Short Book Reviews On-Line 2006

COMPUTATIONAL GENOME ANALYSIS. An Introduction.	R.C. Deonier, S. Tavaré and M.S. Waterman.
MODELING FINANCIAL TIME SERIES WITH S-PLUS, 2nd edition.	E. Zivot and J. Wang.
COMPUTATIONAL METHODS FOR OPTION PRICING.	Y. Achdou and O. Pironneau.
MATHEMATICS OF FINANCIAL MARKETS, 2nd edition.	R.J. Elliott and P.E. Kopp.
FINANCIAL ENGINEERING WITH FINITE ELEMENTS.	J. Topper.
COMMODITIES AND COMMODITY DERIVATIVES: MODELING AND PRICING FOR AGRICULTURALS, METALS, AND ENERGY.	H. Geman.
QUANTITATIVE RISK MANAGEMENT. Concepts, Techniques and Tools.	A.I. McNeil, R. Frey and P. Embrechts.
BAYESIAN STATISTICS AND MARKETING.	P.E. Rossi, G.M. Alleny and R. McCullough.
THEORY OF NEURAL INFORMATION PROCESSING SYSTEMS.	A.C.C. Coolen, R. Kühn and P. Sollich.
SIGNAL PROCESSING: A MATHEMATICAL APPROACH.	C.L. Byrne
LINEAR PROGRAMMING AND NETWORK FLOWS, 3rd edition.	M.S. Bazaraa, J.J. Jarvis and H.D. Sherali.
UNIVARIATE DISCRETE DISTRIBUTIONS, 3rd edition.	N.L. Johnson, A.W. Kemp and S. Kotz.
CELEBRATING STATISTlCS. Papers in Honour of Sir David Cox on the Occasion of his 80th Birthday.	A. Davison, Y. Dodge and N. Wermuth (Eds.).
SYMMETRY AND THE MONSTER. One of the Greatest Quests of Mathematics.	M. Ronan.
JAMES JOSEPH SYLVESTER, JEWISH MATHEMATICIAN IN A VICTORIAN WORLD.	K.H. Parshall.
GENERATING FUNCTIONOLOGY, 3rd edition.	H.S. Wilf.
BAYESIAN NETWORKS AND PROBABILISTIC INFERENCE IN FORENSIC SCIENCE.	F. Taroni, C. Aitken, P. Garbolino and A. Biedermann.
HANDBOOK OF UNIVARIATE AND MULTIVARIATE DATA ANALYSIS AND INTERPRETATION WITH SPSS.	R. Ho.
MODELS FOR DISCRETE DATA, Revised Edition.	D. Zelterman.
ESTIMATION OF DEPENDENCES BASED ON EMPIRICAL DATA. Reprint of 1982 edition. Empirical Inference Science. Afterword of 2006.	V. Vapnik.
FUNCTIONAL APPROACH TO OPTIMAL EXPERIMENTAL DESIGN.	V.B. Melas.
DESIGN AND MODELING FOR COMPUTER EXPERIMENTS.	K.-T. Fang, R. Li and A. Sudjanto.
INTRODUCTION TO CODING THEORY.	R.M. Roth.
RESPONSE SURFACE METHODOLOGY AND RELATED TOPICS.	A.I. Khuri (Ed.).
SCREENING. METHODS FOR EXPERIMENTATION IN INDUSTRY, DRUG DISCOVERY, AND GENETICS.	A.M. Dean and S.M. Lewis, Eds.
DOSE FINDING IN DRUG DEVELOPMENT.	N. Ting (Ed.).
BIOEQUIVALENCE AND STATISTICS IN CLINICAL PHARMACOLOGY.	S. Patterson and B. Jones.
MODELS FOR DISCRETE LONGITUDINAL DATA.	G. Molenberghs and G. Verbeke.
EXTENDING THE LINEAR MODEL WITH R: GENERALIZED LINEAR, MIXED EFFECTS AND NONPARAMETRIC REGRESSION MODELS.	J.J. Faraway.
ROBUST STATISTICAL METHODS WITH R.	J. Jureckova and J. Picek.
GENERALIZED ADDITIVE MODELS - AN INTRODUCTION WITH R.	S.N. Wood.
LATENT CURVE MODELS.	K.A. Bollen and P.J. Curran.
INFERENCE IN HIDDEN MARKOV MODELS.	O. Cappé, E. Moulines and T. Rydén.
STOCHASTIC MODELLING FOR SYSTEMS BIOLOGY.	D.J. Wilkinson.
DYNAMIC REGRESSION MODELS FOR SURVIVAL DATA.	T. Martinussen and T.H. Schelke.
STOCHASTIC AGEING AND DEPENDENCE FOR RELIABILITY.	C-D. Lai and M. Xie.
ANALYSIS OF INTEGRATED AND COINTEGRATED TIME SERIES WITH R.	B. Pfaff.
SELECTED STATISTICAL PAPERS OF SIR DAVID COX.	D.J. Hand and A.M. Herzberg (Eds.).
DATA MINING ET STATISTIQUE DÉCISIONELLE: L'INTELLIGENCE DANS LES BASES DU DONNÉE.	S. Tufféry.
THE ANALYSIS OF MEANS. A Graphical Method for Comparing Means, Rates, and Proportions.	P.R. Nelson, P.S. Wludyka and K.A.F. Copeland.
STOCHASTIC PROCESSES IN SCIENCE, ENGINEERING AND FINANCE.	F. Beichelt.
BINOMIAL MODELS IN FINANCE.	J. van der Hoek and R.J. Elliott.
MONTE CARLO SIMULATION AND FINANCE.	D.L. McLeish.
SELECTED PAPERS OF FREDERICK MOSTELLER.	S.E. Fienberg and D.C. Hoaglin (Eds.).
UNDERSTANDING UNCERTAINTY.	D.V. Lindley.
PRINCIPLES OF STATISTICAL INFERENCE.	D.R. Cox.
DESIGN AND ANALYSIS OF GAUGE R&R STUDIES.	R.K. Burdick, C.M. Borror and D.C. Montgomery.
LINEAR MODELS FOR OPTIMAL TEST DESIGN.	W.J. van der Linden.
COINCIDENCES, CHAOS AND ALL THAT MATH JAZZ. Making Light of Weighty Ideas.	E.B. Burger and M. Starbird.
INFORMATION GENERATION: How Data Rule Our World.	D.J. Hand.
GRAPHICS OF LARGE DATASETS: VISUALIZING A MILLION.	A. Unwin, M. Theus and H. Hofmann.
STATISTICAL EVIDENCE IN MEDICAL TRIALS: WHAT DO THE DATA REALLY TELL US?	S.D. Simon.
STATISTICAL MONITORING OF CLINICAL TRIALS: A UNIFIED APPROACH.	M.A. Proschan, K.K.G. Lan and J.T. Wittes.
INTRODUCTION TO RANDOMIZED CONTROLLED CLINICAL TRIALS, 2nd edition.	J.N.S. Matthews.
RANDOMIZATION, BOOTSTRAP AND MONTE CARLO METHODS IN BIOLOGY, 3rd edition.	B.F.J. Manly.
THE STATISTICAL ANALYSIS OF INTERVAL-CENSORED FAILURE TIME DATA.	J. Sun.
RANDOM FRAGMENTATION AND COAGULATION PROCESSES.	J. Bertoin.
VISUAL STATISTICS: SEEING DATA WITH DYNAMIC INTERACTIVE GRAPHICS.	F.W. Young, P.M. Valero-Mora, and M. Friendly.
MULTIDIMENSIONAL NONLINEAR DESCRIPTIVE ANALYSIS.	S. Nishisato.
A HANDBOOK OF STATISTICAL ANALYSES USING R.	B.S. Everitt and T. Hothorn.
SEMI-PARAMETRIC THEORY AND MISSING DATA.	A.A. Tsiatis.
NONPARAMETRIC FUNCTIONAL DATA ANALYSIS. THEORY AND PRACTICE.	F. Ferraty and P. Vieu.
HANDBOOK OF STATISTICAL DISTRIBUTIONS WITH APPLICATIONS.	K. Krishnamoorthy.
INTRODUCTORY STATISTICAL INFERENCE.	N. Mukhopadhyay.
AN INTRODUCTION TO RANDOM SETS.	H.T. Nguyen.
AN INTRODUCTION TO BAYESIAN ANALYSIS. THEORY AND METHODS.	J.K. Ghosh, M. Delampady and T. Samanta.
MARKOV CHAIN MONTE CARLO. STOCHASTIC SIMULATION FOR BAYESIAN INFERENCE, 2nd edition.	D. Gamerman and H.F. Lopes.
MEASURE THEORY AND PROBABILITY THEORY.	K.B. Athreya and S. Lahiri.
STATISTICAL ANALYSIS OF ENVIRONMENTAL SPACE-TIME PROCESSES.	N.D. Le and J.V. Zidek.
BENCHMARKING, TEMPORAL DISTRIBUTION, AND RECONCILIATION METHODS FOR TIME SERIES.	E.B. Dagum and P.A. Cholette.
A MODERN THEORY OF FACTORIAL DESIGNS.	R. Mukerjee and C.F.J. Wu.
DESIGN AND ANALYSIS OF EXPERIMENTS: Volume 2. Advanced Experimental Design.	K. Hinkelmann and O. Kempthorne.
GENERALIZED LINEAR MODELS WITH RANDOM EFFECTS.	Y. Lee, J.A. Nelder, and Y. Pawitan.
MEASUREMENT ERROR IN NONLINEAR MODELS. A MODERN PERSPECTIVE, 2nd edition.	R.J. Carroll, D. Ruppert, L.A. Stefanski and C.M. Crainiceanu.
SAMPLING ALGORITHMS.	Y. Tillé.
STATISTICS FOR REAL LIFE SAMPLE SURVEYS: NON-SIMPLE-RANDOM SAMPLES AND WEIGHTED DATA.	S. Dorofeev and P. Grant.
AMERICAN-STYLE DERIVATIVES. VALUATION AND COMPUTATION.	J. Detemple.
BRANCH-AND-BOUND APPLICATIONS IN COMBINATORIAL DATA ANALYSIS.	M.J. Brusco and S. Stahl.
EXACT ANALYSIS OF DISCRETE DATA.	K.F. Hirji.
MODERN MULTIDIMENSIONAL SCALING: THEORY AND APPLICATIONS, 2nd edition.	I. Borg and P.J.F. Groenen.
THE TOTAL SURVEY ERROR APPROACH: A GUIDE TO THE NEW SCIENCE OF SURVEY RESEARCH.	H.F. Weisberg.
MODELING LONGITUDINAL DATA.	R.E. Weiss.
MISSING DATA AND SMALL-AREA ESTIMATION. Modern Analytical Equipment for the Survey Statistician.	N.T. Longford.
SAMPLING METHODS: EXERCISES AND SOLUTIONS.	P. Ardilly and Y. Tillé.
CAUSALITY. New Statistical Methods.	I. Barukcic.
STOCHASTIC PROCESSES AND MODELS.	D. Stirzaker.
ALL OF NONPARAMETRIC STATISTICS.	L. Wasserman.
ESSENTIALS OF STATISTICAL INFERENCE.	G.A. Young and R.L. Smith.
STATISTICAL INFERENCE BASED ON DIVERGENCE MEASURES.	L. Pardo.
PRODUCTS OF RANDOM VARIABLES. Applications to Problems of Physics and to Arithmetical Functions.	J. Galambos and I. Simonelli.
STATISTICAL ANALYSIS OF MEDICAL DATA USING SAS.	G. Der and B.S. Everitt.
STATISTICAL MONITORING OF CLINICAL TRIALS FUNDAMENTALS FOR INVESTIGATORS.	L.A. Moyé.
DATA MONITORING IN CLINICAL TRIALS: A CASE STUDIES APPROACH.	D.L. DeMets, C.D. Furberg, L.M. Friedman (Eds.).
PUBLICATION BIAS IN META-ANALYSIS.	H.R. Rothstein, A.J. Sutton, and M. Borenstein (Eds.).
ALGEBRAIC STATISTICS FOR COMPUTATIONAL BIOLOGY.	L. Pachter and B. Sternfels (Eds.).

