ALEXANDER PHILIP DAWID Emeritus Professor of Statistics University of Cambridge E-Mail: [email protected] Web
Total Page:16
File Type:pdf, Size:1020Kb
Load more
Recommended publications
-
September 2010
THE ISBA BULLETIN Vol. 17 No. 3 September 2010 The official bulletin of the International Society for Bayesian Analysis AMESSAGE FROM THE membership renewal (there will be special provi- PRESIDENT sions for members who hold multi-year member- ships). Peter Muller¨ The Bayesian Nonparametrics Section (IS- ISBA President, 2010 BA/BNP) is already up and running, and plan- [email protected] ning the 2011 BNP workshop. Please see the News from the World section in this Bulletin First some sad news. In August we lost two and our homepage (select “business” and “mee- big Bayesians. Julian Besag passed away on Au- tings”). gust 6, and Arnold Zellner passed away on Au- gust 11. Arnold was one of the founding ISBA ISBA/SBSS Educational Initiative. Jointly presidents and was instrumental to get Bayesian with ASA/SBSS (American Statistical Associa- started. Obituaries on this issue of the Analysis tion, Section on Bayesian Statistical Science) we Bulletin and on our homepage acknowledge Juli- launched a new joint educational initiative. The an and Arnold’s pathbreaking contributions and initiative formalized a long standing history of their impact on the lives of many people in our collaboration of ISBA and ASA/SBSS in related research community. They will be missed dearly! matters. Continued on page 2. ISBA Elections 2010. Please check out the elec- tion statements of the candidates for the upco- In this issue ming ISBA elections. We have an amazing slate ‰ A MESSAGE FROM THE BA EDITOR of candidates. Thanks to the nominating commit- *Page 2 tee, Mike West (chair), Renato Martins Assuncao,˜ Jennifer Hill, Beatrix Jones, Jaeyong Lee, Yasuhi- ‰ 2010 ISBA ELECTION *Page 3 ro Omori and Gareth Robert! ‰ SAVAGE AWARD AND MITCHELL PRIZE 2010 *Page 8 Sections. -
IMS Bulletin 39(4)
Volume 39 • Issue 4 IMS1935–2010 Bulletin May 2010 Meet the 2010 candidates Contents 1 IMS Elections 2–3 Members’ News: new ISI members; Adrian Raftery; It’s time for the 2010 IMS elections, and Richard Smith; we introduce this year’s nominees who are IMS Collections vol 5 standing for IMS President-Elect and for IMS Council. You can read all the candi- 4 IMS Election candidates dates’ statements, starting on page 4. 9 Amendments to This year there are also amendments Constitution and Bylaws to the Constitution and Bylaws to vote Letter to the Editor 11 on: they are listed The candidate for IMS President-Elect is Medallion Preview: Laurens on page 9. 13 Ruth Williams de Haan Voting is open until June 26, so 14 COPSS Fisher lecture: Bruce https://secure.imstat.org/secure/vote2010/vote2010.asp Lindsay please visit to cast your vote! 15 Rick’s Ramblings: March Madness 16 Terence’s Stuff: And ANOVA thing 17 IMS meetings 27 Other meetings 30 Employment Opportunities 31 International Calendar of Statistical Events The ten Council candidates, clockwise from top left, are: 35 Information for Advertisers Krzysztof Burdzy, Francisco Cribari-Neto, Arnoldo Frigessi, Peter Kim, Steve Lalley, Neal Madras, Gennady Samorodnitsky, Ingrid Van Keilegom, Yazhen Wang and Wing H Wong Abstract submission deadline extended to April 30 IMS Bulletin 2 . IMs Bulletin Volume 39 . Issue 4 Volume 39 • Issue 4 May 2010 IMS members’ news ISSN 1544-1881 International Statistical Institute elects new members Contact information Among the 54 new elected ISI members are several IMS members. We congratulate IMS IMS Bulletin Editor: Xuming He Fellow Jon Wellner, and IMS members: Subhabrata Chakraborti, USA; Liliana Forzani, Assistant Editor: Tati Howell Argentina; Ronald D. -
STATISTICAL SCIENCE Volume 36, Number 3 August 2021
STATISTICAL SCIENCE Volume 36, Number 3 August 2021 Khinchin’s 1929 Paper on Von Mises’ Frequency Theory of Probability . Lukas M. Verburgt 339 A Problem in Forensic Science Highlighting the Differences between the Bayes Factor and LikelihoodRatio...........................Danica M. Ommen and Christopher P. Saunders 344 A Horse Race between the Block Maxima Method and the Peak–over–Threshold Approach ..................................................................Axel Bücher and Chen Zhou 360 A Hybrid Scan Gibbs Sampler for Bayesian Models with Latent Variables ..........................Grant Backlund, James P. Hobert, Yeun Ji Jung and Kshitij Khare 379 Maximum Likelihood Multiple Imputation: Faster Imputations and Consistent Standard ErrorsWithoutPosteriorDraws...............Paul T. von Hippel and