsletter – February 2005 SSAI’s Canberra office has moved – The new postal address is: PO Box 5111, Braddon ACT 2612 11 The telephone remains the same (02) 6249 8266 – Our new Fax (02) 6249 6558 Three Doors with Borek Puza (Edition 1)
Welcome to a new feature of the SSAI Newsletter, (a) the car was hidden randomly; (b) you were definitely going titled Three Doors. From now on, each issue will contain to pick a door randomly; (c) the host was definitely going to this regular column whose purpose is the discussion of open a goat door (other than the door picked by you) and offer interesting questions, paradoxes, conundrums, etc. in the the option to switch; (d) the host is one of two (called M1 and field of probability and statistics. In each edition there will M2) who take turns at staging the show on alternate nights; (e) be a puzzle for readers to solve. All persons who submit whenever presented with a choice of two doors to open, M1 opens a correct solution will be eligible for a fabulous mystery the door with the lowest number, and M2 decides by flipping a prize donated by the Statistical Society. That prize, its coin; and (f) you randomly chose a night on which to play and winner (or winners), and of course the solution, will be have no extra information regarding the host’s identity. announced in the subsequent edition. For your chance to win a fabulous mystery prize, send your solution by email to
12 SSAI’s Canberra office has moved – The new postal address is: PO Box 5111, Braddon ACT 2612 SSAI Newsletter – February 2005 The telephone remains the same (02) 6249 8266 – Our new Fax (02) 6249 6558 Election of SSAI Office Holders
Election of SSAI Office Holders A Discussion Paper
This discussion paper was prepared in response to concerns b. The opportunity for members of the Society to nominate expressed by members at the 2004 SSAI Annual General Meeting candidates for election was not publicized in the most in Cairns. The Society would like to hear the opinions of members recent elections about the options outlined in the paper below. If you wish to express c. For the most recent elections, the Nominating Committee a view, please contact the Secretary of the Society, or the President did not meet as a Nominating Committee. The members or Secretary of your Branch. Email addresses for Society and Branch of the Nominating Committee were asked to put forward Secretaries are listed on page 15 of this Newsletter. nominations for Vice-President, three nominations were Doug Shaw made and all three went forward (one nominee withdrew Secretary, SSAI before the ballot was held). 1. Background 2. Practice in other Societies At the SSAI Annual General Meeting in July, 2004 several members expressed the view that SSAI elections would be The Rules of some other bodies were examined to obtain a more democratic if office bearers were elected by a vote view of common practice in this area. amongst all members, rather than by a vote amongst Central New Zealand Statistical Association Council members as at present. The meeting unanimously The President, Secretary, Treasurer and a committee are passed a motion requesting that Central Council consider elected, after being proposed and seconded, by members this issue. attending the Annual General Meeting. The current Rules of the Society, adopted at the 2004 International Biometric Society Annual General Meeting, make the following provisions for A Vice-President is elected every second year by a ballot the election of office bearers. of the Council; the person elected serves one year as Vice- (1) The Council shall elect the office-bearers of the Society President, two years as President and a fourth year as from nominations in accordance with Sub-Rules (2) and Vice-President. A nominating committee puts forward a (5) as follows. nomination or nominations for Vice-President, and a ballot (2) There shall be a Nominating Committee consisting of the is held even if there is only one nominee. The nominating President as convenor, the Vice-President, the Secretary, committee writes to all Regions seeking nominations; the Treasurer, the Australian Editor of The Australian and nominations must be supported by more than one Region. New Zealand Journal of Statistics, and the President for The nominating committee also puts forward nominations the time being of each Branch of the Society. for Secretary; id there is more than one nominee a ballot of (3) By 28th February in each year, the Nominating Committee the Council is held. shall submit to Council a list of not more than three names for each of those of the offices of Vice-President, the Australian Mathematical Society Secretary, the Treasurer that require election for the Elections are held for President-elect (in odd years), Vice- ensuing year. In the event that an election is necessary, a President and ordinary members of Council. The Council acts ballot shall be conducted before 31 March. Council shall as a nominating committee; in addition any two members appoint a member of the Society as returning officer; the can make a nomination. The ballot contains the names of the returning officer shall determine the manner of conducting Council nominees and the names of members’ nominees. the ballot. Voting is by postal ballot of members, although e-voting (4) Nominations of candidates for election as office-bearers is allowed. The Secretary and Treasurer are appointed by of the Society may be made by any member of the Council. Society or by any Branch Council of the Society to any American Statistical Association member of the Nominating Committee, by 31st January Postal ballot elections are held annually for President-elect in each year. and Vice-President. A Committee on Nominations is required (5) The Nominating Committee, before 31st October of the to submit at least two candidates for each of President- previous year, shall seek from each Branch nominations elect and Vice-President. In addition, nominations may be of candidates for election as office-bearers of the Society, made by members’ petition; a petition requires at least 100 for those offices that require election for the ensuing signatures. The Association employs an Executive Officer year. who is the Secretary; the Board of the Association appoints a Treasurer. The following observations can be made in relation to the application of these clauses, noting that the only change made to them in the most recent revision of the Rules was in relation to the returning officer for the ballot amongst Central Council members. a. Elections for Society office bearers, such as that for Vice- President in 2004, are rare continued on page 14