Contents:
1. Biology in a nutshell
2. Words
3. Word distributions and occurences
4. Physical mapping of DNA
5. Genome rearrangements
6. Sequential alignment
7. Rapid alignment methods: FASTA and BLAST
8. DNA sequence assembly
9. Signals in DNA
10. Similarity, distance and clustering
11. Measuring expression of genome information
12. Inferring the past: phylogenetic trees
13. Genetic variation in populations
14. Comparative genomics

APPENDIX A: A Brief Introduction to R
APPENDIX B: Internet Bioinformatics Resources
APPENDIX C: Miscellaneous Data

Readership: Upper-level undergraduate students and investigators in disciplines with applications to genomics

This book provides an introduction to a broad spectrum of the biological and computation background required for genome analysis. Topics are illustrated with examples and exercises. The latter are both mathematical and computational. The computational problems encourage the reader to investigate concepts using R.
The book is useful for its breadth. An impressive variety of topics are surveyed; some only lightly. For example, the reader is taken from "an introduction to probability" through to "simulating Markov chains" in 14 pages. Investigators facing real challenges will need to refer to sources covering particular methods in greater depth. This book is a useful starting point.

Contents:
1. S and S-Plus
2. Time series specification, manipulation, and visualization in S-Plus
3. Time series concepts
4. Unit root tests
5. Modeling extreme returns
6. Time series regression modeling
7. Univariate GARCH modeling
8. Long memory time series modeling
9. Rolling analysis in time series
10. Systems of regression equations
11. Vector autoregressive models for multivariate time series
12. Cointegration
13. Multivariate GARCH modeling
14. State space models
15. Factor models for asset returns
16. Term structure of interest rates
17. Robust change detection
18. Nonlinear time series models
19. Copula
20. Continuous-time models for financial time series
21. Generalized method of moments
22. Semi-nonparametric conditional density models
23. Efficient method of moments

Readership: Graduate (advanced MBA and PhD) students specializing in finance and financial economics in general, and in financial econometrics/statistics/time series in particular, researchers and practitioners in the finance industry

It provides theoretical and empirical discussions on exhaustive topics in modern financial econometrics, statistics and time series. To take full advantage of the book, the readers should have basic familiarity with S-Plus and have some background in matrix algebra and statistics (in particular regression theory) at the undergraduate level. As the book exclusively relies on S-Plus and S+FinMetrics as implementation tools, it comes with the shortcomings usually associated with S-Plus and S+FinMetrics. But it is definitely a good reference book for use in studying and/or researching in modern empirical finance if you like to work with S-Plus.