Jonathan W. Bartlett 400 RandomMatrixTheoryandItsApplications.............................Alan Julian Izenman 421 The GENIUS Approach to Robust Mendelian Randomization Inference .....................................Eric Tchetgen Tchetgen, BaoLuo Sun and Stefan Walter 443 AGeneralFrameworkfortheAnalysisofAdaptiveExperiments............Ian C. Marschner 465 Statistical Science [ISSN 0883-4237 (print); ISSN 2168-8745 (online)], Volume 36, Number 3, August 2021. Published quarterly by the Institute of Mathematical Statistics, 9760 Smith Road, Waite Hill, Ohio 44094, USA. Periodicals postage paid at Cleveland, Ohio and at additional mailing offices. POSTMASTER: Send address changes to Statistical Science, Institute of Mathematical Statistics, Dues and Subscriptions Office, PO Box 729, Middletown, Maryland 21769, USA. Copyright © 2021 by the Institute of Mathematical Statistics Printed in the United States of America Statistical Science Volume 36, Number 3 (339–492) August 2021 Volume 36 Number 3 August 2021 Khinchin’s 1929 Paper on Von Mises’ Frequency Theory of Probability Lukas M. Verburgt A Problem in Forensic Science Highlighting the Differences between the Bayes Factor and Likelihood Ratio Danica M. -
A Matching Based Theoretical Framework for Estimating Probability of Causation
A Matching Based Theoretical Framework for Estimating Probability of Causation Tapajit Dey and Audris Mockus Department of Electrical Engineering and Computer Science University of Tennessee, Knoxville Abstract The concept of Probability of Causation (PC) is critically important in legal contexts and can help in many other domains. While it has been around since 1986, current operationalizations can obtain only the minimum and maximum values of PC, and do not apply for purely observational data. We present a theoretical framework to estimate the distribution of PC from experimental and from purely observational data. We illustrate additional problems of the existing operationalizations and show how our method can be used to address them. We also provide two illustrative examples of how our method is used and how factors like sample size or rarity of events can influence the distribution of PC. We hope this will make the concept of PC more widely usable in practice. Keywords: Probability of Causation, Causality, Cause of Effect, Matching 1. Introduction Our understanding of the world comes from our knowledge about the cause-effect relationship between various events. However, in most of the real world scenarios, one event is caused by multiple other events, having arXiv:1808.04139v1 [stat.ME] 13 Aug 2018 varied degrees of influence over the first event. There are two major concepts associated with such relationships: Cause of Effect (CoE) and Effect of Cause (EoC) [1, 2]. EoC focuses on the question that if event X is the cause/one of the causes of event Y, then what is the probability of event Y given event Email address: [email protected], [email protected] (Tapajit Dey and Audris Mockus) Preprint submitted to arXiv July 2, 2019 X is observed? CoE, on the other hand, tries to answer the question that if event X is known to be (one of) the cause(s) of event Y, then given we have observed both events X and Y, what is the probability that event X was in fact the cause of event Y? In this paper, we focus on the CoE scenario. -
Tea: a High-Level Language and Runtime System for Automating
Tea: A High-level Language and Runtime System for Automating Statistical Analysis Eunice Jun1, Maureen Daum1, Jared Roesch1, Sarah Chasins2, Emery Berger34, Rene Just1, Katharina Reinecke1 1University of Washington, 2University of California, 3University of Seattle, WA Berkeley, CA Massachusetts Amherst, femjun, mdaum, jroesch, [email protected] 4Microsoft Research, rjust, [email protected] Redmond, WA [email protected] ABSTRACT Since the development of modern statistical methods (e.g., Though statistical analyses are centered on research questions Student’s t-test, ANOVA, etc.), statisticians have acknowl- and hypotheses, current statistical analysis tools are not. Users edged the difficulty of identifying which statistical tests people must first translate their hypotheses into specific statistical should use to answer their specific research questions. Almost tests and then perform API calls with functions and parame- a century later, choosing appropriate statistical tests for eval- ters. To do so accurately requires that users have statistical uating a hypothesis remains a challenge. As a consequence, expertise. To lower this barrier to valid, replicable statistical errors in statistical analyses are common [26], especially given analysis, we introduce Tea, a high-level declarative language that data analysis has become a common task for people with and runtime system. In Tea, users express their study de- little to no statistical expertise. sign, any parametric assumptions, and their hypotheses. Tea A wide variety of tools (such as SPSS [55], SAS [54], and compiles these high-level specifications into a constraint satis- JMP [52]), programming languages (e.g., R [53]), and libraries faction problem that determines the set of valid statistical tests (including numpy [40], scipy [23], and statsmodels [45]), en- and then executes them to test the hypothesis. -