SSAI Newsletter – February 2005 SSAI’s Canberra office has moved – The new postal address is: PO Box 5111, Braddon ACT 2612 13 The telephone remains the same (02) 6249 8266 – Our new Fax (02) 6249 6558 Election of SSAI Office Holders Continued
Royal Statistical Society b. The process of nomination by members should be The Council consists of a President, an Honorary Treasurer, clarified to ensure that such nominations are in addition three Honorary Secretaries, twenty-six Ordinary Members, to the nominations made by the Nominating Committee. three Past Presidents, and co-opted members. Council Requests for such nominations should be advertised nominates a candidate for President; in addition, any six to all members, perhaps in the November issue of the members can nominate a Fellow for President. Council Newsletter. A decision must be made as to how many calls for suggestions for the remaining elected positions, members must support a valid nomination; two or five and forwards to members one nomination for Honorary seem to be the right sort of numbers. Treasurer, one nomination for each of the Honorary Secretary positions, and twice as many nominations as there are With this background, three options are considered. Note Ordinary Member positions to be filled. Any four members that, in all options, an election would only be required if there can nominate a Fellow for one of these positions. Voting is by is more than one nomination for a position. a postal ballot of all members. Institute of Mathematical Statistics Option 1 – The Status Quo Elections are held for a President-elect and fifteen council members. The President selects a Nominating Committee Under this option, elections for Vice-President, Secretary which produces at least one nomination for President-elect and Treasurer would continue to be held amongst the and at least two nominations for each council position. Other members of Central Council. nominations may be made by a petition of at least twenty Advantages: members. The election is by postal ballot of all members. Minimises changes to the Rules The council elects administrative and editorial officers by Cheaper than a ballot of all members majority vote. Disadvantages: International Statistical Institute Does not address member concerns about the current Elections are held for a President-elect and three Vice- ‘undemocratic’ process Presidents, as well as Council members. Council appoints Leaves SSAI visibly out of line with contemporary a Nominating Committee to propose officers and council practice members. Nominations can also be made by a petition of five or more individual members. Voting is by a secret mail Option 2 – Election amongst all members for Vice-President ballot. The roles of Secretary and Treasurer are taken on by only. the Institute’s Permanent Office. Under this option, the Vice-President would be elected by a ballot of all members, but the Secretary and Treasurer Setting aside the practices of the New Zealand Statistical would continue to be elected by a ballot of Central Council Association and the International Biometric Society, the clear members. commonalities amongst the other bodies are Advantages: a. A ballot of all members for Presidential positions Satisfies member requirements for a democratic process b. A Nominating Committee to nominate candidates for in respect of the most visible office elected positions May ensure more continuity in the ‘operational’ positions of Secretary and Treasurer c. Provision for some number of members to make The model is adopted by some other bodies nominations in addition to those made by the Nominating Committee Disadvantages: Members may still feel disenfranchised in respect of the positions of Secretary and Treasurer There does not appear to be unanimity about the election There would be a significant cost associated with the of Secretaries and Treasurers for those bodies where these election of a Vice-President positions are elected rather than being paid positions. Some include them in the general election process, others reserve their election for the body’s council. Option 3 – Election amongst all members for Vice-President, Secretary and Treasurer 3. Options for SSAI Under this option, all of the Vice-President, Secretary and Treasurer would be elected by a ballot of all members. Before considering options relating to elections, it would appear that Advantages: Satisfies member requirements for a democratic process a. There should be a Nominating Committee charged with This model is adopted by some bodies producing at least one nomination for each position to be filled by election. Mandating that the Nominating Disadvantages: Committee should produce at least two nominations, and Introduces a possibility of loss of continuity in the hence ensure an election, appears to be unnecessary – it ‘operational’ offices of Secretary and Treasurer is usually all that the Committee can do to come up with There would be a significant cost associated with an one nomination! election.
