GREEN FILM C2B 14/3/03 10:54 Page 1 Richardson

Total Page：16

File Type：pdf, Size：1020Kb

GREEN FILM C2B 14/3/03 10:54 Page 1 Richardson Hjort and Hjort oxford statistical science series • 27 Green, Highly Structured Stochastic Systems (HSSS) is a modern strategy for building statistical models for real-world problems, for computing with them, and for interpreting the resulting inferences. Complexity is handled by working up from simple local assumptions in a coherent way, and that is the key to modelling, computation, inference and interpretation; the unifying framework is that of Bayesian hierarchical models. The aim of this book is to make recent developments in HSSS Highly Structured Stochastic Systems Stochastic Structured Highly accessible to a general statistical audience. Graphical modelling and Markov chain Monte Carlo (MCMC) methodology are central to the field, and in this text they are treated in depth. These underlying methodologies are balanced by substantive areas of application: to spatial statistics in epidemiology, ecology and image analysis; to genetics; and to infectious disease modelling. The book Highly Structured concludes with two topics (model criticism and Bayesian nonparametrics) that seek to challenge the parametric assumptions that otherwise underlie most HSSS models. Stochastic Systems For each of the 15 topics in the book, there is a substantial article by a leading author in the field, and two invited commentaries that comple- ment, extend, and discuss the topic, and should be read in parallel. All authors are distinguished researchers in the field, and were active participants in an international research programme on HSSS. ALSO AVAILABLE FROM OXFORD UNIVERSITY PRESS Graphical models Steffen L. Lauritzen Neural networks for pattern recognition Christopher M. Bishop Statistical models in epidemiology David Clayton and Michael Hills Analysis of longitudinal data: Second Edition Peter Diggle, Patrick Heagerty, Kung-Yee Liang, and Scott Zeger Models for repeated measurements: Second Edition J. K. Lindsey Edited by Peter J. Green ISBN 0-19-851055-1 Nils Lid Hjort 1 and 9 780198 510550 2 Sylvia Richardson.

Copy Link

Recommended publications

Causal Diagrams for Interference Elizabeth L
STATISTICAL SCIENCE Volume 29, Number 4 November 2014 Special Issue on Semiparametrics and Causal Inference Causal Etiology of the Research of James M. Robins ..............................................Thomas S. Richardson and Andrea Rotnitzky 459 Doubly Robust Policy Evaluation and Optimization ..........................Miroslav Dudík, Dumitru Erhan, John Langford and Lihong Li 485 Statistics,CausalityandBell’sTheorem.....................................Richard D. Gill 512 Standardization and Control for Confounding in Observational Studies: A Historical Perspective..............................................Niels Keiding and David Clayton 529 CausalDiagramsforInterference............Elizabeth L. Ogburn and Tyler J. VanderWeele 559 External Validity: From do-calculus to Transportability Across Populations .........................................................Judea Pearl and Elias Bareinboim 579 Nonparametric Bounds and Sensitivity Analysis of Treatment Effects .................Amy Richardson, Michael G. Hudgens, Peter B. Gilbert and Jason P. Fine 596 The Bayesian Analysis of Complex, High-Dimensional Models: Can It Be CODA? .......................................Y.Ritov,P.J.Bickel,A.C.GamstandB.J.K.Kleijn 619 Q-andA-learning Methods for Estimating Optimal Dynamic Treatment Regimes .............. Phillip J. Schulte, Anastasios A. Tsiatis, Eric B. Laber and Marie Davidian 640 A Uniformly Consistent Estimator of Causal Effects under the k-Triangle-Faithfulness Assumption...................................................Peter Spirtes
[Show full text]
Design and Analysis Issues in Family-Based Association