Computational and Financial Econometrics (CFE 2016)
CFE-CMStatistics 2016 PROGRAMME AND ABSTRACTS 10th International Conference on Computational and Financial Econometrics (CFE 2016) http://www.cfenetwork.org/CFE2016 and 9th International Conference of the ERCIM (European Research Consortium for Informatics and Mathematics) Working Group on Computational and Methodological Statistics (CMStatistics 2016) http://www.cmstatistics.org/CMStatistics2016 Higher Technical School of Engineering, University of Seville, Spain 9 – 11 December 2016 c CFE and CMStatistics networks. All rights reserved. I CFE-CMStatistics 2016 ISBN 978-9963-2227-1-1 c 2016 - CFE and CMStatistics networks Technical Editors: Angela Blanco-Fernandez and Gil Gonzalez-Rodriguez. All rights reserved. No part of this book may be reproduced, stored in a retrieval system, or transmitted, in any other form or by any means without the prior permission from the publisher. II c CFE and CMStatistics networks. All rights reserved. CFE-CMStatistics 2016 International Organizing Committee: Ana Colubi, Stella Hadjiantoni, M. Dolores Jimenez-Gamero, Erricos Kontoghiorghes, George Loizou and Herman Van Dijk. CFE 2016 Co-chairs: Peter Boswijk, Jianqing Fang, Alain Hecq, Michael Smith. CFE 2016 Programme Committee: Alessandra Amendola, Ansgar Belke, Monica Billio, Josep Lluis Carrion-i-Silvestre, Roberto Casarin, Veronika Czellar, Serge Darolles, Manfred Deistler, Jean-David Fermanian, Sylvia Fruehwirth-Schnatter, Marc Hallin, Matthew Harding, Jan Jacobs, Menelaos Karanasos, Daniel Kaufmann, Robert Kunst, Degui Li, Marco Lippi, Zudi Lu, Luis Filipe Martins, Gian Luigi Mazzi, Tucker McElroy, J Isaac Miller, Bent Nielsen, Yasuhiro Omori, Michael Owyang, Gareth Peters, Michael Pitt, Stephen Pollock, Tommaso Proi- etti, Artem Prokhorov, Anders Rahbek, Francesco Ravazzolo, Jeroen Rombouts, Eduardo Rossi, Christos Savva, Willi Semmler, Laura Spierdijk, Giuseppe Storti, Jonathan Stroud, Genaro Sucarrat, Jean-Pierre Ur- bain, Shaun Vahey, Helena Veiga, Carlos Velasco, Helga Wagner, Martin Wagner, Toshiaki Watanabe and Peter Winker. -
Amstat News May 2011 President's Invited Column
May 2011 • Issue #407 AMSTATNEWS The Membership Magazine of the American Statistical Association • http://magazine.amstat.org Croatia Bosnia Serbia Herzegovina ThroughPeace Statistics ALSO: Meet Susan Boehmer, New IRS Statistics of Income Director Publications Agreement No. 41544521 Mining the Science out of Marketing Math Sciences in 2025 AMSTATNews May 2011 • Issue #407 Executive Director Ron Wasserstein: [email protected] Associate Executive Director and Director of Operations features Stephen Porzio: [email protected] 3 President’s Invited Column Director of Education Martha Aliaga: [email protected] 5 Pfizer Contributes to ASA's Educational Ambassador Program Director of Science Policy 5 TAS Article Cited in Supreme Court Case Steve Pierson: [email protected] 7 Writing Workshop for Junior Researchers to Take Place at JSM Managing Editor Megan Murphy: [email protected] 8 Meet Susan Boehmer, New IRS Statistics of Income Director Production Coordinators/Graphic Designers 10 Peace Through Statistics Melissa Muko: [email protected] 15 Statistics in Biopharmaceutical Research Kathryn Wright: [email protected] May Issue of SBR: A Festschrift for Gary Koch Publications Coordinator Val Nirala: [email protected] 16 Journal of the American Statistical Association March JASA Features ASA President’s Invited Address Advertising Manager Claudine Donovan: [email protected] 19 Technometrics Fingerprint Individuality Assessment Featured in May Issue Contributing Staff Members Pam Craven • Rosanne Desmone • Fay Gallagher • Eric Sampson Amstat News welcomes news items and letters from readers on matters of interest to the association and the profession. Address correspondence to Managing Editor, Amstat News, American Statistical Association, 732 North Washington Street, Alexandria VA 22314-1943 USA, or email amstat@ columns amstat.org. -