Contents:
1. Option pricing
2. The Black-Scholes equation: Mathematical analysis
3. Finite differences
4. The finite element method
5. Adaptive mesh refinement
6. American options
7. Sensitivites and calibration
8. Calibration of local volatility with European options
9. Calibration of local volatility with American options

Readership: Computational mathematicians, graduate students in mathematics, quants

This book focusses on the algorithms and the C++ code necessary for the pricing of options by numerical solution of the appropriate partial differential equation or integro-differential equations. The authors do not attempt to cover Monte-Carlo methods or tree methods, or to discuss the theory of finance which leads to particular models. The finite difference method and finite element method are applied to a number of different option types including basket options and options on assets driven by a Levy process. Other generalizations of the Black-Scholes model, including stochastic and local volatility models, are considered. The calibration of the local volatility using market information on option prices is discussed in detail. Adaptive methods which permit control on the numerical error are recommended.

Contents:
1. Pricing by arbitrage
2. Martingale measures
3. The first fundamental theorem
4. Complete markets
5. Discrete-time American options
6. Continuous-time stochastic calculus
7. Continuous-time European options
8. The American put option
9. Bonds and term structure
10. Consumption-investment strategies
11. Measures of risk

It is well-known that the main stream of option- pricing theory is based on a continuous-time framework for which sophisticated probability tools are required. This book begins with a presentation of pricing models for typical derivative securities in modern financial markets, such as options, futures and swaps within a discrete-time framework with a requirement of limited mathematical background of measure theory.
The main ideas of pricing models presented in this book are clearly presented in a rigorous and systematic way. This will give the reader a strong motivation to learn its counterpart of continuous-time finance. Necessary stochastic calculus is introduced rigorously. The beginner will, however, feel that most of these preparations are meaningful with regards of applications to finance. Some concepts fundamentally important both in probability and in finance are introduced in the places where they are needed.
The second edition adds new material from current active research areas. A new chapter on coherent risk measures for instance reflects the recent trend in research and applications in the area of risk management.
In summary, this is an excellent textbook in mathematical finance, and I can definitely recommend it.

Contents:
PART I: Preliminaries
1. Introduction
2. Some prototype models
3. The conventional approach: Finite differences
PART II: Finite Elements
4. Static 1D problems
5. Dynamic 1D problems
6. Static 2D problems
7. Dynamic 2D problems
8. Static 3D problems
9. Dynamic 3D problems
10. Nonlinear problems
PART III: Outlook
11. Future directions of research
PART IV: Appendices
A: Some Useful Results from Analysis
B: Some Useful Results from Stochastics
C: Some Useful Results from Linear Algebra
D: A Quick Introduction to PDE2D

Readership: Financial engineers, quants, students of economics and quantitative finance academics

The investment banking industry experiences a diversification and expansion of derivative instruments. With the increase in the volume and complexity of these products the demand for fast and accurate pricing and hedging techniques increases as well and advanced computational methods well established in engineering start to become an industry standard in quantitative finance. The book of Jürgen Topper is one of the first so far to present a comprehensive approach for pricing and hedging options by Galerkin or collocation type finite element solution of the corresponding Black-Scholes equation. The author demonstrates the generality of this approach which can handle in a unified way stochastic volatility models, multi-asset options with stochastic correlation, baskets with barriers or American like exercise conditions. Last but not least, the book includes a useful tool-kit of methods from stochastics, analysis and numerics and a quick introduction to Sewell's PDE2D, the finite element solver with which the examples in this book have been computed.

Contents:
Foreword
1. Fundamentals of commodity spot and future markets: instruments, exchanges and strategies
2. Equilibrium relationships between spot prices and forward prices
3. Stochastic modeling of commodity price processes
4. Plain-vanilla option pricing and hedging: from stocks to commodities
5. Risk-neutral valuation of plain-vanilla options
6. Monte Carlo simulations and analytical formulae for Asian, Barrier, and Quanto options
7. Agricultural commodity markets
8. The structure of metal markets and metal prices
9. The oil market as a world market
10. The gas market as the energy market of the next decade
11. Spot and forward electricity markets
12. Commodity swaptions, swing contracts and real options in the energy industry
13. Coal, emissions, and weather
14. Commodities as a new asset class

Readership: Energy companies and utilities practitioners, commodity and cash derivatives traders in investment banks, the agrifood business, commodity trading advisors, and hedge funds

The aim of the book is to describe the three fundamental groups of commodities: agriculturals, metals, and energy. Emphasis is placed on the similarities and differences between the commodity markets and stock and bond markets with the differences including such things as the physical delivery of commodities, and the existence of quantity risk, in addition to price risk in commodity markets, so that they include contracts of different kinds from those in the financial markets.
The author describes the book as attempting 'to bring together the fundamental results from economic theory, the constraints of physical delivery and the lessons learned in modern finance and option pricing'. She avoids too deep a level of mathematical abstraction, and has produced a work appropriate for practitioners as well as those wishing to learn the theory of the area.
Overall, it will make both a superb text and reference book. I expect to see this book become the bible of the field.

Contents:
1. Risk in perspectives
2. Basic concepts in risk management
3. Multivariate models
4. Financial time series
5. Copulas and dependence
6. Aggregate risk
7. Extreme value theory
8. Credit risk management
9. Dynamic credit risk models
10. Operational risk and insurance analytics

APPENDIX:
Miscellaneous Definitions and Results
Probability Distributions
Likelihood Inference

Readership: Advanced undergraduates and graduate students and finance industry professionals

Risk is the potential for or exposure to loss. This book provides a state-of-the-art discussion of the three main categories of risk in financial markets, market risk (the risk of changes in portfolio value due to fluctuations in the value of the underlying components), credit risk (financial risk due to default of a counterparty) and operational risk (financial risk due to failure of internal processes, systems or individuals or external events). The authors suggest that a truly satisfactory approach to risk would require a holistic approach, including all three types of risk and their interactions, but this is largely beyond the scope of the current models. This is a high level, but well-written treatment, rigorous (sometimes succinct), complete with theorems and proofs. There are no end-of-chapter exer cises but there are a large number of interesting examples. It runs the gamut from industry regulation and the Basel accords to dynamic credit risk models.

Contents:
1. Introduction
2. Bayesian essentials
3. Markov chain Monte Carlo methods
4. Unit-level models and discrete demand
5. Hierarchical models for heterogeneous units
6. Model choice and decision theory
7. Simultaneity

Case study 1: A choice model for packaged goods: Dealing with discrete quantities and quantity discounts
Case study 2: Modelling interdependent consumer preferences
Case study 3: Overcoming scale usage heterogeneity
Case study 4: A choice model with conjunctive screening rules
Case study 5: Modelling consumer demand for variety

APPENDIX A: An Introduction to Hierarchical Bayes Modelling in R
APPENDIX B: A Guide to Installation and Use of Bayesm

Customers make choices in the context of a particular market environment, and the aim of the marketer is to decide how to configure the aspects of that environment under their control in order to optimize some outcome. On the other hand, since all customers are different, some highly sophisticated models aimed at elucidating the nature of these differences are needed. ln particular, marketing applications are often naturally hierarchical, thus hierarchical models are particularly appropriate - and Bayesian methods provide a natural approach.
The first half of the book is really an introduction to Bayesian ideas and methods, with the second half illustrating their application to a range of real case studies. The book scores highly in both halves: the first half would serve as a very nice course text for anyone who wanted to introduce Bayesian ideas, while the second would be useful to demonstrate how these ideas are applied in practice. The examples illustrate how powerful these tools can be, and go significantly further than most standard applications of statistics in marketing contexts.
Since this book is driven from an applications perspective, the authors stress the necessity of being able to do the calculations in a reasonable time, and provide (on the Comprehensive R Archive Network on the web) an extensive R package for undertaking the sorts of analyses described in the book.
This book deserves to be widely adopted by business schools, and widely read by more numerate marketing practitioners.