14 SSAI’s Canberra office has moved – The new postal address is: PO Box 5111, Braddon ACT 2612 SSAI Newsletter – February 2005 The telephone remains the same (02) 6249 8266 – Our new Fax (02) 6249 6558 Conferences Society Secretaries Central Council IWMS-2005: 14th International Workshop on Matrices and Statistics President: Mr N. Bartlett 29 March – 1 April 2005, Auckland, New Zealand Secretary: Dr D. Shaw http://www.iwms2005.massey.ac.nz Email: doug.shaw@csiro.au
55th Session of the International Statistical Institute New South Wales 5-12 April 2005, Sydney President: Mr Alun Pope [email protected] Secretary: Dr Neville Weber Email: [email protected] ISBIS – Fourth International Symposium on Business and Industrial Statistics 13-16 April 2005, Palm Cove, Queensland Canberra http://www.action-m.com/isbis4/ President: Dr Ann Cowling Secretary: Ms Anna Poskitt IASS 55: Complex sampling, retrospective sampling and missing data Email: [email protected] A conference in honour of Alastair Scott, International Association of Survey Statisticians Victoria 13-14 April 2005, Auckland, New Zealand President: Dr K. Lipson http://www.stat.auckland.ac.nx/iass55 Secretary: Mr B. Fraser Email: [email protected] Recent Advances in Biometrics, Bioinformatics and Markov Chain Monte Carlo 7-8 July 2005, Sydney South Australia [email protected] President: Dr Alan Branford Secretary: Mr Patrick Lim 20th International Workshop on Statistical Modelling Email: [email protected] 10-15 July 2005, Sydney [email protected] Western Australia President: Dr Aloke Phatak Stochastic Modelling of Complex Systems (SMOCS-05) Secretary: Ms Jodie Thompson 10-16 July 2005 Daydream Island, Queensland Email: [email protected] www.conferences.unimelb.edu.au/smocs05 Queensland President: Dr Bronwyn Harch Secretary: Professor Kerrie Mengersen Email: [email protected]
Section Chairs
Statistics in the Medical Sciences Mr Peter Howley [email protected] Statistics in the Biological Sciences Dr Simon Barry Hosted By Email: [email protected] The Statistical Society of Australia (SSAI) and The New Zealand Statistical Survey and Management Dr Robert Clark Association (NZSA). Email: [email protected] Contact Details Statistical Education Dr Michael Martin ASC/NZSA 2006 Conference Managers [email protected] GPO Box 128 Expression of Interest Statistical Computing Sydney NSW 2001 If you are interested in attending Associate Professor Kuldeep Kumar Phone: +61 2 9265 0700 the Conference, please register Email: [email protected] Fax: +61 2 9267 5443 your interest on-line Industrial Statistics Email: [email protected] www.statsnz2006.com Dr Aloke Phatak Email: [email protected] Scientific Program Young Statisticians A stimulating and cutting edge Scientific Program is being developed to cover a Ms Anna Munday wide range of topics relevant to all statisticians. The program will provide practical Email: [email protected] knowledge and insights from prominent international and Australasian speakers and Bayesian will address the latest developments in statistical research, education and practice. Professor Kerrie Mengersen Email: [email protected] Workshops Further contact details for Technical workshops that are of particular interest to practitioners will be included Society Secretaries and in the Conference Program. The Scientific Program Committee is seeking potential Section Chairs can be obtained workshop presenters. If you are interested in contributing please contact David by contacting the Society on Scott at [email protected]. (02) 6249 8266
SSAI Newsletter – February 2005 SSAI’s Canberra office has moved – The new postal address is: PO Box 5111, Braddon ACT 2612 15 The telephone remains the same (02) 6249 8266 – Our new Fax (02) 6249 6558 SSAI Review of Statistics at Australian Universities Update
The Review is now well underway with the review team visiting Canberra, Sydney, Brisbane, Melbourne, Adelaide and Perth between 14th and 25th February 2005. Meetings for individuals and groups with the team are being arranged in each city.