December 2000
THE ISBA BULLETIN Vol. 7 No. 4 December 2000 The o±cial bulletin of the International Society for Bayesian Analysis A WORD FROM already lays out all the elements mere statisticians might have THE PRESIDENT of the philosophical position anything to say to them that by Philip Dawid that he was to continue to could possibly be worth ISBA President develop and promote (to a listening to. I recently acted as [email protected] largely uncomprehending an expert witness for the audience) for the rest of his life. defence in a murder appeal, Radical Probabilism He is utterly uncompromising which revolved around a Modern Bayesianism is doing in his rejection of the realist variant of the “Prosecutor’s a wonderful job in an enormous conception that Probability is Fallacy” (the confusion of range of applied activities, somehow “out there in the world”, P (innocencejevidence) with supplying modelling, data and in his pragmatist emphasis P ('evidencejinnocence)). $ analysis and inference on Subjective Probability as Contents procedures to nourish parts that something that can be measured other techniques cannot reach. and regulated by suitable ➤ ISBA Elections and Logo But Bayesianism is far more instruments (betting behaviour, ☛ Page 2 than a bag of tricks for helping or proper scoring rules). other specialists out with their What de Finetti constructed ➤ Interview with Lindley tricky problems – it is a totally was, essentially, a whole new ☛ Page 3 original way of thinking about theory of logic – in the broad ➤ New prizes the world we live in. I was sense of principles for thinking ☛ Page 5 forcibly struck by this when I and learning about how the had to deliver some brief world behaves. -
ALEXANDER PHILIP DAWID Emeritus Professor of Statistics University of Cambridge E-Mail: [email protected] Web
ALEXANDER PHILIP DAWID Emeritus Professor of Statistics University of Cambridge E-mail: [email protected] Web: http://www.statslab.cam.ac.uk/∼apd EDUCATION 1967{69 University of London (Imperial College; University College) 1963{67 Cambridge University (Trinity Hall; Darwin College) 1956{63 City of London School QUALIFICATIONS 1982 ScD (Cantab.) 1970 MA (Cantab.) 1967 Diploma in Mathematical Statistics (Cantab.: Distinction) 1966 BA (Cantab.): Mathematics (Second Class Honours) 1963 A-levels: Mathematics and Advanced Mathematics (Double Distinction), Physics (Distinction) State Scholarship HONOURS AND AWARDS 2018 Fellow of the Royal Society 2016 Fellow of the International Society for Bayesian Analysis 2015 Honorary Lifetime Member, International Society for Bayesian Analysis 2013 Network Scholar of The MacArthur Foundation Research Network on Law and Neuroscience 2002 DeGroot Prize for a Published Book in Statistical Science 2001 Royal Statistical Society: Guy Medal in Silver 1978 Royal Statistical Society: Guy Medal in Bronze 1977 G. W. Snedecor Award for Best Publication in Biometry EMPLOYMENT ETC. 2013{ Emeritus Professor of Statistics, Cambridge University 2013{ Emeritus Fellow, Darwin College Cambridge 2007{13 Professor of Statistics and Director of the Cambridge Statistics Initiative, Cambridge University 2011{13 Director of Studies in Mathematics, Darwin College Cambridge 2007{13 Professorial Fellow, Darwin College Cambridge 1989{2007 Pearson Professor of Statistics, University College London 1983{93 Head, Department of Statistical Science, University College London 1982{89 Professor of Probability and Statistics, University College London 1981{82 Reader in Probability and Statistics, University College London 1978{81 Professor of Statistics, Head of Statistics Section, and Director of the Statistical Laboratory, Department of Mathematics, The City University, London 1969{78 Lecturer in Statistics, University College London VISITING POSITIONS ETC. -
Theory and Applications of Proper Scoring Rules
Theory and Applications of Proper Scoring Rules A. Philip Dawid, University of Cambridge and Monica Musio, Università di Cagliari Abstract We give an overview of some uses of proper scoring rules in statistical inference, including frequentist estimation theory and Bayesian model selection with improper priors. 