Contents:
PART I: Introduction to Neural Networks
1. General introduction
2. Layered networks
3. Recurrent networks with binary neurons
4. Notes and suggestions for further reading
PART II: Advanced Neural Networks
5. Competitive unsupervised learning
processes
6. Bayesian techniques in supervised learning
7. Gaussian processes
8. Support vector machines for binary
classification
9. Notes and suggestions for further reading
PART III: Information Theory and Neural Networks
10. Measuring information
11. Identification of entropy as an information
measure
12. Building blocks of Shannon's information theory
13. Information theory and statistical inference
14. Applications to neural networks
15. Notes and suggestions for further reading
PART IV: Macroscopic Analysis of Dynamics
16. Network operation: Macroscopic dynamics
17. Dynamics of online learning in binary
perceptrons
18. Dynamics of online gradient descent learning
19. Notes and suggestions for further reading
PART V: Equilibrium Statistical Mechanics of Neural Networks
20. Basics of equilibrium statistical mechanics
21. Network operation: Equilibrium analysis
22. Gardner theory of task realizability
23. Notes and suggestions for further reading

APPENDIX A: Historical and Bibliographical Notes
APPENDIX B: Probability Theory in a Nutshell
APPENDIX C: Conditions for Central Limit Theorem to Apply
APPENDIX D: Some Simple Summation Identities
APPENDIX E: Gaussian Integrals and Probability Distributions
APPENDIX F: Matrix Identities
APPENDIX G: The Delta-Distribution
APPENDIX H: Inequalities Based on Convexity
APPENDIX I: Metrics for Parametrized Probability Distributions
APPENDIX J: Saddle-Point Integration

Readership: Graduate students of mathematics,
computer science, physics, engineering, biology

This is a weighty, authoritative and comprehensive textbook-treatment of the theory behind information processing in complex networks of simple interacting decision-making units. It is very clearly written, exhaustive in its coverage, and provides a very up-to-date account of its subject. It is specifically targeted at an interdisciplinary audience. Key mathematical ideas are developed fully within the text, though some level of prior mathematical knowledge is assumed. It is this reviewer's opinion that the clarity of presentation, coupled with inclusion of a series of appendices which provide essential background material on probability theory, linear algebra, numerical techniques etc., will ensure that the text will be found accessible and useful to a very wide audience.

Contents:
Introduction
1. Complex exponential function models
2. Matching and filtering
3. Fourier analysis
4. Analysis and synthesis
5. Fourier transforms and estimation
6. Matrix methods
7. Prediction and estimation
8. Nonlinear methods
9. Probabilistic methods
10. Iterative algorithm
11. More applications

Readership: Readers who wish to get a general overview of modern signal processing

This book presents a personal introduction to the mathematics of signal processing. It is intended to provide the necessary mathematical background for people engaged in using signal processing techniques in various applications. Topics covered include Fourier analysis, wavelets, bandlimited extrapolation, Kalman filtering, maximum a posteriori estimation, maximum likelihood, iterative algorithms plus an interesting set of applications (tomography, array processing, acoustic signal processing, etc). The book gives a readable general overview without getting bogged down in details.

Contents:
1. Introduction
2. Linear algebra, convex analysis, and polyhedral sets
3. The simplex method
4. Starting solution and convergence
5. Special simplex implementations and optimal conditions
6. Duality and sensitivity analysis
7. The decomposition principle
8. Complexity of the simplex algorithm and polynomial algorithms
9. Minimal-cost network flows
10. The transportation and assignment problems
11. The out-of-kilter algorithm
12. Maximal flow, shortest path, multicommodity flow, and network synthesis problems

The focus of this text is algorithmic techniques for linear programming and network flow problems. The simplex method is presented in detail and its adapatation to structured models such as network flow problems. The chapter on polynomial algorithms studies Karmakar's algorithm and its variants: interior point methods. This book is suitable as a reference or course text for a postgraduate course in mathematical programming. The reader needs to be fluent in linear algebra and calculus. Each chapter has a set of exercises; the bibliography of forty-eight pages is comprehensive.

Contents:
1. Preliminary information
2. Families of discrete distributions
3. Binomial distribution
4. Poisson distribution
5. Negative binomial distribution
6. Hypergeometric distributions
7. Logarithmic and Lagrangian distributions
8. Mixture distributions
9. Stopped-sum distributions
10. Matching, occupancy, runs and q-series distributions
11. Parametric regression models and miscellanea

Readership: Statisticians, experimental scientists,
social scientists, anyone interested
in statistics

The first edition of this classic book appeared in 1969, and it was first revised in 1992. The latest edition remains true to the guiding principles of the earlier editions, in that it provides a detailed, informative source of reference, without getting bogged down in intricate detail; applications are many and varied. The book is now some eighty pages longer than the last edition, and has over four hundred new references, most of which have appeared since 1992. The basic structure has changed little, with the additions of Lagrangian distributions, q-Series distributions and notably Parametric Regression Models to the titles of the eleven chapters. The motivation for the last of these is the availability of large computerized sets of data that may be analyzed by the appropriate methods. There has been much rearrangement of material, and rewriting to incorporate the latest research. A useful improvement is the new labelling of sections and sub-sections. With virtually every page packed with fascinating information, it is hard to put this book down. It is dedicated to the first author, who died when this edition was in its production stages.

David R. Cox: A brief biography (Y. Dodge)
1. Stochastic models for epidemics (V. Isham)
2. Stochastic soil moisture dynamics and vegetation response (A. Porporato, I. Rodriguez-Iturbe)
3. Theoretical statistics and asymptotics (N. Reid)
4. Exchangeability and regression models (P. McCullagh)
5. On semiparametric inference (A. Rotnitzky)
6. On non-parametric statistical methods (P. Hall)
7. Some topics in social statistics (D. Firth)
8. Biostatistics: The near future (S. Zeger, P. Diggle and K.-Y. Liang)
9. The early breast cancer trialists' collaborative group: A brief history of results to date (S. Darby, C. Davies and P. McGale)
10. How computing has changed statistics (B.D. Ripley)
11. Are there discontinuities in financial prices? (N. Shepherd)
12. On some concepts of infinite divisibility and their roles in turbulence, finance and quantum stochastics (O.E. Barndorff-Nielsen)

This volume is based on invited talks given at a conference to celebrate David Cox's 80th birthday. In the preface the choice of invited speakers and thus the range of topics of the included papers is acknowledged to be, in some sense, arbitrary. For the topics included, reflected in the chapter titles, the authors were to provide surveys of the research area. The extent to which the surveys are 'broad' and 'personal' varies but I enjoyed this sample of perspectives on various topics, some very particularly.
Of course, the brief biography of David Cox is a very welcome inclusion. It is particularly appropriate that, in a volume reflecting the wide range of David Cox's collaborations, there is an example of his handwritten notes as well as other photographs. Perhaps it also should be noted that the biography is shorter than the list of David Cox's publications up to 2004. I suspect this would meet with his approval.