Canberra – 14th/15th February 2005 Melbourne – 21st/22nd February 2005 Contact: Alan Welsh ([email protected]) Contact: Neville Bartlett ([email protected])
Sydney – 16th/17th February 2005 Adelaide – 23rd/24th Febraury 2005 Contact: David Steel ([email protected]) Contact: (Academics) Chris Brien ([email protected]) or (Industry) Margaret Swincer ([email protected] Brisbane – 18th February 2005 Contact: Tony Pettitt ([email protected]) or Perth – 25th February 2005 Bronwyn Harch ([email protected]) Contact: Jodie Thompson ([email protected])
The current timetable for the remainder of the review is as follows: Visit by Review Team: 14th – 25th February 2005 Draft report sent to SSAI: 15th April 2005 SSAI response sent to Review Team: 17th June 2005 Final report sent to SSAI: 29th July 2005 AusCan Scholar Program
A scheme to promote scientific exchanges between outstanding Australian and Canadian statistical researchers.
The AusCan Scholar program was developed by the The Scholar will be required to provide a 2-page report Statistical Society of Australia Inc and Statistical Society within one month of completing the visit outlining activities of Canada to foster young researchers and promote carried out, key contacts made and expected follow-up collaborative activities between the societies. collaborative activities. The report will also be published on the SSAI and SSC websites and the Scholar may be The objectives of the AusCan Scholar program are: requested to make a presentation at his/her Society’s national conference. 1. Promote scientific interaction between the Australian and Canadian statistical communities, particularly in Selection of the 2005 scholar will be made by a Committee areas of Statistics relating to current and important chaired by the SSAI President and a representative of the practical problems. Canadian Statistical Society. The costs of this program will be shared by SSAI and SSC. Funding is available to 2. Provide the opportunity for outstanding young Canadian/ the Scholar to cover airfares, accommodation, meals and Australian researchers in these areas to visit a number incidental expenses. Scholars will be responsible for their of leading research centres in Australia/Canada, to own travel insurance. present their current research and interact with a number of researchers in that country. Applications (maximum of two pages addressing the eligibility criteria detailed above) current CV and the names Applications are sought from Australian statisticians and contact details of three referees should be forwarded to: who wish to be considered for the position of the inaugural Jane Waslin AusCan Scholar to visit Canada in 2005. The visit will be Executive Officer of 4-8 weeks’ duration, one week in each of a few cities, Statistical Society of Australia Inc and extendable if made in conjunction with participation PO Box 111 in a national conference. The primary objective is to move Braddon ACT 2612 around the country to meet people. Australia or faxed to (02) 6249 6558 Eligible scholars will be post-PhD with a strong by close of business, Friday 11 March 2005. preference given to researchers who are within five (active research) years of gaining their PhDs, with a demonstrated The Scholar for 2005 will be announced by the end of strong interest in both theory and applications and members April 2005 and the visit must be completed by the end of of the Statistical Society of Australia. 2005.
16 SSAI’s Canberra office has moved – The new postal address is: PO Box 5111, Braddon ACT 2612 SSAI Newsletter – February 2005 The telephone remains the same (02) 6249 8266 – Our new Fax (02) 6249 6558 Branch Reports
WESTERN if so what is the information going to be Tropics” was held on North Stradbroke used for? Island, just off the coast of Brisbane. A AUSTRALIA cluster of statisticians, some like me with November 2004 very vague priors, boated across Moreton In the October talk of the SSAIWA After the interest created in microarray Bay to North Stradbroke Island to attend Branch Dr Nazim Khan spoke to the analysis in the October talk, Marty Firth title “On a resampling approach for tests a 3 day workshop. The meeting started on stepped into the breach to speak at the on the number of clusters with mixture Tuesday the 5th and over the following November meeting of the SSAIWA model-based clustering of tissue samples” 3 days participants were engrossed in branch to the title ‘A Statistician’s which in fact is the title of a joint paper by discussion of Bayesian statistics from Introduction to Microarrays’. Marty was McLachlan and Khan (2004) in volume 90 advanced topics to practical problem well equipped to speak to this topic as pp.90-105 of the Journal of Multivariate solving. The workshop provided a forum he is employed as a biostatistician at Analysis. This work came about when for the analysis of several case studies in the Telethon Institute for Child Health Nazim was doing postdoctoral work the areas of ecology, epidemiology/health Research where his interests include at the Department of Mathematics, and