ewsletter December 2007 Number 121 N statistical society of australia incorporated Counting Australia In

The Editors invited Professor Eugene Seneta to review “Counting Australia In”, the first and only book on the history Graeme Cohen. Counting Australia In. The People, Organisations and Institutions of Australian Mathematics. With a foreword by of the mathematical sciences, including many aspects of Lord Robert May of Oxford. Halstead Press in association with statistics, in Australia. The result is a combination of personal the Australian Mathematical Society. Sydney, 2006. 431pp. history, reflection on the content of the book, and expansion ISBN 1 920831 39 8. on the book’s statistical aspects. The author stresses in his Preface, dated August 2006, that Chapter 10 is the requested history of the Australian Mathematical Society, Introduction but the book should not be regarded as a history of the Australian mathematics generally. Its scope is indicated by its subtitle, and it There is a great deal about statistics, from official to covers the period from Australia’s colonial beginnings (1788) up mathematical, in the book, and I write this review primarily to about 2005. as a statistician for Australian statisticians. We are part of The following is a table of contents and brief description of the the mathematical community. Evgenii Slutsky, in the Preface chapters. to his first publication, a book in Russian of 1912 entitled Foreword, Preface, Main abbreviations. Correlation Theory which to a large extent brought Pearsonian Chapter 1. Mathematics and the Beginnings of the Colonies. statistics to the Russian Empire, said “Every statistician should Chapter 2. Mathematics and the Rise of the Universities. (Sydney, be a mathematician first, for our science is a mathematical Melbourne, Adelaide, Tasmania). science.” Slutsky started his career as an economic statistician, Chapter 3. Mathematics Outside the Universities. (Royal Societies, and oscillated between this and mathematical statistics, notably Australasian Association for the Advancement of Science, time series analysis and related probability theory dealing with Government and other statisticians, William Farrer.) statistical dependence. Economic statistics was perhaps the most Chapter 4. Mathematics in the Universities in the First Half of the important stream of statistics to interact with Pearson’s English Twentieth Century. (Queensland, Western Australia, Sydney, New Biometric School to form a unified coherent discipline, while England University College, Melbourne, Adelaide, Tasmania.) retaining a separate quantitative identity. While the present Chapter 5. Australia’s Mathematicians in World War 2. book addresses economic statistics/econometrics up to mid 20th Chapter 6. Post-War Mathematics in the Older Universities. century, more would have been welcome about the academic Chapter 7. Mathematics in Canberra’s Colleges and Universities. subject area in Australia, as also about public health statistics, Chapter 8. National Organisations and Mathematics. (Australian biostatistics and epidemiology here in the latter part of the Bureau of Statistics, CSIR and CSIRO, Academy of Science, “Adding to Australia.”) century. Chapter 9. Mathematics and the Later Universities. (Newcastle My review of this record is circumscribed by my own and Wollongong, Monash, other Universities of the 1960’s and background experience within the community whose social 1970’s, former Institutes of Technology, Universities founded since history it chronicles. I was a student in the then Department of the late 1980’s.) Mathematics, University of Adelaide in the years 1960-1964, Chapter 10. The Australian Mathematical Society. with a 4th Year Honours project on random walks supervised Appendix 1. A Snapshot of Australian Mathematics, 1914. by Professor Ren Potts, and considerably influenced by Mrs. Appendix 2. “The Spirit of Applied Mathematics,” by K.E. Bullen. Marta Sved. My M.Sc. work in 1964 was supervised by Dr. Appendix 3. Lists (mainly relating to the Australian Mathematical John Darroch, who was in the Adelaide department for two Society) . years within the period 1962-1964. He returned to that city Endnotes. (pp.381-410. These contain sources.) Name Index. (pp. as Professor (1966-1996) at the newly established Flinders 411-423.) University. The 4th year Honours class in 1963, due to hard work General Index. (pp.424-431.) on restructuring courses by the two Professors, Eric Barnes and The author retired from the Department of Mathematical Sciences Ren Potts, consisted of a record number (about 16) of students, at the University of Technology, Sydney (UTS) in 2002 after 36 and included Rod Worley, Ann Pearson, Ken Pearson, Pender years there, experiencing its name changes and fortunes. This book, his first serious publication outside mathematics, has led Pedler and Tony Andrews who went on to academic careers in to an M.A. in Public History from UTS to add to his M.Sc. (Sydney) Australia.. Others in the group included Brian Bennett, Glen and Ph.D. (NSW). The book is the culmination of the Australian Thompson, C. K. Cheong, Winnie Tye, P.S. Khoo, T.C. Chye, Mathematical Society’s History Project, and has its origins in late E.B. Ooi, Baiba Vitols, and John Tomlin. 1963 was a fateful year 1999. The book is sui generis. Nothing of this scope has been in many ways. Returning by car from my last examination, in undertaken previously, and for one researcher to undertake it and to succeed is astonishing. It brings the author’s career to an Science German, I saw a poster for Rupert Murdoch’s Adelaide illustrious end. He will be remembered. continued on page 2 Counting Australia In

From page 1 Alfred (Alf) Cornish (1909-1973), a Statistics, and arrived at about the same disciple of the great statistician Ronald time, first meeting in Pat Moran’s office News outside a shop, that President John Aylmer Fisher who was spending the last in January 1965. Cheong, my classmate F. Kennedy had been assassinated. years of his life in Adelaide at the time. from Adelaide, had just completed his In the years 1960-1964 the academic Kerwin Morris encouraged the Statistical M.Sc. in the SGS Department, and was staff of the Mathematics Department at Methods class to attend a seminar by likewise about to start his Ph.D. enrolled various times also included Jim Michael, Fisher, but I saw him only through the in Moran’s department. Maurice Brearley, David Elliott, Basil open door of the seminar room. I did I came to the University of Sydney Rennie, George and Esther Szekeres, later attend one lecture by Cornish, but it as Professor and Head of the then Reyn Keats, Maurice Grey, Kerwin was too full of trigonometric expressions Department of Mathematical Statistics Morris, Alf Cornish , Jane Pitman, and for me to continue. in mid-1979, succeeding the foundation Rainer Radok (who had been one of the In late January 1965 I joined the then Professor, Oliver Lancaster, with whom arrivals in Australia on the Dunera). Most Department of Statistics of the then I shared an interest in the history of of these occupy prominent positions in School of General Studies (SGS), at the probability and statistics. The Professors Cohen’s account. Australian National University (ANU), as in the Department of Pure Mathematics Some younger postwar immigrants Temporary Senior Tutor, and stayed until at the time were Tim Wall and Max from Europe round the time of the war, mid-1979. This teaching department, Kelly, and in the Department of Applied like myself, were just then completing whose Head was Ted Hannan, was Mathematics, Peter Wilson. John their undergraduate studies. Another formally in the Faculty of Economics, Robinson was promoted to a Personal example, though not of Adelaide, who and for all of my stay there was composed Chair in Mathematical Statistics in 1991. figures prominently throughout the book, of both mathematical and economic I formally retired as Emeritus Professor and was closely associated with George statisticians, and catered also to students in mid-2004. The vacated position of Szekeres, a European refugee, is Alf van in the Faculties of Arts and Science. The Professor of Mathematical Statistics was der Poorten. The Adelaide 4th year names 4th year Honours students generally were filled by Neville Weber in August 2005. also show strong evidence of Colombo to be employed by the Canberra-based Both John and Neville were members of Plan students, which resulted in many Australian Bureau of Census and Statistics. the Department under Oliver Lancaster. friends for Australia, mainly in Malaysia The ones I supervised for their projects The book misses Neville’s elevation to and Singapore. I still maintain contact included Kathy Kang, Susan Pentony Professor, although it captures a few with Cheong (formerly Executive Officer (now Linacre), Phil Hughes and Suzanne events even from 2006, such as Peter of Singapore Airlines, and now on the Sheridan. My colleagues in the SGS Hall’s Flinders Medal and Lecture at the Board of OCBC) and Eric Ooi (we called Department over those years included Australian Academy of Science which him 001). The White Australia Policy Chip Heathcote, Warren Ewens, Chris actually occurred in April 2007, together was in effect at the time: I remember Heyde, Deane Terrell, Tom Valentine, with other awards of importance to some wag writing on the blackboard Ray Byron, Des Nicholls and Terry Australian statistics. O’Neill, all of whom eventually attained before one of John Darroch’s lectures My meeting with the book’s author, Professorial positions in Australia. in mathematical statistics “Keep out Graeme Cohen, took place over lunch on Bayesians, keep Australia bwhite [sic]”. The other statistical groupings in Wednesday, 30 June, 2004. Subsequently I was first taught mathematical statistics Canberra during my life there were Pat I provided hard copy of obituaries of at Adelaide University by Kerwin Moran’s Department of Statistics at the Harry Mulhall and Stephen Lipton, Morris, in a course then called Statistical Institute of Advanced Studies at ANU both of which had appeared in the SSAI Methods, a first course in the subject and the CSIRO Division of Mathematics Newsletter, and of my two obituaries of given at second year undergraduate level. and Statistics under Joe Gani. Oliver Lancaster, the more biographical Formally, the Professor of Mathematical Don McNeil and I had both came of which, written jointly with Geoff Statistics in Adelaide was Edmund to Canberra to undertake our PhD’s in Eagleson, appeared in the Historical