Emerging Challenges in Statistical Genetics Duncan Thomas University of Southern California Human Genetics in the Big Science Era • “Big Data” – large n and large p and complexity e.g., NIH Biomedical Big Data Initiative (RFA-HG-14-020) • Large n: challenge for computation and data storage, but not conceptual • Large p: many data mining approaches, few grounded in statistical principles • Sparse penalized regression & hierarchical modeling from Bayesian and frequentist perspectives • Emerging –omics challenges Genetics: from Fisher to GWAS • Population genetics & heritability – Mendel / Fisher / Haldane / Wright • Segregation analysis – Likelihoods on complex pedigrees by peeling: Elston & Stewart • Linkage analysis (PCR / microsats / SNPs) – Multipoint: Lander & Green – MCMC: Thompson • Association – TDT, FBATs, etc: Spielman, Laird – GWAS: Risch & Merikangas – Post-GWAS: pathway mining, next-gen sequencing Association: From hypothesis-driven to agnostic research Candidate pathways Candidate Hierarchical GWAS genes models (ht-SNPs) Ontologies Pathway mining MRC BSU SGX Plans Objectives: – Integrating structural and prior information for sparse regression analysis of high dimensional data – Clustering models for exposure-disease associations – Integrating network information – Penalised regression and Bayesian variable selection – Mechanistic models of cellular processes – Statistical computing for large scale genomics data Targeted areas of impact : – gene regulation and immunological response – biomarker based signatures – targeting
[Show full text]
A Joint Spatio-Temporal Model of Opioid Associated Deaths and Treatment Admissions in Ohio
A JOINT SPATIO-TEMPORAL MODEL OF OPIOID ASSOCIATED DEATHS AND TREATMENT ADMISSIONS IN OHIO BY YIXUAN JI A Thesis Submitted to the Graduate Faculty of WAKE FOREST UNIVERSITY GRADUATE SCHOOL OF ARTS AND SCIENCES in Partial Fulﬁllment of the Requirements for the Degree of MASTER OF ARTS Mathematics and Statistics May 2019 Winston-Salem, North Carolina Approved By: Staci Hepler, Ph.D., Advisor Miaohua Jiang, Ph.D., Chair Robert Erhardt, Ph.D. Acknowledgments First and foremost, I would like to express my sincere gratitude and appreciation to my advisor, Dr. Staci A. Hepler. Without your continuous support and inspiration, I could not have achieved this thesis research. Thank you so much for your patience, motivation, enthusiasm, and immense knowledge, which made this research even more enjoyable. In addition to my advisor, I would also like to thank Dr. Miaohua Jiang and Dr. Robert Erhardt, for the constant encouragement and advice, and for serving on my thesis committee. Furthermore, this thesis beneﬁted greatly from various courses that I took, including Bayesian statistics from Dr. Hepler, factor analysis and Markov Chain from Dr. Berenhaut, poisson linear model from Dr. Erhardt's Generalized Linear Model class, R programming skills from Dr. Nicole Dalzell and also reasoning skills from Dr. John Gemmer's Real Analysis class. Additionally, I would like to extend my thanks to Dr. Ellen Kirkman, Dr. Stephen Robinson and Dr. Sarah Raynor for their special attention and support to me, as an international student. Last but not least, I would like to thank my family for their sincere mental support throughout my life.
[Show full text]
Alan Agresti
Historical Highlights in the Development of Categorical Data Analysis Alan Agresti Department of Statistics, University of Florida UF Winter Workshop 2010 CDA History – p. 1/39 Karl Pearson (1857-1936) CDA History – p. 2/39 Karl Pearson (1900) Philos. Mag. Introduces chi-squared statistic (observed − expected)2 X2 = X expected df = no. categories − 1 • testing values for multinomial probabilities (Monte Carlo roulette runs) • testing fit of Pearson curves • testing statistical independence in r × c contingency table (df = rc − 1) CDA History – p. 3/39 Karl Pearson (1904) Advocates measuring association in contingency tables by approximating the correlation for an assumed underlying continuous distribution • tetrachoric correlation (2 × 2, assuming bivariate normality) X2 2 • contingency coefficient 2 based on for testing q X +n X independence in r × c contingency table • introduces term “contingency” as a “measure of the total deviation of the classification from independent probability.” CDA History – p. 4/39 George Udny Yule (1871-1951) (1900) Philos. Trans. Royal Soc. London (1912) JRSS Advocates measuring association using odds ratio n11 n12 n21 n22 n n n n − n n θˆ = 11 22 Q = 11 22 12 21 = (θˆ − 1)/(θˆ + 1) n12n21 n11n22 + n12n21 “At best the normal coefficient can only be said to give us. a hypothetical correlation between supposititious variables. The introduction of needless and unverifiable hypotheses does not appear to me a desirable proceeding in scientific work.” (1911) An Introduction to the Theory of Statistics (14 editions) CDA History – p. 5/39 K. Pearson, with D. Heron (1913) Biometrika “Unthinking praise has been bestowed on a textbook which can only lead statistical students hopelessly astray.” .