genetics. Special topics included applied statistics in the areas of vaccine University of Queensland following the trials, asthma, and leukaemia research Bayesian modelling, expert opinion and completion of his PhD at the University and microarray analysis. He has been priors, computational issues and MCMC. of Western Australia. Nazim currently involved in microarray analysis for the The workshop presentations began with holds a lecturing position in the School last 2-3 years and in fact some of the guest speaker Professor Judith Rousseau, of Mathematics and Statistics at the graphs shown in “Chloe’s Story” on University of Western Australia. Dauphine University, CEREMADE, Channel 2 were his. Paris and followed with participant The talk was essentially on the use of After giving a brief view of the Telethon presentations with almost all workshop mixture model-based clustering to analyse Institute for Child Health Research and participants presenting some work gene expression micro array data. It was his background as a statistician Marty inferred that the statistical methodology directly involving Bayesian methods talked biology explaining DNA, RNA, mentioned has an important role in the or analyses that they believed could be and gene expression arrays. He talked discovery, validation and understanding improved using Bayesian methods but about array design, genechip probe of various classes and subclasses of cancer. were not sure how they should proceed. arrays and cDNA microarrays. Then In particular the talk considered a mixture Approximately 30 participants enjoyed turning probe intensities into a single the informal atmosphere that generated model-based approach to clustering of probe set expression, and including great discussions. tissue samples of a very large number of background correction, summarizing and genes from microarray experiments. A normalization finally leads to statistical It was not all hard work. Thursday feature of microarray data that one has analysis. The latter involves common evening saw participants test their skill to deal with is that the data involve n, the questions: Which genes are differently (mainly skills associated with cheating) at number of tissue samples in the order of expressed? How do the patients cluster? How Pictionary. A singing competition saw Rob 10 – 102, and the number of variables p, do the genes cluster? Which genes best classify Reeves’ musical talents prove themselves the number of gene expressions in the 3 4 the groups? Which genes are changing over by providing witty and poignant lyrics to order of 10 – 10 . Standard statistical time? methodology is appropriate for n>>p but the song `There is a chapel in the town’ Marty pointed out the importance we have here p>>n. for his group to take the prize for the of working with biologists in his talk best song of the workshop. I have placed In classifying tissues, e.g. one may have and noted the vast number of statistical a copy of the words at the end of this 2 classes such as disease free or metastases, procedures applied to microarray data. He report. Nazim introduced in his talk the use then spoke of his own research involving of mixture models, adopting elliptically classification trees and random forests. Preceding the Stradbroke Island symmetric components such as the After the talk interest appeared to be in workshop a 1-day workshop was held normal or t-densities. By including an whether the different statistical analyses in Brisbane titled `Practical Bayes for additional parameter, the degrees of were arriving at the same conclusions Beginners’. Some of the participants from freedom, these fall into the one framework and indeed whether there is an ‘optimal’ this workshop found their way out to and the degrees of freedom can be approach. `Straddie’. estimated from the data. By including Brenton R. Clarke The venue, the University of t-densities the assertion is that the methodology is robust against outliers. Queensland’s Moreton Bay Research Questions of the order of a mixture model Station, was a great choice. Lunch and involve bootstrapping of the Likelihood QUEENSLAND dinner were provided at the Little Ships Ratio Statistic. A discussion of published Club a small walk from the Research data and comparison using the EMMIX- What were those statistical Station. Several participants accepted GENE software ensued. sightseers doing? Kerrie Mengersen’s offer of transport Questions following the talk reflected It is hard to resist using more catchy to several swimming spots at various the growing interest in microarray data headlines like `Moreton Bay BUGs sheltered and not-so-sheltered beaches and in particular the motivation behind Meet’ or `BAYESWATCH’ to describe and lakes. Many thanks go to Petra such studies. Just as an example, are we the Bayesian workshop held in October Kuhnert and Kerrie Mengersen for interested in identifying particular groups on Moreton Island. The Inaugural ASBA their efforts in organising such a great of genes as reflecting a link to cancer and Bayesian Retreat: “Bayesian Topics in the workshop.