In this issue Counting Australia In 1 Society Awards 6 Editorial 4 Three Doors 7 Conferences 4 ANZJS Corner 8 President’s Corner 5 Farewell Jane Waslin 9 Why Accreditation is Critical 6 Spring Bayes Conference 9 Election of Executive Members 6 Branch Reports 13

 SSAI Newsletter – December 2007 Records of Australian Science (HRAS) late mainly in the Transactions/Proceedings of with the CSIRO, are well remembered in in 2004. Contact by email continued the Royal Societies of New South Wales, the book. The first two, under conditions till 12 July 2005, when I made some Victoria and Queensland. Publications of prevailing at the time, had to resign from minor corrections to paragraphs relating such “local” Royal Societies, in existence the CSIRO following their marriages. to Harry Mulhall and Oliver Lancaster. before the Federation in 1901 of the They were replaced by George McIntyre Michael Adena said when requesting six separate colonies, continued to play and Evan Williams respectively, both of a review, that I have experienced a an important part in the publication of whom had distinguished statistical careers. substantial part of the post-war history Australian research until airmail made McIntyre had studied under Weatherburn of mathematical science, and especially of access to newly “international” journals in Western Australia, and Williams under statistics, in Australia, and my interest in easier. A case in point is that of Dansie Pitman in Tasmania. history made me an eligible reviewer for Thomas Sawkins (1880-1950), who from the SSAI Newsletter. Michael and Alice 1922 to 1946 alone carried responsibility Early History and the Richardson wanted a less formal and of statistics teaching at Sydney University, Cambridge Influence chatty review than one which might be and published in the local journal which Chapter 1 is a prehistory of mathematics appropriate, say, for the Australia and New carried his obituary in 1950. (What in Australia, and is presented by the Zealand Journal of Statistics. Its personal evolved into the Australian Journal of author with a historian’s dedication. It perspective is also intended to provide Statistics did not come into existence till foreshadows its bright statistical future. additional information, in a small way, to the late 1940’s.) Arthur Phillip’s instructions on his that in the book. Sawkins’ work was publicized in the book appointment as “Captain-General and A First Course in Mathematical Statistics Governor –in-Chief of New South Some Statistical Highlights (1946), of his Australian contemporary Wales” in 1787 required him to make In his Foreword, Lord Robert May, Charles Ernest Weatherburn (1884- regular statistical reports. He is rightly formerly a Professor of Physics at Sydney 1974) whose name occurs at many remembered as “the first Australian University, finds a great positive in the points in Cohen’s book. Weatherburn statistician”. The first census of the “Australian ecumenism” of mathematics was Professor of Mathematics at the Colony of New South Wales was held which the book reflects, and points out the University of Western Australia since in November, 1828. The first Blue Books impact of Australian applied mathematics 1929. Both Sawkins and Weatherburn of this colony and the Colony of Van and statistics on the world stage, singling had applied unsuccessfully for the Chair Dieman’s Land were produced in 1822. out the mathematical genetics group of (Pure) Mathematics at the University Christopher Rolleston (1817-1888) stemming from Pat Moran. He points of Sydney after Carslaw’s retirement in presented a paper on the “science of out, however, that the predominant 1935. Sawkins was offered a Chair in statistics” at one of the first meetings of focus of the book on institutions is on Statistics in the Faculty of Economics, the Philosophical Society of New South mathematics departments. but took instead a Readership in 1938. Wales on 10 December, 1856. The first lecture on mathematics was given at the Some of the book’s highlights come Weatherburn had at one time taught University of Sydney in October, 1852, from archival research, especially on staff Edwin J. G. Pitman (1897-1993), but the emphasis in the early colonies was records of universities. Among these Professor of Mathematics for 36 years in application to areas of concern to the was the discovery of an application of from about 1925 at the University of colonists, such as astronomy, commerce, Thomas Gerald Room for the Chair of Tasmania, who was also an applicant for surveying and statistics, and was Mathematics in the University of Western the Chair at Western Australia. Together supported by enlightened administrators Australia in 1928, six years before he with Maurice Henry Belz (1897-1975) such as Governor Brisbane. The means gained the Chair at the University of and Cornish, they formed the nucleus of of elementary mathematical education Sydney, where he was Professor until Australian mathematical statistics in the were the schools and establishments of 1968. Material was also found illuminating first half of the 20th century. adult education, all amidst considerable the “background to the antipathy Weatherburn’s book is dedicated to sectarian influence. between Room and Keith Bullen” at the great statisticians of that period Karl this university. Other highlight material Pearson and Ronald Aylmer Fisher. The early history of mathematical includes the appointment of Thomas In “an exhaustive and masterly report science in Australian university Cherry as Professor of Mathematics at in 17 typed pages”, Karl Pearson had departments was considerably influenced the University of Melbourne, ahead of recommended Milne-Thompson by Cambridge University, especially its Norbert Wiener who was supported for the Tasmanian chair; Pitman had St. John’s and Trinity Colleges. The first by a stellar cast of European referees, not yet become interested in statistics. two youthful Professors of Mathematics including G.H. Hardy, David Hilbert, Later, Fisher played a decisive role in at Sydney University, Maurice B. Pell Henri Lebesgue, Paul Lévy and Maurice the appointment of Oliver Lancaster to (actually born in the U.S.) and Theodore Fréchet. Winer’s name now permeates the foundation Chair of Mathematical T. Gurney were Fellows of St. John’s, and the study of stochastic processes in Statistics at the University of Sydney over some of the other names of the early times continuous time and harmonic analysis, Geoff Watson in 1959. (G.R. Smalley, F.B. Horner, and E.M. Moors – the first locally born academic and he is credited with the discovery of The Cohen book has (p.255), courtesy mathematician and statistician) were also cybernetics. of the CSIRO, a photo of Helen Newton St. John’s men. The first two mathematics The book’s author was particularly Turner and Sir Ronald Fisher sailing on Professors at Melbourne University, excited about the discovery of the Sydney Harbour in 1960. Betty Allan, William P. Wilson and Edward J. Nanson, mathematical work of the surveyor Mildred Barnard and Helen Newton were associated respectively with St. John’s Martin Gardiner in 1850-1870. He Turner , who had all studied with Fisher lists some 16 of Gardiner’s publications, and who were the first three biometricians continued on page 10

SSAI Newsletter – December 2007  Editorial

In this newsletter we are delighted to please inform the Society office so that PO Box 5111, feature a book review that we hope will arrangements can be made for you to be of interest to a wide range of Society receive the newsletter. Braddon ACT 2612 members. Professor Eugene Seneta has Phone (02) 6249 8266 To close this editorial, we note that by done so much more than merely tell the time this newsletter arrives, the silly Fax (02) 6249 6558 potential what to expect in the book. season will be under way. The Editors Email: [email protected] If you have read a book that would be would like to take this opportunity Society Web Page of broad interest to Society members, to wish all Society members a happy http://www.statsoc.org.au please write in and tell the Editors! Or, Christmas, and a pleasant New Year. if you have read a book that deals with a We would also like to thank all those subject about which you have some extra who have contributed to the newsletter, Editors information to impart, write in and tell whether by writing reports, taking the Editors too! photos or entering competitions. Alice Richardson, School of ISE, This will be the last time a paper copy Thankyou also to the organizations University of Canberra, of the Newsletter arrives in your mailbox. who have supported the activities of PO Box 1, Belconnen ACT 2616 Please remember to keep the Society up- the Statistical Society and its members to-date with you current email address, – your input into successful Society Michael Adena, Covance Pty Ltd and check the SSAI website (http:// activities is much appreciated. PO Box 5125, Braddon, ACT 2612 www.statsoc.org.au) regularly in order to The deadline for copy for the next keep in touch with Society activities. If newsletter is 10 February 2008 – enjoy Correspondence you do not have access to web or email, the summer! Please direct all editorial correspondence to Alice Richardson. Email: [email protected] Member News Disclaimer The views of contributors to this President of the Islamic Countries Newsletter should not be attributed to Society of Statistical Sciences (ISOS), Dr the Statistical Society of Australia, Inc. Shahjahan Khan (pictured), is scheduled to present a keynote speech in the First Subscriptions Arab Statistics Conference (FASC) organised by the Arab Institute for Training The Newsletter of the Statistical and Research in Statistics (AITRS) to Society of Australia is supplied free to be held during 12-13 November 2007 in all members of the society. Any others Radisson SAS Hotel, Amman Jordan. wishing to subscribe to the Newsletter The AITRS was established in 1971 to may do so at an annual cost of raise competencies and skills of the Arab A$30.00 for an issue of four numbers. Statistical Organisation. The conference is organised under the Patronage of his Royal Advertising Majesty King Abdullah II Ibn Al-Hussain Advertising will be carried in the of Jordan, and the session in which Dr based academic, Dr Khan will speak on the Newsletter on any matters which Khan will present his keynote speech will Importance of Statistics in Development in the Editors feel are of interest to be chaired by the Minister of Planning line with the theme of the conference “No the members of the Society. For of Jordan, Dr Suhair Al-Ali. Australia Development without Statistics.” details of advertising rates, etc. contact the SSAI Executive Officer at [email protected]

Printer Conferences National Capital Printing 22 Pirie Street, Fyshwick ACT 2609 Australian Statistical Conference 2008 30 June-3 July 2008, Melbourne, VIC — http://www.asc2008.com.au International Society for Bayesian Analysis Conference DEADLINE FOR 21-25 July 2008, Hamilton Island, QLD — http://www.isba2008.sci.qut.edu.au/ NEXT ISSUE: Australian Statistical Conference 2010 10 February 2008 6-10 December 2010, Perth, WA