[Show full text]
The Stata Journal
The Stata Journal Editor Editor H. Joseph Newton Nicholas J. Cox Department of Statistics Department of Geography Texas A & M University Durham University College Station, Texas 77843 South Road 979-845-3142; FAX 979-845-3144 Durham City DH1 3LE UK [email protected] [email protected] Associate Editors Christopher F. Baum Jens Lauritsen Boston College Odense University Hospital Rino Bellocco Stanley Lemeshow Karolinska Institutet, Sweden and Ohio State University Univ. degli Studi di Milano-Bicocca, Italy A. Colin Cameron J. Scott Long University of California–Davis Indiana University David Clayton Thomas Lumley Cambridge Inst. for Medical Research University of Washington–Seattle Mario A. Cleves Roger Newson Univ. of Arkansas for Medical Sciences Imperial College, London William D. Dupont Marcello Pagano Vanderbilt University Harvard School of Public Health Charles Franklin Sophia Rabe-Hesketh University of Wisconsin–Madison University of California–Berkeley Joanne M. Garrett J. Patrick Royston University of North Carolina MRC Clinical Trials Unit, London Allan Gregory Philip Ryan Queen’s University University of Adelaide James Hardin Mark E. Schaffer University of South Carolina Heriot-Watt University, Edinburgh Ben Jann Jeroen Weesie ETH Zürich, Switzerland Utrecht University Stephen Jenkins Nicholas J. G. Winter University of Essex University of Virginia Ulrich Kohler Jeffrey Wooldridge WZB, Berlin Michigan State University Stata Press Production Manager Lisa Gilmore Stata Press Copy Editor Gabe Waggoner Copyright Statement: The Stata Journal and the contents of the supporting files (programs, datasets, and help files) are copyright c by StataCorp LP. The contents of the supporting files (programs, datasets, and help files) may be copied or reproduced by any means whatsoever, in whole or in part, as long as any copy or reproduction includes attribution to both (1) the author and (2) the Stata Journal.
[Show full text]
Statistical Methods for the Analysis of the Genetics of Gene Expression
Statistical methods for the analysis of the genetics of gene expression Matthias Alexander Heinig April 13, 2011 Dissertation zur Erlangung des Grades eines Doktors der Naturwissenschaften (Dr. rer. nat.) am Fachbereich Mathematik und Informatik der Freien Universität Berlin Gutachter: Prof. Dr. Martin Vingron Prof. Dr. Norbert Hübner 1. Referent: Prof. Dr. Martin Vingron 2. Referent: Prof. Dr. Norbert Hübner Tag der Promotion: 13. Dezember 2010 Preface The work that led to this thesis is part of several collaborative projects in which I participated. For the sake of clarity this thesis presents all results that led to the conclusions drawn and individual contributions will be detailed here. For each project I mention publications, highlight my contributions and acknowledge the work of my collaborators. Star project The work presented in section 3.3.1 was published in Nature Genetics [1] by the Star Consortium. It was a large EU funded project initiated by Norbert Hübner and led by Kathrin Saar aiming to assess genetic variability among rat strains and to provide haplotype and genetic maps. My contribution was the design of the strategy for the construction of the maps and its implementation and application to three different populations of rats. Moreover I performed the comparative analysis of these maps. I would like to acknowledge Herbert Schulz, Kathrin Saar, Norbert Hübner, Edwin Cuppen, Richard Mott, Domenique Gaugier and Marié-Thérèse Bihoreau for discussions and guidance and Denise Brocklebank for discussion and contributions to the implementation of the analysis. I would like to thank Oliver Hummel and Herbert Schulz for the preprocessing of genotype data.
[Show full text]
A Comparison of Bayesian and Likelihood-Based Methods for fitting Multilevel Models
Bayesian and likelihood-based methods in multilevel modeling 1 A comparison of Bayesian and likelihood-based methods for fitting multilevel models William J. Browne University of Nottingham, UK and David Drapery University of California, Santa Cruz, USA Summary. We use simulation studies, whose design is realistic for educational and medical re- search (as well as other fields of inquiry), to compare Bayesian and likelihood-based methods for fitting variance-components (VC) and random-effects logistic regression (RELR) models. The like- lihood (and approximate likelihood) approaches we examine are based on the methods most widely used in current applied multilevel (hierarchical) analyses: maximum likelihood (ML) and restricted ML (REML) for Gaussian outcomes, and marginal and penalized quasi-likelihood (MQL and PQL) for Bernoulli outcomes. Our Bayesian methods use Markov chain Monte Carlo (MCMC) estimation, with adaptive hybrid Metropolis-Gibbs sampling for RELR models, and several diffuse prior distri- 1 1 butions (Γ− (, ) and U(0; ) priors for variance components). For evaluation criteria we consider bias of point estimates and nominal versus actual coverage of interval estimates in repeated sam- pling. In two-level VC models we find that (a) both likelihood-based and Bayesian approaches can be made to produce approximately unbiased estimates, although the automatic manner in which REML achieves this is an advantage, but (b) both approaches had difficulty achieving nominal coverage in small samples and with small values of the intraclass correlation. With the three-level RELR models we examine we find that (c) quasi-likelihood methods for estimating random-effects variances perform badly with respect to bias and coverage in the example we simulated, and (d) Bayesian diffuse-prior methods lead to well-calibrated point and interval RELR estimates.