SSAI Newsletter – February 2005 SSAI’s Canberra office has moved – The new postal address is: PO Box 5111, Braddon ACT 2612 17 The telephone remains the same (02) 6249 8266 – Our new Fax (02) 6249 6558 Branch Reports
“There is a chapel in the town” lyrics by CANBERRA The Knibbs Lecture 2004: On Rob Reeves. selling cheap things, by Adrian Baddeley There is a chapel in the town, in the town Talk on the analysis of At the Australian National University And there my true Tom lays him down, lays microarray data by Peihua Qiu (ANU) in Canberra on Tuesday the 2nd him down At the monthly meeting of the Canberra of November, Professor Adrian Baddeley And I’ll hang my career on subjectivity Branch of the SSAI on 26 October 2004, of the University of Western Australia And may the world go well with thee Associate Professor Peihua Qiu of the presented the Knibbs Lecture 2004, titled: A man, Ron Fisher was his name, was his School of Statistics at the University of “The Difficulty of Selling Cheap Things”. name Minnesota gave a talk titled “Segmentation Adrian began by noting that although Thought that he would change the game, Of cDNA Microarray Images By Local statistics is potentially useful in every change the game Smoothing”. This presentation was a field of science, there is no obligation “It isn’t real till there’s a frequency” summary of joint work with Mr Jingran on scientists to use it. Indeed there is And may the world go well with thee Sun and a foretaste of Peihua’s book plenty of resistance to it, as typified by “Image Processing and Jump Regression Einstein’s famous quip that God does He had hypotheses to test, yes to test Analysis” which is soon to be published not play dice. There is also no guarantee His way seemed better than the rest, yes the by Wiley. that statistics will be used correctly. For rest Gene microarray data are widely example, it is widely agreed that roughly With rejections based on alpha and p employed in pharmaceutical and clinical 75% of medical publications contain We’ll know the world objectively research. By comparing gene expression statistical flaws which are severe enough But experiments and trials must be designed, in normal and abnormal cells, microarrays to potentially invalidate the conclusions. yes designed can be used to identify genes involved in One of the most perplexing forms of And there we’re sure that you will find, you particular diseases, allowing these genes resistance to statistical thinking is the will find to be targeted by therapeutic drugs. Most disregard for efficiency, in the sense of A healthy dose of subjectivity gene expression data are produced from performance per unit effort. For example, To make the world go well with thee cDNA microarray images. A microarray many scientists would instinctively image consists of thousands of spots, disbelieve that it is possible to accurately Now when I’ve a matter to infer, to infer with individual DNA sequences printed estimate the average income of all one To Reverend Bayes I will defer, will defer at each spot first and then equal amount billion persons in India using a sample And I’ll draw my conclusions a-posteriori of cDNA samples from treatment and fraction of only one millionth. Such And praise the word MCMC. control cells mixed and hybridized skepticism is related to a general tendency October meeting with the printed DNA sequences. To within science to prefer expensive and obtain gene expression data, the image complex technology over cheaper, simpler Members were addressed at the needs to be segmented first to separate and equally effective methods for the October 12th meeting by Professor Judith foregrounds from backgrounds for same task. Thus, increasing the efficiency Rousseau (Unniversité de Paris 5) on individual spots, and then averages of the of statistical methods may in fact make work done by herself and Brunero Liseo foreground pixels are used for computing them less attractive. This paradox is (Università di Roma “La Sapienza”) titled the gene expression data. Several image part of a broader phenomenon called “Nonparametric Bayesian estimation of segmentation procedures have been “the difficulty of selling cheap things”, the spectral density for Gaussian long- suggested and included in software whereby consumers do not always memory processes and semiparametric packages for handling gene microarray choose the commodity which offers the estimation of the long-memory data, such as ScanAlyze, Genepix, best value for money (e.g. a cheap wine or parameter” at the Queensland University Quantarray and SRG (the Seeded Region car), and whereby lovers sometimes play of Technology. Growing procedure). From the audience ‘hard to get’. The talk discussed their study of Professor Sue Wilson of the Australian One prominent example of this the estimation of the spectral density National University also mentioned Spot, phenomenon is in the field of stereology, of a Gaussian long memory processes a package developed at the CSIRO. which has to do with extracting which has been extensively studied in Having outlined the nature and uses of quantitative information from microscope the frequentist literature, where rates microarray data, Peihua presented a new images. Suppose we are interested in V, of convergence have been obtained for image segmentation methodology based the average fraction of volume in a solid some classes of functions. However the on local linear kernel smoothing. He rock that is occupied by some mineral speakers considered a Bayesian non- described several theoretical arguments of interest M. Then one estimate of V parametric approach and discussed how and numerical studies which show is A, the fraction of area occupied by