 SSAI Newsletter – December 2007 President’s Corner

Shortly you will receive an invitation by Since the last newsletter, I have emailed and workshop email to participate in a member survey members seeking ideas for professional activities. Before commissioned by Council in order to obtain development and other specialised Council proceeds your views. Data Analysis Australia has workshops. I have also emailed the Chairs with filling this agreed to administer this web based survey of Sections seeking their input. There has position, on a and I thank them for their generous support been a good response to these requests part or full time basis, it has of this activity. Please take the 10 to 15 and I am currently compiling a summary of suggestions for wider circulation and requested that a minutes that will be required to complete it. discussion. If you have not already contacted business plan for It is critical that we have relevant and up- me about possible workshop or professional these activities to-date data on the membership and their development ideas I would be pleased to be prepared by views. Like all surveys, there may be areas hear from you. To those who took the the Executive against which the costs of that you particularly care about but which trouble to contact me, I am very grateful such a position would be allocated. are not covered. We plan future, shorter and and ask for your patience if you have not The August decision of Central Council more focussed surveys over the coming year. received a reply from me as yet. to move the Newsletter to electronic access However, please bring any areas of concern Council met via teleconference in early only was discussed and it was decided to to your Branch or Section representatives or November to pursue the main action items revisit that decision once the results of to members of Council including me. arising from the August Central Council the member survey are available, and after If you do not receive the email regarding meeting. We plan on meeting every three evaluation of lower cost options for printing the survey, please contact the SSAI office to months in this way to ensure that plans the newsletter are assessed. request a survey form to complete. and actions are pushed forward. Council As many of you will be aware Jane Waslin, amended regulations to formalize the our Executive Officer for the past 4 years In this newsletter you will find registration role of Branch Presidents’ Representative, has resigned to pursue a full-time position information about ASC2008 to be held in which Alan Branford has been doing – for details see the article by Doug Shaw. I Melbourne. Planning for the conference, superbly for the past few years. In future, a personally would like to thank Jane for her as well as for various satellite activities, is similar position for Section Chairs will be wonderful contributions to the professional well advanced and I look forward to a very established to assist in the communication running of the Society and wish her well for successful professional meeting. Planning between Sections and with the Executive. the future. has also commenced for ASC2010 in Perth. A Professional Officer Position Description This will be held in December, concurrently was discussed in readiness for planning William Dunsmuir with the Genstat Conference. for continuing professional development Email: [email protected]

When you hear something like “for Member News making statistics interesting, relevant and fun!” you might wonder “can this be true? Michael Martin Interesting and relevant, sure, but fun??” Professor Michael Martin of the Australian National University or even “it’s got to be some sales pitch!” recently received several awards from the Carrick Institute for And maybe it is a sales pitch, of sorts, Learning and Teaching in Australian Higher Education. In for what statistics as a profession might August, Michael was awarded a Carrick Citation for Outstanding do to attract more people to study it and choose it as a job. Contributions to Student Learning with a citation that read “For There is a (wrong) perception out there that statistics is either nearly two decades of making statistics interesting, relevant and frightening or dull (or both) and one of the main challenges in fun!”, and in November, Michael received a Carrick Award for teaching statistics is to overcome that perception and leave people Teaching Excellence in the discipline category Law, Economics, understanding that statistics is directly relevant to their lives, vital (in Business and Related Studies. the “alive” sense), and necessary in order to begin to understand our Carrick Citations are awarded annually to up to 210 people over increasingly complex world. all disciplines at Australian universities who “have made a significant This is my starting point for teaching statistics to undergraduates, contribution to the quality of student learning in a specific area of responsibility over a sustained period.” Carrick Awards for Teaching one that I try to combine with a sense of humour and a determination Excellence “celebrate a group of the nation’s most outstanding not to get bogged down in jargon. university teachers in their fields. The awards give recognition to It is an approach that has worked well for me over many years, teachers (individuals and teams) renowned for the excellence of and, more importantly, one that I hope has worked well for my their teaching, who have outstanding presentation skills and who students. I’ve also worked hard at developing lots of good analogies have made a broad and deep contribution to enhancing the quality for statistical thinking and techniques from real-life experiences. of learning and teaching in higher education.” (see http://www. These have really helped me show students that statistics is neither carrickinstitute.edu.au ) mystical nor magical, but rather a way of looking at the world that Michael reflected on his teaching and his Carrick Citation: unlocks (some of) its random secrets.

SSAI Newsletter – December 2007  Why Accreditation is Critical

The statistical profession in Australia when beginning study that it will lead bridge they use has been designed by a and around the world is under pressure to automatic recognition. A number of qualified engineer to be safe. They should from many directions. It sees statistical universities in Australia already have have the same expectation that decisions work being done by non-statisticians accreditation or are in the process of that affect their lives are based, where (often poorly) and new areas of research applying for it. appropriate, on the proper statistical such as data mining passing to other Accredited Statistician (AStat) status evaluation of the evidence carried out professions in a way that often prevents provides a higher level of accreditation by a properly qualified statistician. We statisticians being able to contribute. Ever and applies only to individuals. It requires are a long way from achieving that but it growing computing availability increases a combination of qualifications, ability should remain our goal. these pressures. and experience, and evidence must be I strongly urge all members of the Accreditation is a key element in the provided in the form of documented Society to support our accreditation statistical profession’s action against work (papers or reports) and referees. An system. Statisticians who do applied these trends. In Australia the system of Accredited Statistician takes responsibility work should seriously consider personal accreditation evolved by the Statistical for their work and can be trusted for the accreditation. Staff in universities should Society of Australia emphasises the quality of their advice. This is backed up consider having their courses accredited. outward face of statistics, providing by a requirement to uphold the Society’s Employers should recognise the value in standards so that those needed statistical Code of Conduct and the review of accredited applicants where appropriate. advice know where to turn and providing accreditation every five years. (We often overlook that some statisticians a structure useful to those who want to The Accreditation Committee of the are employers of other statisticians.) We develop a career in applied statistics. Society administers the accreditation should all publicise it more widely. Two levels of accreditation are system. While the Committee sees its role The Accreditation Committee is provided. Graduate Statistician (GStat) of encouraging the accreditation process currently updating the content of the status recognises the completion of an (and will often suggest improvements to Society’s website relevant to accreditation, undergraduate level course with sufficient applications if it feels that an applicant including fuller guidelines for both content and level in statistics, both theory has not done themselves justice), it individual and institutional applicants. and application. It marks the individual rigorously follows standards so that the We aim to make the processes and expectations clearer. Anyone interested in as being able to start a statistical career. terms GStat and AStat are seen as highly accreditation is also welcome to contact An applicant may ask for accreditation desirable and meaningful qualifications. me ([email protected]). on their individual merits, or through Much remains to be done. Ideally completing a course accredited by the the profession should achieve a full John Henstridge Society. The standards are the same but professional status as is held by architects Chair, Accreditation Committee, an accredited course provides the surety and engineers. The public expects that a Statistical Society of Australia.

Election of Executive Society Awards

Members The Society awards a gold medal, the Pitman Medal, at Members are advised that the Executive positions of Vice- most once annually, in recognition of outstanding achievement President (President Elect), Secretary and Treasurer will become in, and contribution to, the discipline of Statistics. Honorary vacant at the Society’s Central Council Annual General Meeting Life Membership honours outstanding contribution to the in 2008. profession and the Society, while a Society Service Award The SSAI Rules provide for a Nominating Committee, may be awarded to a Society member in recognition of consisting of the current Executive and the Branch Presidents, sustained and significant service to the Society. to solicit nominations and submit a list of nominees to Central An Awards Committee, chaired by the President of the Council. Should an election be required, Central Council will Society, makes recommendations to the Society’s Central then arrange a ballot of all financial members of the Society. Council as to appropriate Award recipients. Pitman Medals Members of SSAI are invited to submit nominations for the and Honorary Life Memberships are usually announced at three positions to be vacated. Nominations must be in writing the Society’s Conference. and signed by the nominator(s), and must be accompanied by Members of the Society are encouraged to propose a written and signed statement from the nominee accepting the suitable recipients of the Pitman Medal, Honorary Life nomination. Membership or a Society Service Award. Suggestions, Nominations should be submitted to the SSAI President with brief supporting information, should be emailed (William Dunsmuir) or to a Branch President before to the President, William Dunsmuir, as Chair of the 31st January, 2008. Awards Committee. Doug Shaw Doug Shaw Secretary Secretary