[Show full text]
Equivalence of Random-Effects and Conditional Likelihoods for Matched Case-Control Studies Ken Rice
Equivalence of random-effects and conditional likelihoods for matched case-control studies Ken Rice MRC Biostatistics Unit, Cambridge, UK January 8th 2004 Random effects derivations of conditional likelihoods Motivation Study of genetic c-erbB-2 ‘exposure’ and breast cancer (Rohan et al, JNCI, 1998) Number of cases exposed 0 1 0 1 0 1 0 1 0 1 0 1 0 0 2 1 0 8 2 0 15 1 0 12 0 Number of 1 0 0 1 3 0 1 1 0 1 6 1 1 6 1 controls 2 0 0 2 1 1 2 3 1 2 3 0 exposed 3 0 0 3 1 0 3 0 0 4 0 0 4 0 0 5 0 0 Each exposure measured imperfectly; Rate of False Positive Exposure ≈ 0.49 Rate of False Negative Exposure ≈ 0.00 (External validation study, 187 subjects) Is any useful inference possible? 1/24 Random effects derivations of conditional likelihoods Matched case-control studies Some disease of interest, want to find if a binary ‘exposure’ is associated For each diseased case, find a control matched for other covariates; age, sex, etc Then measure exposure of interest The exposures are outcomes of interest, not the disease status - must build a model for Pr(exposure), not Pr(disease) Common study design, efficient, simple, popular 2/24 Random effects derivations of conditional likelihoods Formal description Control exposure (1 or 0) is Z1k, Case exposure is Z2k, for pair k Z1k ~Bern( p1k ), Z2k ~Bern( p2k ) Assume odds ratio identical in all pairs k; − ψ = p2k 1 p1k − 1 p2k p1k ψ i.e.
[Show full text]
Regional Newsletter, June 2013
Regional Newsletter, June 2013 Regional Newsletter, June 2013 France, Netherlands and the BIR. This year the meeting In this issue will take place from 3rd-5th July at the University of St DANIEL FAREWELL & MICHAEL SWEETING Andrews. The keynote speaker for the meeting is Geert Welcome to the June 2013 edition of the BIR newsletter! Verbeke with further invited sessions on Statistical Ecol- As incoming editors we would like to thank Sara Genelet- ogy (David Borchers, Eleni Matechou and Jerome Dupuis), ti for all her efforts in producing the newsletter for the Advances in Genomics (Mark van de Wiel, Hae Won Uh last few years, and for helping us hit the ground running. and Hongzhe Li) and Mixed Modelling (Simon Wood, Ja- We would welcome any comments on the structure and nine Illian and Adeline Samson). The meeting will also content of the current newsletter, and contributions for host the Fisher Memorial Lecture given by Prof. David the next issue! Names in blue (such as ours at the top of Spiegelhalter in the evening of the 3rd July, which will this section) provide email links. be followed by a poster and buffet session. Prior to the In this issue we hear for the first time from our new lecture a presentation will be made to the winner of the regional president, Simon Thompson, and learn about Young Biometrician Prize 2013, which is awarded jointly upcoming regional society events including meetings by the BIR and the Fisher Memorial Trust. There will be around Species Distribution Modelling and Statistical a further 60 contributed talks within 12 contributed ses- Software Interoperability.
[Show full text]
IBC Dublin Is Here!