Contents:
Prologue
1. Theaetetus's icosahedron
2. Galois: Death of a genius
3. Irrational solutions
4. Groups
5. Sophus Lie
6. Lie groups and physics
7. Going finite
8. After the war
9. The man from Uccle
10. The big theorem
11. Pandora's box
12. The leech lattice
13. Fischer's theorem
14. The atlas
15. A monstrous mystery
16. Construction
17. Moonshine
Notes

APPENDIX 1: The Golden Section
APPENDIX 2: The Witt Design
APPENDIX 3: The Leech Lattice
APPENDIX 4: The 26 Exceptions

Readership: Students and lay people interested in important mathematical concepts presented in a nontechnical manner

This book describes the historical development leading up to the discovery of the Monster - 'the most exceptional finite symmetry group in mathematics'. It interweaves personal narratives about the lives of those responsible for the several hundred year history culminating in this discovery, with the mathematical ideas and concepts. Because of the importance of group theory in modern mathematics, physics, and other areas, an impressive panoply of eminent names appears. The human stories are interesting, albeit brief, and the book nicely conveys the flavour and excitement of mathematical research, and the fact that mathematicians are not of a uniform kind, but come in all shapes, sizes, attitudes and opinions. It assumes essentially no prior knowledge of mathematics, even going so far as to present a footnote showing how to solve a quadratic equation, and to describe Cartesian coordinate systems.

Contents:
Introduction
1. Born to "the Faith in which the Founder of Christianity was Educated"
2. A price of dissent
3. The hollow walls of academe
4. Actuary by day ... Mathematician by night
5. Into the invariant-theoretic unknown
6. A new beginning
7. At war with the military
8. The uneasy years
9. Exploring familiar ground on unfamiliar territory
10. Tackling new challenges in a home away from home
11. A bittersweet victory
12. The final transition
13. Epilogue

James Joseph Sylvester (1814-1897) was born in London as James Joseph and the surname "Sylvester" was adopted from his elder brother Sylvester Joseph. It seems that the brothers "wanted an unequivocally non-Jewish surname as their calling card in life". This change in name, minor though it be, signifies much more in the tempestuous life of Sylvester. As the author writes, "Anglo-Jewry did not then exist as a well-defined social and cultural category. Essential tensions were nevertheless in evidence that would soon effect key changes."
This comprehensive biography is a sequel to the author's earlier work, James Joseph Sylvester: Life and Work in Letters. "The book aims to tell, for the first time, the complex story of Sylvester's life by situating that life as fully as possible within the political, religious, mathematical and social currents of the nineteenth century England. It aims to demythologize the man by placing him in his milieux at the same time that it demystifies his mathematics by revealing it as a very human endeavor. It aims to show how the man lived his life, what choices he made and why, how the world in which he lived affected him, and how he affected that world."
This fascinating biography traces the early days of Sylvester in Cambridge, when he came second in the mathematical tripos examination surpassing George Green, of "Green's theorem" fame. However, he was unable to graduate since being of Jewish faith, he could not sign up to the Thirty-Nine Articles of the Church of England which was a necessary condition for graduation. Never-theless, he was able to secure a position at the University of London, one of the few places which did not bar his entry based on his religion. But this position was in physics and he was more a mathematician. So he emigrated to the United States in the hope of finding a position in a mathematics department. It appears that at this time, he became romantically involved with a "Miss Marston" who, when he proposed marriage, turned him down, again on the grounds of religion! Dejected, Sylvester returned to England and took up an actuarial position by day and gave mathematical tuition by night. Apparently, one of his pupils was Florence Nightingale.
A turning point seems to have come in 1877, when Sylvester accepted a position at Johns Hopkins University where he founded the now famous American Journal of Mathematics. This was the first mathematical journal published in the United States. At Johns Hopkins, he could teach and research unfettered by the religion bias he encountered before. This period can be described as his most productive and Sylvester is credited with developing matrix theory, theory of equations, partition theory and some early foundational work in algebra at that time.
What emerges in this biography is Sylvester, the man, his encounters with the social, cultural and political milieu. Its extensive research into the vast archival sources gives the reader a detailed view of the Jewish mathematician living in the Victorian era, as the appropriate subtitle suggests. Both mathematicians and historians of mathematics would benefit from this book. It is well-written and engaging.

Contents:
1. Introductory ideas and examples
2. Series
3. Cards, decks, and hands: The exponential formula
4. Applications of generating functions
5. Analytic and asymptotic methods

The title of this book refers to the general method of associating a generating function to study a sequence of numbers, often defined by combinatorial constraints. Depending on the context, the generating function could be a power series, an exponential power series, a Dirichlet series or something more exotic. For instance, if the sequence is given by a recurrence relation, the power series associated with it turns out to be a rational function. One can then determine the singularities of this function which in turn can be used to give an explicit expression for the n-th term of the sequence.
Wilf's book is very well-written and easy to read by any serious mathematics student. Scientists in other disciplines often encounter the need to study sequences that naturally arise in their own discipline. The book is well-suited to them also. Just to give an indication of the friendly style of the book, I quote the opening line: "A generating function is a clothesline on which we hang up a sequence of numbers for display." For many who consult the "On-line Encylopedia of Integer Sequences", see
www.research.att.com/~njas/sequences/,
this book is a welcome guide.
The book is also suitable for a senior level course in combinatorics and is filled with numerous exercises. Unfortunately, the solutions are also included and so the instructor may have to create some new exercises for the class.

Contents:
1. The logic of uncertainty
2. The logic of Bayesian networks
3. Evaluation of scientific evidence
4. Bayesian networks for evaluating scientific evidence
5. DNA evidence
6. Transfer evidence
7. Aspects of the combination of evidence
8. Pre-assessment
9. Qualitative and sensitivity analysis
10. Continuous networks
11. Further applications

Readership: Forensic scientists, applied statisticians working in evidence evaluation, graduate students in these areas. Scientists, lawyers, and others interested in the evaluation of forensic evidence and/or Bayesian networks

This book eloquently argues the case for the use of Bayesian networks in forensic science. It uses the HUGIN system to illustrate, but, of course, alternatives could be used. It should be highly accessible by anyone with a fairly basic grasp of probability and a willingness to work through the many illustrations. It is clearly written, and nicely produced.
Statistics is, of course, the primary discipline concerned with the weighing and analysis of evidence. In recent years one has witnessed a rise in awareness of the importance of bringing this expertise to bear in courts of law, which are also concerned with weighing evidence, though there is still some way to go. This book will significantly advance that process. More generally, it also provides one of the best introductions to the theory and principles of Bayesian belief networks that I have seen, and could be used for a course in that, regardless of its particular application to forensic science.

Contents:
1. Inferential statistics and test selection
2. Introduction to SPSS
3. Multiple response
4. T-test for independent groups
5. Paired-samples t-test
6. One-way analysis of variance, with post hoc comparisons
7. Factorial analysis of variance
8. General linear model (GLM) multivariate analysis
9. General linear model: repeated measures analysis
10. Correlation
11. Linear regression
12. Factor analysis
13. Reliability
14. Multiple regression
15. Structural equation modelling
16. Nonparametric tests

Readership: Students of statistics and applied researchers in social science and psychology

This book offers an easy-to-read coverage of the uses and interpretation of basic statistical methods as applied to experimental data. The methods are presented using the software package SPSS (Statistical Product and Service Solutions, formerly Statistical Package for the Social Sciences) for Windows and the text proceeds through the analysis of a variety of sets of data (each accessible from the Web), interspersed throughout the text with illustrative SPSS code (also available from the Web). The main features of the package are presented and illustrated with examples using both the Windows method (point-and-click) and the traditional syntax method.
The mathematical prerequisites for using this book are minimal (appreciation of mathematical formulae and graphs) and calculus is not required. Some familiarity with using a computer is necessary in order to gain the most benefit from the text, and some previous experience of using a statistical software package would be advantageous. The text is suitable for private study.