 SSAI Newsletter – December 2007 Three Doors with Borek Puza Society Secretaries Central Council (Edition 12) President: Prof William Dunsmuir Secretary: Dr Doug Shaw [email protected] Welcome to the 12th edition of width 2.8, and if =0.7 its expected width k Canberra Three Doors. Last time I presented The is 1.4. Like the alternative CI, the lower President: Dr Glenys Bishop Alternative Confidence Interval Puzzle CI exists for all possible values of y. Secretary: Dr Ray Lindsay and am now pleased to announce Terry Figures 1, 2 and 3 show the bounds, [email protected] Neeman as the next winner of the Three widths and expected widths of the above New South Wales Doors Prize, a cheque for $60 as donated CIs. It will be seen that the lower CI is President: Ms Caro Badcock by SSAI. The solution to that puzzle is uniformly the best of the three as regards Secretary: Dr Eric Beh given below, and the next puzzle follows. expected width (Figure 3) but not as [email protected] regards actual width (Figure 2). (Note: Queensland The Alternative Confidence In our calculations the expected width of President: Ross Darnell Interval Puzzle the ordinary CI was taken as 0 for y>4.5.) Secretary: Ms Helen Johnson Suppose that a number y is randomly chosen For further discussion of ‘alternative’ CIs [email protected] from the uniform distribution from 0 to k, - and examples of their usefulness - see the South Australian where k is an unknown real constant between following papers: President: Dr Andreas Kiermeier 0 and 5. Consider the statement Puza, B.D., and O’Neill, T.J. (2006). Secretary: Dr Paul Sutcliffe [email protected] 0.8 = P(0.2T(k) < y/k < 0.8 + 0.2T(k)), Generalised Clopper-Pearson confidence intervals for the binomial proportion. Victoria where T(k) is a suitable function of k President: Professor Mervyn Silvapulle or a constant. For the case T(k)=0.5, the Journal of Statistical Computation and Simulation, 76(6), 489-508. Secretary: Lucy Busija statement can be inverted to yield an 80% [email protected] confidence interval (CI) for k given by (y/0.9, Puza, B., and O’Neill, T. (2006). Interval Western Australian estimation via tail functions. The Canadian min(5,y/0.1)). Now suppose that T(k)=k/5. President: Dr Brenton Clarke Then the statement can be inverted to yield Journal of Statistics, 34(2), 299-310. Secretary: Ms Pamela McCaskie an alternative CI for k. Find the general Puza, B.D., and O’Neill, T.J. Optimal [email protected] form of the alternative CI and evaluate it constrained confidence estimation via tail for y=3.0. Why might the alternative CI be functions. To appear in The Mathematical Section Chairs considered preferable? Scientist in December 2008. Bayesian Statistics Kerrie Mengersen Solution to The Alternative [email protected] Confidence Interval Puzzle Biological Sciences With T(k)=k/5, the statement becomes Ari Verbyla [email protected] 0.8 = P(0.2k/5 < y/k < 0.8 + 0.2k/5), which, by solving y/k = 0.8 + 0.2k/5 and Environmental Statistics 0.2k/5 = y/k, leads to 0.8 = P(L < k < U), Petra Kuhnert – [email protected] where Industrial Statistics L=5sqrt(4+y)-10 and U=min(5,5sqrt(y)) Ross McVinish – [email protected] are the bounds of the required alternative Statistical Computing Kuldeep Kumar CI. Thus, when y=3.0 the alternative CI is [email protected] (L,U)=(3.23,5), whereas the ‘ordinary’ CI is (3/0.9,min(5,3/0.1))=(3.33,5). Statistical Education (co-chairs) Michael Martin An advantage of the alternative CI is [email protected] that it exists for all values of y. E.g., if y=4.7 Peter Howley then (L,U)=(4.75,5), whereas the ordinary [email protected] CI is empty (because 4.7/0.9=5.22>5). Surveys and Management Also, the alternative CI is shorter for many Veronica Rodriguez values of y. E.g., if y=0.7 it is (0.84,4.18) [email protected] with width 3.34, whereas the ordinary CI Young Statisticians (co-chairs) is (0.78,5) with width 4.22. Furthermore, Richard Hutchinson the alternative CI is shorter on average for [email protected] many values of k. For example, if k=0.7 The Two Uniforms Puzzle Pamela McCaskie its expected width is 2.36, whereas the Suppose that X and Y are independent [email protected] expected width of the ordinary CI is 2.83. random variables, and each is uniformly Social Sciences This discussion would not be complete distributed from 0 to 1. Find the density Michele Haynes – [email protected] without mention of the one-sided ‘lower’ function of U = X/(X-Y). Further contact details for Society 80% CI for k, defined by T(k)=1, namely For your chance to win a fabulous Secretaries and Section Chairs (y, min(5,5y)). E.g., when y=3.0 the lower mystery prize, please send your solution to can be obtained by contacting the CI is (3,5), when y=0.7 it is (0.7,3.5) with [email protected]. Society on (02) 6249 8266

SSAI Newsletter – December 2007  ANZJS Corner

Over the last year the ANZJS best mix of Associate Editors to cover A central, more long term question Editors (Kerrie Mengersen, Jeff Wood, the very wide range of statistics papers concerns Open Access, and whether Ken Russell and Steve Haslett) have submitted to ANZJS. We are working this is the best mode for future held a number of meetings, including toward focusing some journal issues publication of ANZJS. For those of the first face-to-face meeting of the around invited papers and special topics, you wanting to know more, there is current editorial panel in Sydney on with submitted papers on the same or a useful background article to which 24 May 2007. similar topics in the same issue. We are Murray Jorgensen provided a link. A number of issues were discussed considering how best to support the at length. editors and control workload through This is Jim Pitman’s IMS presidential address devoted to “Open Access to The Editors have been considering widening the group of people involved, content and direction for ANZJS. Of for example, through assistance, Professional Information”: http://www. necessity, this is driven in part by citation particularly for Kerrie and Ken. imstat.org/news/ims_pres_address07. indices, but we also recognise that there We were pleased to note that pdf The SSAI and NZSA executive could be more local content, both in Blackwell, the ANZJS publishers, committees plan to make sure there is terms of authorship and focus, without provide partial funding toward ANZJS extensive discussion on Open Access detriment to either the Theory and operational expenses, and once this and ANZJS at the next joint meeting of Methods or the Applications sections. is available to the Editors from the NZSA and SSAI, likely mid-2008. We have also discussed how to get two associations, there can be more more high quality applications papers We think putting a regular column operational flexibility than is possible submitted, and whether the criteria on ANZJS in the SSAI and NZSA at present. This could usefully include for acceptance of such papers needs more regular meetings of the editors. newsletters could be a possibility. That reconsideration. We are advertising way, we can outline what is happening for a new Book Review Editor. We We are considering whether a at ANZJS in more detail, and get more are looking medium term at a web- questionnaire to you, the journal’s timely feedback from you all. accessible system so authors and editors subscribers, would be a useful way to can better monitor submitted papers, canvas your views, and if so, what best Steve Haslett and at whether we currently have the to put in it. T&M Editor, ANZJS

ANZJS – Technical Editor ANZJS

The Australian & New Zealand Journal of Statistics is looking for an assistant Technical Editor. You will help the Technical Editor, Ken Russell, to – Book maintain a high standard of presentation and readability for the Journal. The work will allow you to get a good overview of what is included in Review ANZJS, before publication, and to understand the journal publication process more fully. You need to be good at spelling, grammar and Editor punctuation, and meticulous in carefully reading manuscripts from authors and Uncorrected Proofs from the publisher. You need to care that the format The Australian & New is correct. It helps if you have a sense for when something in a manuscript Zealand Journal of Statistics ‘seems wrong’ (for example, a set indexed by a variable that doesn’t appear needs a new Book Review in the definition of the set). You must have the time available to do the Editor. Responsibilities will work well, and be able to schedule your time when there are production include maintaining contact deadlines to be met. with publishers, finding Please contact Ken Russell ([email protected]) for further details. Please suitable reviewers, monitoring contact the Managing Editor, Kerrie Mengersen ([email protected]), reviews, and ensuring a steady if you are interested in being considered for the position. stream of high quality book reviews for each issue of ANZJS. If interested, please contact Prof Kerrie Mengersen, [email protected]. WORKSHOP au. There is no fixed closing Advanced regression models with R date. The vacancy will remain 13 February 2008, Melbourne open until filled by a suitable candidate. For further information: www.statsoc.org.au/whatsnew

 SSAI Newsletter – December 2007 Farewell Jane Waslin

The Society has regretfully said farewell to its Executive Officer, Jane Waslin, who has left us to take up a new challenge in the business world. Jane accepted the position of Executive Officer in early 2003, at a time when SSAI had taken a number of strategic decisions to increase the professionalism of the Society and the profile of the Statistics profession. That we have achieved as much of this strategic intent as we have in the subsequent years is due in no small part to Jane’s efforts. Jane has been responsible for the running of the SSAI Central Office, operating largely independently. During her time at SSAI, she has systematised and streamlined most of the office processes, resulting in efficient and well-understood procedures. As a consequence, the Central Office has been able to provide significantly improved assistance to the Branches, to the Society’s committees, to organisers of conferences and workshops, and to the individual members of the Society. Jane also ably represented the Society at meetings and at public events such as the Society’s conferences. All of this was done in a cheerful and helpful manner, which reflected Jane’s personality and her commitment to the success of the Society. We wish her ‘all the best’ for her new endeavours. Doug Shaw Secretary Spring Bayes Conference: Coolangatta, September 2007