The International Biometric Society British & Irish Region Newsletter — July 2008 IBC Dublin is here! With around 800 participants from all over the world coming to savour the delights of Dublin, including many from special circumstance countries, this promises to be a truly international meeting. The excellent programme represents well the broad scope of biometry today. Thanks in particular to those BIR members who have worked hard to make the meeting a success. On the local organising committee, John Hinde (chair), An- drew Mead, Gabrielle Kelly, Adele Marshall, Patrick Murphy, John Whittaker and Angela Wood. On the scientific committee, David Balding and Andrew Mead. It's particularly appropriate that in Andrew we have a BIR international president, and we look forward to South County, Dublin. Kevin McGarry, used Andrew's presidential address. with permission (www.flickr.com/photos/mcgarry) British and Irish Region in Dublin INDEX On Monday 14th July David Bald- drop by the BIR/IBS stand on Mon- ing is inviting all BIR members and day morning/lunchtime. First Regional Seminar p. 3 friends to a half-hour meeting at 6.30pm on the UCD campus, moving This stand will be staffed at break on to a pub later in the evening. and lunchtimes throughout the con- David will talk briefly about some ference. A big thanks to those who In brief p. 4 future plans, and is keen to hear from have volunteered to staff this. Please members about what they would like drop by and say hello, and encour- the BIR to be doing. age potential new members to do the For details of times and places, same.
[Show full text]
Bayesian Demography 250 Years After Bayes
Bayesian Demography 250 Years after Bayes Short title: Bayesian Demography Jakub Bijak1* and John Bryant2 1 University of Southampton, Southampton, United Kingdom 2 Statistics New Zealand, Christchurch, New Zealand, and the University of Waikato, Hamilton, New Zealand * Address for correspondence: Department of Social Statistics and Demography, University of Southampton, Southampton, SO17 1BJ, United Kingdom, email: [email protected] Abstract Bayesian statistics offers an alternative to classical (frequentist) statistics. It is distinguished by its use of probability distributions to describe uncertain quantities, what leads to elegant solutions to many difficult statistical problems. Although Bayesian demography, like Bayesian statistics more generally, is around 250 years old, only recently has it begun to flourish. The aim of this paper is to review the achievements of Bayesian demography, address some misconceptions, and make the case for wider use of Bayesian methods in population studies. We focus on three applications: demographic forecasts, limited data, and highly-structured or complex models. In general, the ability to integrate information from multiple sources, and to describe uncertainty coherently, is the key advantage of Bayesian methods. Bayesian methods also allow for including additional (prior) information next to the data sample. As such, Bayesian approaches are complementary to many traditional methods, which can be productively re- expressed in Bayesian terms. 1 Bayesian Demography 250 Years after Bayes Keywords: Bayesian demography, Bayesian statistics, demographic methodology, population estimates and forecasts, statistical methods 1. Introduction The original paper of Thomas Bayes (1763) establishing the theorem that bears his name was presented to the Royal Society just over 250 years ago, two years after Bayes’s death, by his friend Richard Price (Courgeau 2012).
[Show full text]
A Comparison of Phasing Algorithms for Trios and Unrelated Individuals
A Comparison of Phasing Algorithms for Trios and Unrelated Individuals Jonathan Marchini,1 David Cutler,2 Nick Patterson,3 Matthew Stephens,4 Eleazar Eskin,5 Eran Halperin,6 Shin Lin,2 Zhaohui S. Qin,7 Heather M. Munro,7 Gonc¸alo R. Abecasis,7 and Peter Donnelly,1 for the International HapMap Consortium 1Department of Statistics, University of Oxford, Oxford, United Kingdom; 2McKusick-Nathans Institute of Genetic Medicine, Johns Hopkins University School of Medicine, Baltimore; 3Broad Institute of Massachusetts Institute of Technology and Harvard, Cambridge, MA; 4Department of Statistics, University of Washington, Seattle; 5Computer Science Department, Hebrew University, Jerusalem; 6The International Computer Science Institute, Berkeley; and 7Center for Statistical Genetics, Department of Biostatistics, University of Michigan, Ann Arbor Knowledge of haplotype phase is valuable for many analysis methods in the study of disease, population, and evolutionary genetics. Considerable research effort has been devoted to the development of statistical and compu- tational methods that infer haplotype phase from genotype data. Although a substantial number of such methods have been developed, they have focused principally on inference from unrelated individuals, and comparisons between methods have been rather limited. Here, we describe the extension of ﬁve leading algorithms for phase inference for handling father-mother-child trios. We performed a comprehensive assessment of the methods applied to both trios and to unrelated individuals, with a focus on genomic-scale problems, using both simulated data and data from the HapMap project. The most accurate algorithm was PHASE (v2.1). For this method, the percentages of genotypes whose phase was incorrectly inferred were 0.12%, 0.05%, and 0.16% for trios from simulated data, HapMap Centre d’Etude du Polymorphisme Humain (CEPH) trios, and HapMap Yoruban trios, respectively, and 5.2% and 5.9% for unrelated individuals in simulated data and the HapMap CEPH data, respectively.
[Show full text]