Contents:
1. Introduction
2. Sampling distributions
3. Logistic regression
4. Log-linear models
5. Coordinate-free models
6. Additional topics

APPENDIX A: Power for the Chi-Squared Tests
APPENDIX B: A FORTRAN Program for Exact Tests in Tables
APPENDIX C: S-plus Programs for the Extended Hypergeometric Distribution

Readership: Statisticians, graduate students of statistics, numerate biomedical or sociological research workers

The author hopes this book will be used as a text to accompany a one semester master's graduate level course but it is also relevant to students following a statistics or biostatistics graduate degree. The book contains a large number of practical examples which have a health/medical bias.
The reader is expected to know about topics in elementary statistics such as sample means and variances, the Pearson chi-squared and statistical distributions such as the binomial and Poisson and a basic knowledge of SAS as a minimum. It is desirable for students to be familiar with matrix multiplication, maximum likelihood estimation, sufficient statistics, moment generating functions and hypothesis testing to be able to gain maximum benefit from the text. The reader is also encouraged to attempt all of the exercises that are included at the end of each chapter. At the end of the book there are hints and some solutions to these exercises.
The previous edition [Short Book Reviews, Vol. 19, p. 23] of this text has been used with many students and as a result this revised edition contains many more exercises. There are more applied exercises requiring computer solutions, usually in SAS. The author has also included new sets of data; most frequently from the health and medical sector, and previously unpublished data from a study of Tourette's syndrome in children. A computer file containing programs and sets of data is available by contacting the author.
The author has a wealth of experience in this area and this is demonstrated throughout the text with relevant poignant examples. This revised edition provides a sound introduction to the subject for graduate students and for practitioners needing a review of the methodology. An excellent practical book to be used in conjunction with relevant courses or to be used for reference and updating as required.

Contents:
1. The problem of estimating dependences from empirical data
Appendix to Chapter 1. Methods for Solving Ill-posed Problems
2. Methods of expected risk minimization
3. Methods of parametric statistics for the pattern recognition problem
4. Methods of parametric statistics for the problem of regression estimation
5. Estimation of regression parameters
6. A method of minimizing empirical risk for the problem of pattern recognition
Appendix to Chapter 6. Theory of Uniform Convergence of Frequencies to Probabilities: Sufficient Conditions
7. A method of minimizing empirical risk for the problem of regression estimation
Appendix to Chapter 7. Theory of Uniform Convergence of Means to their Mathematical Expectations: Necessary and Sufficient Conditions
8. The method of structural empirical minimization of risk
9. Solution of ill-posed problems. Interpretation of measurements using the method of structural risk minimization
Appendix to Chapter 9. Statistical Theory of Regularization
10. Estimation of functional values at given points
Appendix to Chapter 10. Taxonomy Problems

Postscript
Addendum I. Algorithms for pattern recognition
Addendum II. Algorithms for estimating nonindicator functions
Bibliographical Remarks

Empirical Inference Science. Afterword of 2006
1. Realism and Instrumentalism: Classical statistics and VC theory
2. Falsifiability and parsimony: VC dimension and the number of entities (1980-2000)
3. Noninductive methods of inference: Direct inference instead of generalization (2000-…)
4. The big picture

The first edition of this book was published in 1982, it was a translation by S. Kotz of the 1979 Russian edition. I reviewed that edition [Short Book Reviews, Vol. 3, p. 3]. This first edition is included in this new volume as it was, i.e. it is reprinted. The author has now added Chapters 1, 2, 3, as listed in the contents under "Empirical Inference Science, Afterword of 2006". These chapters were written as the author says: "to update the technical results presented [in the previous volume] and to describe a general picture of how the new ideas developed over these years." The fourth chapter called "The Big Picture" was written for his students and his students' students to tell "what is going on in the development of this science and in closely related branches of science in general (not only about some technical details)."

Contents:
1. Fundamentals of the optimal experimental design
2. The functional approach
3. Polynomial models
4. Trigonometrical models
5. D-optimal designs for rational models
6. D-optimal designs for exponential models
7. E- and c-optimal designs
8. The Monod model

The book presents a non-traditional approach to optimal design of experiments. The approach, named "functional" by the author and developed by him and other researchers over the last twenty-five years or so, relies on the Implicit Function Theorem which allows one to represent support points of locally optimal designs as functions of parameter values and then use Taylor series techniques. The book starts with a survey of general results of optimal design (various criteria of optimality, the generalized equivalence theorem, locally optimal designs for nonlinear models, numerical algorithms). Then the author illustrates the application of the functional approach to some classical models (polynomial, trigonometric, exponential). The last chapter of the monograph is devoted to the Monod model which is widely used in biological applications.

Contents:
PART I: An Overview
1. Introduction
PART II: Design of Computer Experiments
2. Latin hypercube sampling and its modifications
3. Uniform experimental design
4. Optimization in construction of designs for computer experiments
PART III: Modeling for Computer Experiments
5. Metamodeling
6. Model interpretation
7. Functional response

This book is a very close sibling of The Design and Analysis of Computer Experiments Th.J. Santner, B.J. Williams and W.I. Notz published by Springer in 2003 [Short Book Reviews, Vol. 24, p. 27]. The authors use simple probabilistic models to approximate computationally demanding deterministic models. I did not find any strong motivation why this type of approximation can successfully compete with the methods based on classical approximation theory. The use of the term "metamodeling" is rather confusing. The polynomial, splines, Kriging approximations, etc., which are used as simplified descriptions of more complex models, should be called "secondary" models, not metamodels.

Contents:
1. Introduction
2. Linear codes
3. Introduction to finite fields
4. Bounds on the parameters of codes
5. Reed-Solomon and related codes
6. Decoding of Reed-Solomon codes
7. Structure of finite fields
8. Cyclic codes
9. List decoding of Reed-Solomon codes
10. Codes in the Lee metric
11. MDS codes
12. Concatenated codes
13. Graph codes
14. Trellis and convolutional codes

Although this book is titled Introduction to Coding Theory, it contains topics beyond those normally found in an introductory text on this subject. It is aimed at upper undergraduate and postgraduate level.
It is assumed the reader has a good working knowledge of linear algebra, probability and discrete mathematics, but there are two separate chapters on finite fields which play such an important part in this subject. The first of these lays the framework for generalized Reed-Solomon codes which are commonly used in magnetic and optical storage media. Further material on finite fields, including minimal polynomials and cyclotomic cosets, is introduced later in order to discuss cyclic codes.
The mathematical style of this book is clear, concise and scholarly with a pleasing layout. There are numerous exercises, many with hints and many introducing further new concepts.
Altogether this is an excellent book covering a wide range of topics in this area, and including an extensive bibliography.