In September 2007, Bayesian Topics discussed during the conference Statisticians made their way to obviously centred on Bayesian Statistics Coolangatta, Queensland for three days of but ranged from the highly theoretical, to presentations, workshops and just a little applied areas such as finance, ecology and bit of beach action at the Spring Bayes remote sensing. Delegates seemed pleased Conference. Delegates from Taiwan, to have the three different modes for New Zealand, Perth, Melbourne, Sydney, communication of their research. Further, Brisbane and numerous places in between the “hands-on” nature of the workshops all participated. Keynote speakers for the was well received. Does SSAI event were Adrian Barnett (University of Queensland), Cathy Chen (Feng Chia The next scheduled meeting for have your University, Taiwan) and Jean-Michel Bayesian Statisticians is the World Marin (INRIA, France). It was wonderful Conference of the International Society current to have such talented people sharing their for Bayesian Analysis on the 21 ¬ 25 July, research and we thank them very much. 2008 (ISBAA 2008). The conference is to be held on the beautiful Hamilton email address? The basic format for the event was oral presentations in the morning, workshops Island and please refer to the website in the afternoon and poster presentations http://www.isba2008.sci.qut.edu.au/ for If you are not sure please in the evening (accompanied by a tasty further information. We look forward to cold beverage). As conference numbers seeing your there. let us know by email to: are kept small, each Keynote Speaker was Kerrie Mengersen, Mark Griffin, Kate Lee, [email protected] able to give a workshop based on their Chris Oldmeadow and Matt Falk work to roughly a third of all delegates. Spring Bayes Organising Committee

SSAI Newsletter – December 2007  Counting Australia In

From page 3 Britain and elsewhere.” The government of Statistics was headed by Don McNeil. statisticians included Timothy Coghlan, The account on pp.291-294 of the and Trinity. The University of Adelaide in Sydney born-and–bred, of whom there Pollard family’s statistical careers, partly its search for the first Elder Professor of has also been some mention in the SSAI from Alf Pollard’s private autobiography, Mathematics, consulted Isaac Todhunter, Newsletter (No.94, 28 February, 2001), but makes fascinating reading. a Fellow of St. John’s; and the candidate much more in the Bicentennial History Room’s 4th year Honours class of 1936 chosen was the young Horace Lamb, Issue of the Australian Journal of Statistics included Pat Moran. Harry Mulhall, a Fellow of Trinity. Todhunter is well (30B, August 1988), where Chris Heyde with first class Honours in mathematics remembered by historians of probability states that Coghlan’s “work was hailed and chemistry and university medal in and statistics for his still widely used in the Journal of the Royal Statistical mathematics, had also been instructed by History of the Mathematical Theory of Society in 1898 as a pioneering work with both Carslaw and Room. He joined the Probability (1865); and by this reviewer no equivalent in Britain.” Sydney department in 1941, and took for his contact with, and probabilistic over from Sawkins the task of teaching influence on, the Oxford mathematician Statistical Science in the Universities statistics in 1946. His Ph.D. was from Charles Lutwidge Dodgson (“Lewis Horatio Scott Carslaw, Scottish- Cambridge, where mathematical statistics Carroll”, famed for his Alice books), on born and Fellow of Emmanuel College, as a discipline was evolving under John whose probabilistic work there have been Cambridge, succeeded Gurney at the Wishart. Mulhall transferred to the newly articles in the SSAI Newsletter. Dodgson’s University of Sydney in 1903. He retired created Department of Mathematical 175th birthday anniversary occurs this in 1935, but collaborated with John Statistics in 1959, and retired in the same year, and there have been accompanying Conrad Jaeger, a former undergraduate year, 1978, as Oliver Lancaster. In the first celebrations in England and Canada. on the first book in English on the full year of existence of that department, Lamb was Australia’s most notable Laplace transform, Operational Methods in 1960, the 4th year students included Chris mathematician of the 19th century. Applied Mathematics in 1941. Operational Heyde, Murray Aitkin and M.A.(David) The second Elder Professor, William research overlaps heavily in content with Hamdan. Chris Heyde’s illustrious career H. Bragg, was Australia’s most noted applied probability in which non-negative within Australia has encompassed the physicist of the early 20th. Associated random variables are of primary interest. CSIRO, and accession to Chairs at the with Trinity and selected on the advice My own introduction to probability was University of Melbourne and the A.N.U. of J.J. Thomson, William and his son through Ren Potts who considered himself In regard to statistics at the University Lawrence (born 1890 in Adelaide), won an operational researcher, with the study of Sydney in Chapter 6, which completes the Nobel Prize for Physics in 1911. (The of operational methods encompassed the story of the six state universities, Braggs had left Adelaide in1908.) William by courses in both Pure and Applied one in each capital city, the career and Bragg was a contemporary at Trinity Mathematics at Adelaide University. activities of Oliver Lancaster naturally of the Australian-born and educated First class Honours graduates from dominate. A meeting on 25 September W. F. Sheppard, still remembered for the University of Sydney in the first half 1947 between Lancaster, Stewart “Sheppard’s corrections” in calculating of the 20th century included not only Rutherford and Helen Newton Turner moment estimates from grouped data. later statistically important figures such as was the genesis of the formation of the There was an Australian applicant for the Weatherburn and Belz, but also a name Statistical Society of New South Wales, second Elder Chair, William Sutherland prominent in Australian political history, the predecessor of the Statistical Society (1859-1911) who discovered the diffusion Herbert Vere Evatt, who also obtained of Australia. Rutherford would go on to equation which Einstein developed. first class Honours in English, and was become Professor of Economic Statistics Sutherland’s results were presented to the Leader of the Labor opposition in the era from 1962 to 1980. Other notable meeting of the Australasian Association of Menzies’ Government. names associated with the early period for the Advancement of Science held in Thomas Gerald Room, a Fellow of St. of statistics at the university are Ann Dunedin, N.Z., in 1904. Bragg gave the John’s from 1925, arrived in August 1935 Eyland, Geoff Eagleson, and David B. seminal address, to Section A, of this to succeed Carslaw. Room introduced a Duncan, who had taught Biometry since meeting. separate 4th year Honours in mathematics. 1938, and had returned by 1947 with a The author writes in his brief, but as The sole recipient of 1st class Honours Ph.D. from Iowa State University, one of for all chapters, characteristically incisive, in 1935 was Alf Pollard, who was to the great centres of statistics development lead-in to Chapter 3: “There were four become one of the most colourful in the U.S. The mathematics Honours universities in Australia by the end of and influential figures in Australian class of 1958 included John A. Hartigan. the 19th century and in the University statistics, with career concluded by The Departments of Pure Mathematics, of Adelaide, if not in the others, world retirement from his position as Professor Applied Mathematics and Mathematical class research in mathematics was being of Economic Statistics at Macquarie Statistics were incorporated into a School achieved. In all four, mathematics was University. When Pollard retired, his of Mathematics and Statistics from 1st being taught at a level comparable with department, incorporating actuarial January 1990, to achieve, it was said, British universities. … Government studies, demography and statistics was economies of scale. The Department statisticians in the eastern states in split into two. The new Department of of Economic Statistics, under Aln particular were also setting standards Actuarial Studies was headed by his son Woodland, eventually underwent a name that would be viewed with admiration in John Pollard, while the new Department change to Department of Econometrics,