M they obtained a general result giving that this methodology has some good on a polished plane section of the rock. consistency and rates of convergence, statistical properties. After illustrating Another estimate of V is L, the length under general conditions on the prior with an application to the study of the fraction of equally spaced lines on this distribution. Further work examined a defense reaction of Arabidopsis leaves, plane which lie over M; and yet another class of prior based on the FEXP models, Peihua concluded with some avenues for is P, the fraction of equally spaced points for which they gave precise rates of future research. These avenues include on these lines which cover M. convergence. Semiparametric estimation robustification of the procedure and the It can be shown that if the said plane, of the long memory parameter was also post smoothing of estimated boundary lines and points are selected in a certain discussed and consistency was shown. curves. way, then A, L and P are all unbiased for V. Brenton Clarke Borek Puza Of these three, P is the estimate of choice,
18 SSAI’s Canberra office has moved – The new postal address is: PO Box 5111, Braddon ACT 2612 SSAI Newsletter – February 2005 The telephone remains the same (02) 6249 8266 – Our new Fax (02) 6249 6558 Branch Reports Continued
NEW SOUTH WALES
NSW Branch Annual Dinner Outlier-Robust Model Selection Alan Welsh The J. B. Douglas Postgraduate Awards were followed by the Annual Dinner of the NSW Branch. Guest speaker for the occasion was Professor Alan Welsh from the CMA at ANU. Alan wanted to share with us some of his work on model selection methods that are robust against outliers in the data. However, he began with the unwelcome discovery that his CD, written on a Mac computer, could not be read by the PC-based system Alan Welsh, Adrian Baddeley and Simon Barry at the Knibbs Lecture. at the venue and that his presentation would have to be hand-written on the on the grounds of efficiency, and A is the are typically much easier to compute than two sides of an electronic whiteboard. least efficient. However, since A was first MLEs whilst being reasonably close to In spite of this the audience was given a proposed in 1847 by the French geologist them. For these and other reasons, Adrian wonderfully clear and concise exposition, A.E. Delesse (a great breakthrough at the envisages that future developments in demonstrating that lack of technology is time), it took the scientific community inference for highly structured systems no impediment to quality of presentation about 50 years to accept L as better, and will be working at the “cheap end” for a speaker clearly as robust as his then another 30 years to finally accept P as (pseudolikelihood) rather than the selection methods ! best. Thus, whilst the idea of progressively “expensive end” (MCMC maximum The work Alan presented was motivated subsampling from A to L to P may seem likelihood). by the need for models in large chemical obvious to modern-day statisticians, each Adrian concluded by saying that since data sets for which existing selection successive step encountered long and stiff efficiency seems not to be the highest methods often fail because the expected resistance historically. priority for many users of statistics, we prediction loss function does not have the Adrian went on to discuss the variances need either to understand and accept usual smooth, unimodal variation. The of the three estimators, and noted some these users’ priorities, or to become more criterion function used in model selection interesting paradoxes. For example, in persuasive. contained terms penalising poor fit to some situations P may be more accurate The first discussant, Professor Alan the given data and poor prediction of the than A, which is counterintuitive since A is Welsh of the ANU, commended Adrian response at new data (prediction loss). based on full information from the section for a highly interesting, stimulating and The key to making the selection method plane. Adrian resolved this paradox by provocative lecture, and noted the absence (minimisation of the criterion function) considering covariance contributions, of the frequentist-Bayesian debate. He robust against outliers lay in breaking the and explained why the Rao-Blackwell also raised questions regarding whether link between the form of these penalty theorem is not applicable in this case. some of the ideas in Adrian’s paper functions and that used in the calculation He also gave an example of ignorance extend to the infinite case, and mentioned of model parameters. Thus, if for example regarding sampling methods in the the tendency of users to love simplicity least squares regression is used to obtain journal Trends in Neurosciences. Several the models, the function used in obtaining even when wrong, for example to blindly 2 letters were recently published therein apply ANOVA to everything, and to use the parameters is r(x) = x , but the penalty which heatedly but incorrectly argued linear regression for binary data (which functions used in the selection criterion that sampling techniques are not reliable however is not always wrong). The function may be truncated versions of r for estimating the number of neurons in second discussant, Dr Simon Barry of which are constant if the magnitude of x the brain on account of it having a highly the Bureau of Rural Sciences, noted that exceeds a certain value. In this way, large organised architecture. sometimes statistical methods are used models which “fit” outlying data are not Adrian’s second main illustration of “the to produce an elegant solution to the necessarily favoured over smaller models difficulty of selling