1 Introduction The theory of proper scoring rules (PSRs) originated as an approach to probability forecasting (Dawid 1986), the general enterprise of forming and evaluating probabilistic predictions. The fundamental idea is to develop ways of motivating a forecaster to be honest in the predictions he announces, and of assessing the performance of announced probabilities in the light of the outcomes that eventuate. The early applications of this methodology were to meteorology (Brier 1950) and subjective Bayesianism (Good 1952; de Finetti 1975). However, it is becoming clear that the mathematical theory of PSRs has a wide range of other applications in Statistics generally. The aim of this paper is to give a brief outline of some of these. Some further details will appear in Dawid et al. (2014). After setting out the basic ideas and properties of PSRs in § 2, in § 3 we describe a number of important special cases. Section 4 describes how a PSR supplies an alternative to the likelihood function of a statistical model, and how this can be used to develop new estimators that may have useful properties of robustness and/or computational feasability. In particular, in § 5 we show that, by using an appropriate PSR, it is possible to avoid difficulties associated with intractable normalising constants. In § 6 the same ability to ignore a normalising constant is shown to supply an approach to Bayesian model selection with improper priors. -
The BUGS Project: Evolution, Critique and Future Directions
STATISTICS IN MEDICINE Statist. Med. 2009; 28:3049–3067 Published online 24 July 2009 in Wiley InterScience (www.interscience.wiley.com) DOI: 10.1002/sim.3680 The BUGS project: Evolution, critique and future directions 1, , 1 2 3 David Lunn ∗ †, David Spiegelhalter ,AndrewThomas and Nicky Best 1Medical Research Council Biostatistics Unit, Institute of Public Health, University Forvie Site, Robinson Way, Cambridge CB2 0SR, U.K. 2School of Mathematics and Statistics, Mathematical Institute, North Haugh, St. Andrews, Fife KY16 9SS, U.K. 3Department of Epidemiology and Public Health, Imperial College London, St. Mary’s Campus, Norfolk Place, London W2 1PG, U.K. SUMMARY BUGS is a software package for Bayesian inference using Gibbs sampling. The software has been instru- mental in raising awareness of Bayesian modelling among both academic and commercial communities internationally, and has enjoyed considerable success over its 20-year life span. Despite this, the software has a number of shortcomings and a principal aim of this paper is to provide a balanced critical appraisal, in particular highlighting how various ideas have led to unprecedented flexibility while at the same time producing negative side effects. We also present a historical overview of the BUGS project and some future perspectives. Copyright 2009 John Wiley & Sons, Ltd. KEY WORDS: BUGS; WinBUGS; OpenBUGS; Bayesian modelling; graphical models 1. INTRODUCTION BUGS 1 is a software package for performing Bayesian inference using Gibbs sampling 2, 3 . The BUGS[ ] project began at the Medical Research Council Biostatistics Unit in Cambridge in[ 1989.] Since that time the software has become one of the most popular statistical modelling packages, with, at the time of writing, over 30000 registered users of WinBUGS (the Microsoft Windows incarnation of the software) worldwide, and an active on-line community comprising over 8000 members. -
Conference Flyer
SPONSORED BY IMPORTANT DATES Abstract submission: March 16th, 2012 Early registration: May 4th, 2012 Istanbul is a vibrant, multi-cultural and cosmopolitan city bridging Europe and Asia. It has a unique F‹GÜR cultural conglomeration of east and CONGRESS & ORGANIZATION west, offering many cultural and touristic attractions, such as Hagia CONGRESS & ORGANIZATION Sophia, Sultanahmet, Topkap› e-mail: [email protected] Palace and Maiden's Tower. NAMED LECTURES PROGRAM COMMITTEE Anestis ANTONIADIS Adrian BADDELEY, University of Western Australia Dear Colleagues, University Joseph Fourier Vladimir BOGACHEV, Moscow State University LAPLACE LECTURE Krzysztof BURDZY, University of Washington The eighth World Congress in Probability and Statistics will be in Peter GREEN T. Tony CAI, University of Pennsylvania University of Bristol Elvan CEYHAN, Koç University Istanbul from July 9 to 14, 2012. It is jointly organized by the BERNOULLI LECTURE Probal CHAUDHURI, Indian Statistical Institute Bernoulli Society and the Institute of Mathematical Statistics. Steffen LAURITZEN Mine ÇA⁄LAR, Koç University Scheduled every four years, this meeting is a major worldwide event University of Oxford Erhan ÇINLAR, Princeton University WALD LECTURE for statistics and probability, covering all its branches, including Anthony DAVISON, École Polytechnique Fédérale de Lausanne theoretical, methodological, applied and computational statistics Yves LE JAN Rick DURRETT, Duke University and probability, and stochastic processes. It features the latest Université Paris-Sud