10 SSAI Newsletter – December 2007 and then was incorporated into a larger H.M. Searle and A.W. Jones: Leaving years. Australian-born and educated, Economics framework. Mathematics (Revised) with Answers. It unusually for the times he studied in New Maurice Belz, born and educated in has a preface by Wilton, who died of a York, gaining a Ph.D. from Columbia Sydney, gained a B.Sc. from Sydney in stroke in 1944.) Harold W. Sanders, an University. He introduced new courses in 1918. He won a University Medal in applied mathematician who had been at statistics and actuarial mathematics; his mathematics, and a few years later was Adelaide University since 1923 replaced research output demonstrated a growing awarded a Barker graduate scholarship Wilton, who had been ill for a decade, interest in statistics. He retired in 1955. which saw him enter Gonville and Caius as Elder Professor in 1944. A boost for Later names of importance to Australian College, Cambridge. With an M.Sc. from Adelaide mathematics, and relief for its statistical history include those of Henry there he was appointed to the University hard-pressed small mathematical staff, Finucan, and Stephen Lipton, the of Melbourne in 1923, and in 1929 he came with the appointment of Hans foundation Professor of Statistics from introduced a course in the theory of Schwerdtfeger in 1940. Though not a 1967 to 1991. statistics. As Associate Professor there, Jew, he had fled Naziism, and came to The University of Western Australia Belz was appointed Head of the first Australia through the intercession of Sir was founded in 1913. Weatherburn, autonomous Statistics Department in William Bragg. Schwerdtfeger “left a was the second Professor there. Born in Australia in 1948. He was Professor of remarkable legacy in Adelaide. He had Chippendale, Sydney, and educated at Statistics from 1955 to 1963, the first to enticed Tim Wall away from medicine, Sydney Boys High and the University hold such a position in Australia. The Ren Potts away from engineering and of Sydney, he was a B.A. with first Australian statistician, Herbert A. David, Alan James … away from physics”. class Honours from Trinity College, born in Berlin in 1925, and as a postwar Alan James succeeded Alf Cornish in Cambridge. He retired in 1950. Other student at the University of Sydney, a the Chair of Statistics, in 1965 and retired statistical names associated with this contemporary of Oliver Lancaster (after in 1989. South Australian-born, his return university are those of Joe Gani, Uma Lancaster’s military service as pathologist from the U.S., as that of Rainer Radok, Prabhu and from 1974, Terry Speed. in New Guinea), was appointed Senior was due to Ren Potts. James’ Ph.D., Some temporally intermediate names Lecturer in Belz’s department in 1955. from Princeton, had been supervised by are missing from the author’s account, He left in 1957 to pursue an illustrious S.S. Wilks, a pioneer of mathematical specifically Don McNeil. career in the U.S., and is still active in statistics in that country. There is a feisty Statistics at the University of Tasmania, the history of statistics. Other staff at the “Valedictory Address” in 3 parts by Alan in addition to Pitman, includes in its University of Melbourne during Belz’s James in the SSAI Newsletter (No.51, 31 development Peter Sprent, Don McNeil, time included Joe Gani, Geof Watson, May, 1990; No.52, 31 August 1990; No. Michael Hasofer, Peter Sprent, and Rupert Leslie, Emeric Binet, Alison 53, 30 November, 1990). L.S. Goddard. Doig (who later worked with M. G. Due to Schwerdtfeger’s influence, Chapter 6 also describes mathematics Kendall on statistical bibliography), Betty Sanders was able to argue successfully in what were to be called The University Laby (apparently not mentioned in the that he himself be replaced (as happened of New South Wales and the University book), Daryl Daley, Peter Finch, Stephan in 1948) by a Hungarian Jewish refugee of New England. With the Australian Maritz, Warren Ewens, and C.R. (Chip) living in Shanghai, George Szekeres, who National University, there were nine Heathcote. Both of the last two were had enormous influence on Australian universities and two university colleges (in in Ted Hannan’s Statistics Department mathematics, and on this reviewer Newcastle and Wollongong) before the when I arrived there. personally as a fellow refugee. In a changes brought about by the report of After retirement, Maurice Belz was mathematical sense, the influence came the Murray Committee in 1957. Chapter replaced as Professor of Statistics in 1964, later, through Szekeres’ work on iteration, 7 includes a treatment of the Australian by Evan Williams, who took the position when I was working on the simple National University. Important statistical in preference to an offer of a personal branching process, which is essentially a sources for the author of this book were Chair at the University of Tasmania, his problem in iteration in a stochastic model A Brief History of the Department of alma mater. The staff members which framework. This wiry man of modest Statistics. The University of New South Williams inherited from Belz were Finch, physical stature bestrides the latter parts Wales. 1948-1983. by J.B. (Jim) Douglas later Professor of Mathematical Statistics of this book. George and his wife Esther which appeared in 1996; and Joe Gani’s at Monash University, and Maritz, who both died in Adelaide within hours of Fifty Years of Statistics at the Australian was to hold Chairs at both Monash and each other on 28th August, 2005. In National University, 1952-2002, published Latrobe. Williams retired in 1982. Sydney they had lived in Turramurra, in HRAS in 2005. (A history of statistics On Bragg’s suggestion John R. Wilton, and liked to have coffee at a local pastry at the University of Sydney is envisaged who had spent his early years at Mount shop where I occasionally met with for 2009, the 50th anniversary of the Barker, near Adelaide, after his Honours them. George collaborated in some work creation its Department of Mathematical degree in mathematics and Physics there, in probability with Emeric Binet; and Statistics.) entered Trinity College, Cambridge. perhaps with the Szekeres’ friends from The dominant influence in the After a lectureship in mathematics at their youth, Alfred Renyi and Paul Erdös, development of statistics, is in British Manchester under Horace Lamb, within a two great Hungarian names of probability, sense of including applied probability, year he was offered the Chair in Adelaide. who came to Australia on visits. in Australia was the Department of He reorganized the mathematics courses, The second Professor of Mathematics Statistics in the Research School of Social and was very influential in changes to and Physics at the University of Sciences, ANU, at which Pat Moran took school mathematics in South Australia. Queensland (founded in 1912), Eugene up the Chair in 1951. In what was an (I still have my copy of the 1957 version Francis Simonds, appointed in 1932, had (it was first published in 1941) of already been in the department for some continued on page 12

SSAI Newsletter – December 2007 11 Counting Australia In

From page 11 published in J. Business in 1990. Dilip, and the papers originating from Invited at the time a member of the Economic Paper Meeting 75: “Statistics in Australia Australian tradition by the time, Moran Statistics/Econometrics Department, has Growth and Influence” on Thursday 7th and Freddie Chong (later Professor of subsequently gone onto great things at April 2005 of the Sydney Session of the Mathematics at Macquarie University) the University of Maryland at College International Statistical Institute which who had known each other at Sydney Park, and as consultant to many financial are available on the post-session CD of University, had both entered St. John’s institutions, in the U.S. and Europe, presentations produced by the ABS. The College, Cambridge. One of the effects including Morgan Stanley on Wall papers particularly relevant to Chapter which the newly established research-only Street. are: Ian Castles (formerly Commonwealth ANU was intended to have, was to train Another interesting area, Statistician): “Official Statistics: From and retain in Australia, Ph.D. candidates. conservationism, is treated very briefly on Blue Books to White Papers”; and Murray In regard to statistics, the plan succeeded p.286, in the context of Flinders University, Cameron (Chief, CSIRO Mathematical brilliantly. Eminent overseas and local and includes the names of Keith Tognetti and Information Sciences): “Statistical staff were attracted to the department, as in Wollongong, Shirley Strickland de la Research for Science and Industry.” well as students not only from Australia Hunty (recently deceased) in Perth, and The following criticisms can be easily and New Zealand. By the 1980’s most of most remarkably, John Walmsley. There eliminated with any updating of this the Statistics Chairs in Australia had been is more to tell of Walmsley’s interaction important book. filled by ANU’s Ph.D. graduates. The at Flinders with Rainer Radok, and of story is well told in Joe Gani’s article, and Radok’s interesting career in Australia. Ease of use of the citation system seems is well told in Cohen’s book. to me of paramount importance in such Certain important topics for a reference book. In this one citations are Joe Gani has been activist in the mathematics as broadly defined in this by sequential number of endnote. The development, status and health of book are close to the author’s heart, and Australian mathematics since his early are particularly well treated. The Canberra numbering begins anew for each chapter. years in this country, and the author’s CAE/University of Canberra is used as a At the end of the book the endnotes Preface lists him among those who platform for the Australian Mathematics are listed under chapter number, but the provided exceptional assistance. Since Trust and the Australian Mathematics text pages on which the numbers occur Gani’s return to Australia in 1974 (to Competition. The Australian Academy have at the bottom of each page only the the CSIRO) and in 1994, he has been of Science is used as background for chapter title. Thus one needs to go to the enormously influential in Australian the 120 page report dated January 1996: table of contents (or to flip back) to the statistics, and in his formal retirement Mathematical Sciences: Adding to Australia. chapter’s beginning, to find the chapter remains active and energetic at ANU. number, before being able to access the This report was motivated largely by With the great increase in the number the parlous state of mathematical sciences, endnotes. of universities and appointments from particularly in the newer universities such There is no bibliography: that is, no 1960, emphasis in the book relating to as Wollongong, Newcastle and Monash, overall alphabetical listing by author of this period is on Professors and dominant resulting from the announcement in all the items cited in the book, so it personalities. The author in this part December 1987, which ended the binary is difficult to see, for a given author/ of the book also tends to emphasis on system of universities and Colleges of editor, which items have been used. Of Australian activity and qualifications in Advanced Education (the “Dawkins particular interest to this reviewer was U.S. universities. There is consequently reforms”), and the introduction of citation of writings of historical kind of some unevenness of treatment of HECS. This is described in Chapter 9. H.O. Lancaster, J.M. Gani, I.S. Turner, statisticians, for example Iain Johnstone By 1999 there were, consequently, 37 R. B. Potts and C. Forster. However, for and Barry Quinn, classmates at ANU. public universities across Australia, all example, it’s somewhat difficult to see if competing for government funds for both Gani’s The Making of Statisticians is cited. Recent Topics and teaching and research. The number of The Name Index could have been more Concluding Remarks Professors of Mathematics, including comprehensive. For example, apart from mathematical statistics, at Monash peaked The book’s author uses the Mathematics names which occur in the text (such as at nine in the late 1970’s. At the end Department of the University of New A.G. Pakes on p.201) but are missing of 2005 in its School of Mathematical South Wales as platform for his main completely from the index, the names of Sciences there were five. At the CAE’s discussion of financial mathematics in J.N. Darroch and D.R. McNeil occur on a concluding section of Chapter 6. The which had become universities with the pages (resp. p. 297 and p.301) additional beginnings of this largely probability- end of the binary system, the decline to those listed in this index. based subject area in Sydney were of mathematics, included in Chapter 7, probably in joint work stretching over though on a smaller scale, is no less However, I leave this book with the 1980’s, at the University of Sydney dramatic. profound admiration for both its author between Dilip Madan and the reviewer, In regard to statistical sciences, to the and his creation, and encourage its wide whose collaboration culminated in the sources cited by the author one might dissemination among individuals as well foundation paper on the now widely add the following: P. Groenewegen as libraries. cited and used Variance-Gamma Model and B. McFarlane (1990): A History of Eugene Seneta for the movement of returns, which was Australian Economic Thought (pp.92-117).; University of Sydney