Contents:
1. Two-level factorial and fractional factorial designs in blocks of size two. Part 2,
by Y.J. Yang and N.R. Draper
2. Response surface experiments on processes with high variation,
by G. Gilmour and L.A. Trinca
3. Random run order, randomization and inadvertent split-plots in response surface experiments,
by J. Ganju and J.M. Lucas
4. Statistical inference for response surface optima,
by D.K.J. Lin and J.J. Peterson
5. A search method for the exploration of new regions in robust parameter design,
by G. Miro-Quesada and E. del Castillo
6. Response surface approaches to robust parameter design,
by T.J. Robinson and S.S. Wulff
7. Response surface methods and their application in the treatment of cancer with drug combinations: Some reflections,
by K.S. Dawson, T.J. Eller and W.H. Carter, Jr.
8. Generalized linear models and response transformation,
by A.C. Atkinson
9. GLM Designs: The dependence on unknown parameters dilemma,
by A.I. Khuri and S. Mukhopadhyay
10. Design for a trinomial response to dose,
by S.K. Fan and K. Chaloner
11. Evaluating the performance of non-standard designs: The San Cristobal design,
by L.M. Haines
12. 50 Years of mixture experiment research: 1955-2004, by G.F. Piepel
13. Graphical methods for comparing response surface designs for experiments with mixture components,
by H.B. Goldfarb and D.C. Montgomery
14. Graphical methods for assessing the prediction capability of response surface designs,
by J.J. Borkowski
15. Using fraction of design space plots for informative comparisons between designs,
by C.M. Anderson-Cook and A. Ozol-Godfrey
16. Concepts of slope-rotatability for second order response surface designs,
by S.H. Park
17. Design of experiments for estimating differences between responses and slopes of the response,
by S. Huda

Readership: Design of experiments aficionados, especially those in the response surface area

The seventeen papers cover a wide range of topics, all connected to response surface modeling. This is a useful collection to have on hand, or in the library.

Contents:
1. An overview of industrial screening experiments,
by D.C. Montgomery and C.L. Jennings
2. Screening experiments for dispersion effects,
by D. Bursztyn and D.M. Steinberg
3. Pooling experiments for blood screening and drug discovery,
by J.M. Hughes-Oliver
4. Pharmaceutical drug discovery: Designing the blockbuster drug,
by D.J. Cummins
5. Design and analysis of screening experiments with microarrays,
by P. Sebastiani, J. Jeneralczuk
and M.F. Ramoni
6. Screening for differential gene expressions from microarray data,
by J.C. Hsu, J.Y. Chang

and T. Wang
7. Projection properties of factorial designs for factor screening,
by C.-S. Cheng
8. Factor screening via supersaturated designs,
by S.G. Gilmour
9. An overview of group factor screening,
by M.D. Morris
10. Screening designs for model selection,
by W. Li
11. Prior distributions for Bayesian analysis of screening experiments,
by H. Chipman
12. Analysis of orthogonal saturated designs,
by D.T. Voss and W. Wang
13. Screening for the important factors in large discrete-event simulation models: Sequential bifurcation
and its applications,
by J.P.C. Kleijnen, B. Bettonvil
and F. Persson
14. Screening the input variables to a computer model via analysis of variance and visualization,
by M. Schonlau and W.J. Welch

Readership: Individuals who "sift through a very large number of factors, genes or compounds, in order to discover the few that influence a measured response"

This is an excellent aggregation of fourteen papers about screening, covering general areas. It includes recent work and provides first class summaries of various topics. The fourteen chapters have about twenty-two pages of references between them, helping the reader to delve further as needed. It is definitely an excellent library selection; some may prefer to have it even closer at hand. The cover art comes from p. 201 and is not a chapter summary!

Contents:
1. Introduction and new drug development process,
by N. Ting
2. Dose-finding based on preclinical studies,
by D. Salsburg
3. Dose-finding studies in Phase I and estimation of maximally tolerated dose,
by M. Modi
4. Dose-finding in oncology - nonparametric methods,
by A. Ivanova
5. Dose-finding in oncology - parametric methods,
by M. Tighiouart and A. Rogatko
6. Dose response: Pharmacokinetic-pharmacodynamic approach,
by N. Holford
7. General considerations in dose-response study designs,
by N. Ting
8. Clinical trial simulation - A case study incorporating efficacy and tolerability dose response,
by W. Ewy, P. Lockwood and C. Bramson
9. Analysis of dose-response studies - Emax Model,
by J. MacDougall
10. Analysis of dose-response studies - Modeling approaches,
by J. Pinheiro, F. Bretz, and M. Branson
11. Multiple comparison procedures in dose response studies,
by A.C. Tamhane and B.R. Logan
12. Partitioning tests in dose-response studies with binary outcomes,
by X. Ling, J. Hsu and N. Ting
13. Analysis of dose-response relationship based on categorical outcomes,
by C. Chuang-Stein and Z. Li
14. Power and sample size for dose-response studies,
by M. Chang and S.C. Chow

Readership: Statisticians and biostatisticians seeking an overview of the field of dose response studies, especially in the medical and regulatory contexts of clinical trials

This book is a collection of chapters contributed by different authors, each addressing a different aspect of the important problem of how to identify appropriate doses of medication during the drug development process - mainly in clinical trials from Phase I through Phase III. Most of the chapters are broad surveys of the issues involved in each subtopic. Most also assume prior statistical maturity of the reader. Except for the final chapter on sample size, the discussion is more conceptual than mathematical.
Consequently the most appropriate audience is statisticians and biostatisticians who desire an overview of the medical and regulatory contexts of the design and analysis of dose response studies as well as a digest of the field. Other readers may also benefit, for example clinical scientists, pharmacologists, and regulatory specialists, but most of the chapters imply prior knowledge of underlying statistical concepts that are not directly developed in the book. Most of the chapters provide many references where the reader can pursue further detail.
On the whole the individual chapters are well written, and the book overall is a nice reference from which to begin.

Contents:
1. Drug development and clinical pharmacology
2. History and regulation of bioequivalence
3. Testing for average bioequivalence
4. BE studies with more than two periods
5. Dealing with unexpected BE challenges
6. The future and recent past of BE testing
7. Clinical pharmacology safety studies
8. QTc
9. Clinical pharmacology efficacy studies
10. Population pharmacokinetics
11. Epilogue

Readership: Students, practitioners and researchers of bioavailability and bioequivalence studies

This book covers the statistical tools used in the assessment of bioequivalence and describes the use of statistics in clinical pharmacology studies of safety, QTc prolongation, efficacy, and population pharmacokinetics. The authors provide a historical perspective on the evolution of bioequivalence testing methods in the context of regulatory policies and public debates on these issues. Detailed description of current statistical concepts, methodology, and underlying assumptions of the design and analysis of bioavailability and bioequivalence studies are provided and exemplified using real data.

Contents:
PART I: Introductory Material
1. Introduction
2. Motivating studies
3. Generalized Iinear models
4. Linear mixed models for Gaussian longitudinal data
5. Model families
PART II: Marginal Models
6. The strength of marginal models
7. Likelihood-based marginal models
8. Generalized estimating equations
9. Pseudo-likelihood
10. Fitting marginal models with SAS
PART III: Conditional Models
11. Conditional models
12. Pseudo-likelihood
PART IV: Subject-Specific Models
13. From subject-specific to random-effects models
14. The generalized linear mixed model (GLMM)
15. Fitting generalized linear mixed Models with SAS
16. Marginal versus random-effects models
PART V: Case Studies and Extensions
17. The analgesic trials
18. Ordinal data
19. The epilepsy data
20. Non-linear models
21. Pseudo-likelihood for a hierarchical model
22. Random-effects models with serial correlation
23. Non-Gaussian random effects
24. Joint continuous and discrete responses
25. High-dimensional joint models
PART VI: Missing Data
26. Missing data concepts
27. Simple methods, direct likelihood, and WGEE
28. Multiple imputation and the EM algorithm
29. Selection models
30. Pattern-mixture models
31. Sensitivity analysis
32. Incomplete data and SAS