cheap things” was in wrong problem, and that there is not which do not. the sphere of Markov chain Monte Carlo enough penalty for making mistakes. He The calculation of the prediction (MCMC) methods. After listing several also questioned whether accuracy per loss term, strictly speaking, requires uses of these methods, e.g. for estimating unit of computer time is really a good new data. If it can be assumed that maximum likelihood estimates (MLEs), he measure of efficiency, and drew some the data to which the model will be noted some of their disadvantages, such very entertaining parallels with boating applied has the same distribution as that as being highly computer-intensive. This habits. The lecture and discussion were used to derive the model then bootstrap led him to describe a simple alternative followed by a sumptuous dinner in the sampling from the latter can be used technique called maximum pseudolikelihood Drawing Room of the ANU’s University to simulate “new” data. Ordinary (MPL). Although this technique may be House, with food provided by Boffins bootstrap sampling does not work well, criticised for being an ad hoc invention, in Restaurant. however. Instead, m out of n sampling high dimensional models MPL estimates Borek Puza is needed, where m/n is in the range 0.25 – 0.5. Unfortunately, this introduces the
SSAI Newsletter – February 2005 SSAI’s Canberra office has moved – The new postal address is: PO Box 5111, Braddon ACT 2612 19 The telephone remains the same (02) 6249 8266 – Our new Fax (02) 6249 6558 Branch Reports Continued problem that bootstrap samples may be chosen which contain a large proportion of outliers. The innovation proposed by Alan to overcome this difficulty is the use of stratified bootstrap sampling. A stratification is imposed on the model residuals and bootstrap sample allocated proportionately to the resulting strata. Alan presented some results which indicated that the method works well even when the proportion of outliers in the data set is as high as 25%. For higher values, the method begins to perform poorly, essentially because of bias in the estimation of population variance. These are high values for proportions of outliers, but may not be unrealistic in situations where simple modelling is used to explore large and rich data sets. The presentation gave rise to some penetrating questions and observations from the audience after which technical discussion gave way to the relaxed conviviality of the annual dinner. Jos Beunen
J. B. Douglas Postgraduate Clair Alston (second from right), winner of the Peter Wright Prize, with Christine Awards Curran, Jim Douglas (left) and Alun Pope. The fifth annual J. B. Douglas Postgraduate Awards event was held on students, late in their research programs difficult, because of the high quality of the afternoon of 24 November 2004 at the for masters or PhD degrees, and are the presentations, but the decision was to University of Sydney. It was followed decided on the basis of 20-minute award the Peter Wright Prize and a cheque by a talk by Professor Alan Welsh from presentations. As Branch President, I was for $500 to Clair Alston (Newcastle) for the ANU, and then the Branch Annual chair of the judging panel and the Branch her talk on Bayesian mixture models for CAT Dinner the same evening. was delighted to have Alan Welsh and Dr scan interpretations. Prizes of $400 each Philip McCloud of Roche Pharmaceuticals were awarded to Daniel Melser (UNSW) The NSW Branch created these awards as the other members of the panel. and Carole Birrell (Wollongong). to encourage excellence in postgraduate work in any area of Statistics or Talks – all of a very high standard, both Before the presentation of the awards, Econometrics and to recognise the in presentation and content – were given Associate Professor Ken Russell of contributions to the profession of by the following students, nominated by the University of Wollongong gave a Professor Jim Douglas. This year, for the their departments: presentation entitled Remembering Peter first time, the first prize was dedicated to Alistair Merrifield, University of Wright, in which he painted a quick, the memory of Dr Peter Wright, a former Sydney (Statistics) affectionate picture of the man. The Branch was honoured to have as its Branch Councillor, who died suddenly Shuling Chen, UNSW (Statistics) in May. guest Christine Curran, who presented Piea Peng Lee, Macquarie University the award named in honour of her late The awards are open to graduate (Statistics) husband. Daniel Melser, UNSW After Alan Welsh’s talk (reported (Econometrics) elsewhere in the Newsletter) members Clair Alston, University and guests celebrated the end of the year of Newcastle (Statistics) at a well-attended Annual Dinner. Carole Birrell, University Special mention and thanks must of Wollongong (Statistics) go once again to Dr Shelton Peiris The event was attended (University of Sydney) for organising by about 60 people such a successful event. The event also and it was good to see could not have taken place without that the members and the generous support of our sponsors: supporters included CSIRO, Macquarie University, Roche sizable contingents from Pharmaceuticals, SAS Australia, St George Newcastle, as well as other Bank, UNSW Department of Statistics out of town centres such as and the Wollongong Statisticians. The judges (from left, Philip McCloud, Alan Welsh, Wollongong. The task of Alun Pope Alun Pope) had a difficult task. the judges was extremely
20 SSAI’s Canberra office has moved – The new postal address is: PO Box 5111, Braddon ACT 2612 SSAI Newsletter – February 2005 The telephone remains the same (02) 6249 8266 – Our new Fax (02) 6249 6558