12 SSAI Newsletter – December 2007 Branch Reports

Queensland Branch News president Brenton Clarke; a former honours Australia, presented a talk on Statistical student of Brenton’s at Murdoch University, Consulting – Beyond a Statistical Toolkit at Peter McKinnon; and Geoff Riley of Alcoa the October meeting of the WA Branch. At the Branch’s October ordinary general World Alumina. John had recently presented this talk at meeting, Tim Robinson from the University In order to estimate the size, location the International Symposium on Business of Wyoming presented a seminar titled and grade of bauxite deposits, Alcoa must and Industrial Statistics in the Azores, and, “Bayesian Analysis of Split-Plot Experiments process approximately 500,000 drill samples indeed, he commenced his presentation to with Non-Normal Responses for Evaluating annually. The solids analysis technique of the WA Branch by recommending to us Non-Standard Performance Criteria.” infrared spectroscopy is used along with fast that if we ever got the opportunity to attend Tim began his presentation by explaining Fourier transform spectral analysis to extract and present a paper at a conference in the an example of the use of split-plot chemical information from the samples. Azores, that we should definitely take up experiments in the production of light- The spectra are approximated using the offer! polarizing film used in the manufacture partial least squares or principal components John began his talk by describing how a of items such as sunglasses. The inputs regression and the fitted models are typical statistical consulting model would to the process that could be controlled assessed using the representation indicators not be viable for a fully commercial company were chemical components or mix and Mahalanobis distance and residual ratio. such as Data Analysis Australia. The role process settings. The response is the quality These representation indicators have an of the statistician in the typical model is of polarized light that passes through the approximate gamma distribution and robust limited, with their main input being only film. The distribution of the response methods of estimating the shape and scale at the start (formalising hypotheses and exhibited a constant coefficient of variation parameters of this distribution are required. defining data requirements) and the end suggesting a gamma distributed variable. The study evaluated several estimation (analysing data in terms of hypotheses and For each combination of input mixtures, methods: maximum likelihood, method of discussing the results) of projects or research the process variables were altered according moments, L2 minimum distance, Cramér with limited involvement in any other areas. n to a 2 factorial design, hence the use of the von Mises distance, and Hampel’s B- In reality, the commercial statistician’s term “split-plot” to describe the design. The optimal. Robust methods such as the B- role is much larger than this. In a group of whole plot stratum corresponded to a “roll” optimal estimator perform well in practice, professionals, the statistician is at the very of film and its precision was modelled as a particularly in the presence of outliers, least an equal and often the appropriate random variable. however they are difficult to compute since leader for the project or research, rather Tim then used a Bayesian approach their “tuning parameters” require knowledge than as just the technical advisor in a limited by defining a prior distribution for the of the underlying parameter. Neither the capacity. In this role, the statistician is parameters that controlled the quality of the method of maximum likelihood nor the involved in helping to define the client’s film. The “whole plot” effect distribution method of moments are robust for the problem, reviewing possible new and was defined as normal and the response gamma parametric family since they have existing data, managing the data, analysis was distributed as gamma conditional on unbounded influence functions. On the and presentation of the results. Typically, the whole plot effect. Tim proceeded to fit other hand the two estimators, the L2 these results must be taken beyond the the model using WinBUGS comparing minimum distance and the Cramér von statistical results and formulated into real the output generated with that from a Mises distance estimator, are robust in terms actions and recommendations for the client. PQL approach. The two sets of parameter of having bounded influence functions John summarised the differences between estimates were similar. and perform reasonably well in terms of scientific and commercial consulting models The analysis became more complex when efficiency. Asymptotic theory associated in the following way – scientific consulting the response changed to the number of film with these two estimators is justified using is typically more conservative, based on Fréchet differentiability. The Cramér von pieces in a roll that exceeded a predefined existing hypotheses, is often less time and Mises distance estimator is the most readily stringent specification. The benefit of a cost critical and the statistician’s role is clearly computable, which is important where fast Bayesian approach was then realized when understood, whereas commercial consulting throughput methods are needed. Tim could generate point estimates and tends to be more optimistic, with innovative credible intervals for this parameter. The question was asked about whether approaches favoured, is extremely cost and multivariate methods could be used to The speaker and members adjourned to a time critical and the statistician’s role is identify outliers prior to the summary data local restaurant for further discussions. much larger and less defined. In reality, involving variables with approximate gamma some elements of each are often applicable distributions. It was pointed out that usually Western Australia to most cases, with some problems being identification of outliers in multivariate data closer to one model than the other. Branch News involved computer intensive algorithms The implication of this is that statisticians and this was not always amenable in an in commercial consulting need more than just automated process. August Meeting good statistical skills. Having the statistical The talk presented at the August meeting Alex Stuckey skills is necessary, but not sufficient, to be a of the WA branch was titled “Robust good statistical consultant. The real skill lies Methods for Fitting Gamma Distributions October Meeting in choosing the best tools to fit the problem, with an Application to Spectral Goodness John Henstridge, Managing Director and in order to choose the best tool, you of Fit Data, For Use in Chemometric and Principal Consultant Statistician of need to know the tools – what tools exist, Measurement Quality Assurance”. The Australia’s largest fully commercial statistical what they might do, and perhaps, how to speakers were: the SSAI WA branch and mathematical consultancy, Data Analysis use them. It is unlikely that there will be

SSAI Newsletter – December 2007 13 Branch Reports Continued

any tool that is exactly right, but knowing and extending the work to detection of enjoying her current roles as a data analyst the tools allows a choice to be made as to multiple ‘clumped’ outliers. The moderator and statistician. what is closest to right. What is needed for and the audience then thanked the speaker After the talks, audiences were invited this is a broad appreciation of statistical and for sharing her interesting and important to a free Turkish pizza dinner. Exchange mathematical principles, rather than simply work. The evening was finished with of experience and networking among all the application of the details. Typically, this members of the audience joining the present continued during the dinner. The requires more knowledge, not less. speaker for a delicious dinner at an Indian branch would like to thank Centre for Anna Munday restaurant nearby. Mental Health Research and School of Dr. Alison Smith holds a PhD degree in Finance and Applied Statistics, ANU for Canberra Branch News Statistics from the University of Adelaide. their financial contribution to the event and In fact, as she told the audience, she holds Kevin Wang for organizing the event. August Meeting the distinction of being the last student Angus Salim to be conferred a PhD degree before the The speaker at the Canberra Branch Statistics department was amalgamated. Young Statisticians’ Events August meeting was Dr. Alison Smith from She can be contacted at alison.smith@dpi. in Canberra NSW Department of Primary Industries. nsw.gov.au. This year has seen a considerable amount On the occasion, Alison talked about her of young statisticians’ activity here in proposed approaches for detecting outliers the National Capital. Besides the Young in linear mixed models. She started her talk Canberra September Young by giving a brief overview of mixed models. Statistician Night Statisticians’ Conference held in April this year, we had a Young Statisticians’ Dinner She then presented three practical examples On September 25, the Canberra branch in August and a Young Statisticians’ Night from multi-phase trials of wheat varieties, a organized a special meeting for Young in September. microarray experiment and nicotine content Statisticians. The meeting was an avenue in tobacco, where the presence of outliers is for young statisticians in Canberra and The Young Statisticians’ Dinner was suspected even after clustering effects and nearby regions to meet each other and kindly sponsored by the Canberra Branch the effects of other covariates are accounted exchange ideas and networking. The event and was held Thursday 16 August at UniPub for using linear mixed models. Unlike was very successful and 46 people came to in Civic (downtown Canberra). There were ordinary linear models, where the existence the meeting. On the occasion, we are very 14 attendees, most of whom were from the of an outlier can be easily checked by fortunate to have two promising young ANU. UniPub’s pleasant atmosphere and looking at the residual plot, outlier detection statisticians speak about their young career “cook-your-own-steak” set-up (an interesting in linear mixed models is more complicated. and achievement. Emma Lawrence from the concept to some of us!) proved conducive to There are two factors that complicate outlier Bureau of Rural Sciences (BRS) talked about social mingling as everyone moved around detection in linear mixed models, namely her recent work which shows how statistics to meet new people and chat. the existence of multiple random terms is applied to help answering questions of We would like to thank Professor rather than a single random (error) term high importance to public policy makers. Michael Martin and Mr Ian McDermid and the possibility of correlated rather than She highlighted her experience in using for allowing us to talk to their classes independent effects. She proposed to tackle environmental data to predict the potential and for promoting SSAI on WebCT (the this difficulty by extending outlier models distribution of marine pests such as starfish ANU web-based learning system). Going discussed by Cook, Holschuh and Weisberg in different locations around the country to the classes and speaking to the students (1982) for ordinary linear models, where should an invasion event by these species proved to be fruitful in promoting the outliers are assumed to arise from an error occur. Another highlight was her project on Society and the dinner - most of people term with inflated variance. In order to predicting the effect of over fishing on the who came along said something along the detect an outlier a score test based on the fit stock of commercial fish such as blue tuna. lines of, ‘I heard about the Society and of the null (no outlier) model is proposed. dinner from the talk you gave us at the The second speaker was Jessica Zhang The distribution of the score test statistic beginning of the semester!’ is approximated using a ‘quick’ resampling from Australian Institute of Health and The Young Statisticians’ Night was held scheme or full bootstrap approach. In terms Welfare (AIHW). Jessica shared her long, on Tuesday 25 September. It was held as the of yielding the correct level of Type I error, winding story in becoming a statistician. regular Canberra Branch meeting. With the ‘quick’ resampling method seems to be From interest in mathematics during her 40 people attending, there is no doubt that reliable, except when sample size is small secondary school years she did a degree in this was one of the most successful young in which case full bootstrap approach is Economics and as a top student was offered statisticians’ events held by the Canberra preferred. The applicability of the score a place at an American university to do a Branch. Roughly 2/3 of the attendees were tests was demonstrated by applying them Masters degree. But her failure to obtain a young statisticians and the majority were to the illustrative datasets. Some data US visa and the fact that her husband got a fresh faces. points were identified as outliers and many job in Australia led her to Canberra. Once were subsequently found to correspond to in Australia, she decided to do a Masters Emma Lawrence and Jessica Zhang were erroneous data. After the talk, there was degree in Applied Statistics at the ANU the two young statisticians invited to give a lot of discussion between the speaker because she heard about the shortage of talks. Both speakers gave very interesting and audiences on comparing her approach statisticians in Australia! Upon graduation talks that were well received. Agus Salim’s to potential alternative approaches such as from the ANU she was offered a job with column provides details of their talks. modeling the outliers using distribution AIHW. She said that in statistics she has We would like to give special thanks to with heavier tail than Normal distribution finally found her real interest and she is the Centre for Mental Health Research

14 SSAI Newsletter – December 2007 Branch Reports Continued

and the School of Finance and Applied The South Australian Branch held Other tidbits of advice that the speakers Statistics at the ANU for sponsoring us another successful careers evening for Young imparted to the audience were: a wonderful Turkish Dinner after the Statisticians in September. The concept of a • Keep your uni notes, they may come in talks, and Burgmann College (ANU) for careers evening for students and those early handy! providing us the venue. Indeed, the food in their statistical career was initiated by our • Networking amongst statisticians is was so good that there were sightings of Young Stats reps, Lisa Yelland and Penny invaluable random College residents popping into Bennett, last year which generated a record the room and nicking off with food! The attendance. • Do not underestimate your numerical collegiate environment was very fitting for We had four speakers from different skills and how they can assist others the atmosphere of the evening and the backgrounds to give a broad perspective • Honours and PhDs are worthwhile, they stand-up dinner format allowed everyone on the potential careers for a statistician, are an apprenticeship in research to mix readily rather than being confined particularly focused in SA. Our four • Consider what drives you, what interests to a seat. Senior statisticians shared their speakers were: Ewa Seidel, Methodologist, you experiences in informal chats and were ABS; Jonathan Tuke, PhD student and • Do not compare yourself to others. invaluable to the young statisticians. Lecturer at The University of Adelaide; The night finished off with the speakers Kevin Wang Penny Bennett, PhD student and Statistician at Telstra; and Lynne Giles, Senior Research and attendees joining for some great pizza ACT Young Statisticians’ Representative Fellow, Department of Rehabilitation and and plenty of discussions about career Richard Hutchinson Aged Care, Flinders University. opportunities in statistics. The SA Branch Co-Chair Young Statisticians Section Council would like to thank Lisa Yelland To summarise, the speakers noted some and all her helpers for organising the event outh ustralia ranch ews different aspects of working in different S A B N and to thank all the speakers for their great types of organisations which are worth wisdom and enthusiasm, which made it a considering when thinking about taking a The use of High-Density Genetic great evening. position: Markers to Detect Genes Associated Janine Jones with a Disease • flexible working conditions • freedom to research statistical topics for Huwaida Rabie, a Post Doctoral Fellow Cornish Lecture – Statistical Critique your own interest (academia might be for working with Dr. Ian Saunders at CSIRO, of the International Panel for Climate you) spoke at the September meeting about Change’s work on Climate Change • professional development: do they provide her current project which aims to develop The SA branch inaugurated a series methods for association and variable in house training? Do they support going to conferences? of public lectures on statistical topics of selection on SNPs data. broad interest in 2001. The lecture series SNPs or single nucleotide polymorphisms • Will you be working with other has been named to commemorate Alf are DNA sequence variations that occur statisticians? If not, maybe you will need Cornish, a leading figure in the early years when a single nucleotide within the gene to “network” with other statisticians of the statistical profession in Adelaide. The is altered. These could be used to locate outside your workplace to bounce ideas lectures are held biennially and presented genes associated with a disease of interest. • Good written and verbal communication by eminent statisticians from around the Only a small fraction of SNPs alter gene skills are essential for communicating world. The speaker at the 2007 Cornish function or expression. The aim is to look statistical concepts to non-statistical lecture was former Australian Statistician, for SNPs close to the disease susceptibility colleagues/clients Mr Dennis Trewin. As a current member gene using an association study which • Diversity of statistical projects: consulting of the Australian State of the Environment considers the population and compares provides this diversity - you get to play in Committee, Dennis was invited to give cases with controls. This differs from a everyone else’s backyard for a while. Be a a statistician’s perspective on recent linkage study which involves families. The medical scientist one day, a horticulturalist climate change projections made by the issue of multiple testing is a huge problem the next International Panel for Climate Change in association studies where a large number of SNPs are considered. Huwaida presented findings from a simulation study to compare methods of identifying SNPs associated with the disease susceptibility gene. The example data set related to leukaemia and involved consideration of both genotype and haplotype based data. The purpose of the simulations was to examine the accuracy of predictions and the ability to select the correct model. No evidence of an association between SNPs and disease was observed in the data. Speakers at YS careers night (L-R): Lynne Giles, Jonathan Tuke, Ewa Seidel, Lisa Yelland Penny Bennett; and organiser Lisa Yelland.

SSAI Newsletter – December 2007 15 Branch Reports Continued

(IPCC). His presentation was based on during his very short visit and in order “publication bias”. The main thrust of his an invited paper presented at the OECD to keep his commitments manageable the presentation was to go on and discuss the World Forum. Around 100 SSAI members NSW Branch worked with the Australian recent new evidence that demonstrates that and other interested parties attended the Pharmaceutical Biostatistics Group the selective reporting of trial outcomes lecture – a record attendance for an SA (APBG) to hold a joint meeting. As a within published studies is an additional branch event! result of this collaboration the meeting was threat to validity. Doug called this “reporting Dennis believes that climate change is an held in the very pleasant surroundings of the bias”. Doug showed results of research where important issue which needs to be addressed. Pfizer auditorium, a venue that also allowed they were able to obtain whole protocols and In his talk, he recognised the importance of us to video conference in members of the publications and found that the trial reports the work conducted by the IPCC and Queensland Branch. On behalf of the could include preferentially those outcomes congratulated them for the transparency of SSAI I would like to give a very big thanks with statistically significant results and, their work which includes projections of to Philippa Clark, Deputy Chair of the in addition, analyses in publications may future climate change based on a range of APBG and Senior Statistician for Pfizer’s have differed in important ways from those scenarios. This transparency however, has health outcomes group for suggesting and specified in the trial protocol. For example, allowed people to criticise their work and organising the venue. I would also like to the primary endpoint in the protocol is a major concern is the lack of involvement take this opportunity to thank all the staff at no longer the primary endpoint in the of statistical experts. According to Dennis, Pfizer involved in the evening, including the publication. The research also identified there are statistical flaws in the climate security staff at the front gate who let us in something that we may all have come across change projections which could lead to and helped us park, the IT staff who assisted when reading publications but probably not overestimates of global warming. with the video conference link, Annabel thought that much about – the analyses Peeling, Pfizer Biometrics who collated all specified in the methods section are not the The IPCC have considered six basic the registrations so we could pass through ones presented in the results section! This is scenarios in their climate change projections, security and the catering staff within the obviously of concern but is possibly easier to depending on technology, population function area of Pfizer. remedy through the review process than the growth and use of fossil fuels. The media selective reporting of endpoints. typically report on scenarios resulting in The meeting was held on September 20. higher estimates of climate change. Dennis Doug shared his opinions and insight to Doug explained the Cochrane pointed out that not all scenarios are equally the selective non-reporting of findings of Collaborations new “risk of bias” tool for likely and voiced his concerns about some randomised clinical trials and the subsequent identifying selective outcome reporting and of the assumptions made in the various impact on systematic reviews that are acknowledged that it couldn’t provide the scenarios considered by the IPCC, such as commonly undertaken in the medical field. total solution. He also provided some, absolute convergence of economies. Seventy seven people attended the Sydney what he called, partial solutions such as monitoring and regulation via ethics Addressing climate change requires venue (16 from Pfizer and the rest being committees, data monitoring committees resources which need to be used efficiently from a number of institutions including and funders, registration and publication of and thus, it is important for policy decisions pharmaceutical companies, academia and protocols, replication and availability of the to be made based on appropriate estimates government). Approximately 50% of the raw data - each of these having their own of climate change. Dennis would advise Sydney attendees were statisticians but challenges not least of which is who is going policy makers that policies should be all had an interest in data and obviously to fund and manage the solution(s)! focused on the more likely climate change amalgamating data in order to assist decision scenarios presented by the IPCC, however it making. Another dozen or so joined us via After Doug’s very thought provoking talk should be recognised that much uncertainty video conference from Brisbane. about 20 attendees joined Doug at a local remains and thus, an adaptive approach may Doug started by reminding us that non- Chinese restaurant to continue discussions. be best. publication of the findings of some trials I think that all in all it was a very successful evening and thanks again to everyone who At the end of the lecture, Dennis was has been recognised as a potential threat to attended and assisted. faced with many tough questions from the validity of meta-analysis and reviewed members of the audience who were clearly the evidence of bias from this activity. This Caro Badcock passionate about the issue of climate change. practice is usually, and paradoxically, called President NSW Branch SSAI He then enjoyed a well-deserved dinner at Jahz Café with around 20 society members and invited guests. Lisa Yelland

New South Wales Branch News

Doug Altman who is currently head of the Centre for Statistics in Medicine, University of Oxford recently visited Australia as an invited speaker to the third International Clinical Trials Symposium. Doug kindly Caro and Philippa waiting to catch agreed to speak to the Statistical Society Doug beginning his presentation. Doug’s attention!

16 SSAI Newsletter – December 2007