Readership: Statisticians, especially those working in the pharmaceutical industry, experimental scientists, post-graduate students of statistics

This book complements Verbeke and Molenberghs (2000), which focused on models based on the multivariate normal distribution. As the practical illustrations in the current book demonstrate, many real problems are far more complex in a variety of ways. This book covers the alternative models and approaches in a methodical and accessible manner. The emphasis in the book is on presenting methods for solving practical problems, and the authors succeed admirably in this. Motivation is provided by the wide-ranging and interesting examples in Chapter 2. These examples recur throughout the book, and new sets of data are also introduced as the book progresses. The material on how to use SAS to perform analyses is invaluable. However the book is not devoted to SAS alone, and, for instance, there are illustrations using MLwin and MIXOR. The material is clearly presented, and this undoubtedly reflects the experience derived from the many international workshops and courses that the authors have given on the subject matter. To take just two examples, Chapter 14 deals clearly with GLMMs, and covers the standard Bayesian and classical approaches, while Section VI is comprehensive in its treatment of the taxonomy of missing longitudinal data, and in ways of dealing with such missing data. The importance of this work is clear from the special issue of the Journal of the Royal Statistical Society (2006, Vol 169, Part 3), which is devoted to Attrition and Non-response. The research contributions of the authors of this book to these areas are evident in the papers of that issue. This book is very welcome, and will undoubtedly prove to be useful and influential.

Verbeke, G. and Molenberghs, G. (2000) Linear Mixed Models for Longitudinal Data. New York: Springer-Verlag.

Contents:
1. Introduction
2. Binomial data
3. Count regression
4. Contingency tables
5. Multinomial data
6. Generalised linear models
7. Other GLMs
8. Random effects
9. Repeated measures and longitudinal data
10. Mixed effect models for nonnormal responses
11. Nonparametric regression
12. Additive models
13. Trees
14. Neural networks

This book shows how the many variations that have been developed on the theme of the linear model have been implemented in the R software package. The R package, which has become somewhat ubiquitous in academic departments, is designed as part of the GNU project as freeware similar in appearance to S and hence S-plus. Due to its open design concept there have been very many different packages in R, written by many different groups of researchers, which implement both standard methodology and more cutting edge material. This volume does an excellent job of documenting the use, and existence, of these packages in the context of modern linear modelling using many data based examples. It is not however designed as an introductory text, neither for R or the basics of linear modelling but is written for those with some experience in both areas, and as such is a valuable contribution.

Contents:
1. Introduction
2. Mathematical tools of robustness
3. Basic characteristics of robustness
4. Robust estimators of real parameters
5. Robust estimators in linear model
6. Multivariate location model
7. Some large sample properties of robust procedures
8. Some goodness-of-fit tests
9. Appendix: R system

This book follows a mathematical, but still readable, path in exploring the key ideas behind the theory of robust statistical methods. It explores a wide range of robust methods, including differentiable statistical functions, distance of measures, influence functions, and asymptotic distributions. The mathematical approach is supplemented with code, written in the package R, which illustrates the ideas behind the theory, and the referenced R code can be downloaded from the book's website. The book is cleanly written and might form the theoretical foundation of a modern robustness course.

Contents:
1. Linear models
2. Generalized linear models
3. Introducing GAMs
4. Some GAM theory
5. GAMs in practice: mgcv
6. Mixed models and GAMMs

Readership: Readers who wish to work through generalized linear models into generalized additive models, using R as their computing package

To get the most from this book one must use the R computing system, available on the internet. This is either a positive or negative depending on the reader's feelings and capabilities.
Before putting that issue aside, this book is really excellent. It is a pleasure to read, is extremely clear, and progresses smoothly through topics of increasing complexity. The table of contents is very detailed, making it easy to find topics directly; only the chapter titles are given here. The book is highly recommended.

Contents:
1. Introduction
2. Unconditional latent curve model
3. Missing data and alternative metrics of time
4. Nonlinear trajectories and the coding of time
5. Conditional latent curve models
6. The analysis of groups
7. Multivariate latent curve models
8. Extensions of latent curve models

This book formulates traditional (multilevel) repeated measures models (individuals at level 2 and measurement occasions at level 1) in terms of structural equation modelling (SEM) models using standard notation and diagrams. Polynomial and multivariate models are described with and without covariate terms and a very brief reference to true nonlinear models (not simply polynomial representations) can be found in Chapter 4. Under the topic of multivariate latent curve models, linear first-order autoregressive dependencies are introduced among the level 1 random effects. In addition to the standard model for grouping factors, Chapter 6 has a brief section on mixture models where separate parameters are assumed for each different group where group membership is unknown. Categorical responses using a threshold modelling approach are dealt with in the final chapter which also discusses the case where the responses and covariates are treated as latent variables with associated multiple indicators.
The book will appeal to researchers and advanced students familiar with SEMs. Whether it provides an appropriate general introduction to repeated measures models, however, is open to question.
In many ways, the multilevel approach first clearly articulated by Laird and Ware (1982), provides a more natural characterization for repeated measures data. This approach now has a large literature that deals with topics hardly touched upon in the present book. These include time series models that incorporate random growth terms with general covariance structures that are functions of time and nonlinear response functions as well as Bayesian models using MCMC estimation. The most serious omission from this book, however, is consideration of data where each individual may have a completely unique set of time points. This case is covered only in passing (Section 3.2.5) yet in practice it is extremely common. Multilevel models (mentioned briefly in Section 2.6) take this case as the starting point, leading to general models of which most of those described in this book are just special cases. Multilevel models easily incorporate further levels of nesting and crossing of factors and recent work has begun to extend them to handle the repeated latent variable models discussed in the final chapter.

REFERENCE
Laird, N.M. and Ware, J.H. (1982). Random-effects models for longitudinal data. Biometrics 38, 963-974.

Contents:
1. Introduction
2. Main definitions and notations
3. Filtering and smoothing recursions
PART I: State Inference
4. Advanced topics in smoothing
5. Applications of smoothing
6. Monte Carlo methods
7. Sequential Monte Carlo methods
8. Advanced topics in sequential Monte Carlo
9. Analysis of sequential Monte Carlo methods
PART Il: Parameter Inference
10. Maximum likelihood inference, Part I: Optimization through exact smoothing
11. Maximum likelihood inference, Part II: Monte Carlo optimization
12. Statistical properties of the maximum Iikelihood estimator
13. Fully Bayesian approaches
PART III: Background and Complements
14. Elements of Markov Chain theory
15. An information-theoretic perspective on order estimation
PART IV: Appendices

The authors describe Hidden Markov Models (HMMs) as "one of the most successful statistical modelling ideas ... in the last forty years." The book considers both finite and infinite sample spaces. The first chapter provides the basic ideas, and a wide range of different areas of application; these include the stochastic modelling of biological sequences, ion channel modelling, capture-recapture, speech recognition, change point detection, stochastic volatility and regime switches in econometrics. Illustrative examples from these areas recur throughout the book. The authors recommend that the book is read initially from the algorithmic perspective, and only then should the reader consider the detailed theory. However, an important feature of the book is its theoretical underpinning, and to follow all of the arguments the reader should be familiar with measure theory. Theory is balanced by simulation approaches, which are given in detail for HMMs, including Gibbs sampling, Metropolis Hastings and Sequential Importance Sampling. Reversible jump MCMC is described for choosing between HMMs of different orders. The last section of the book presents background theory, including material on chains on general state spaces. The great scope of this book has been achieved with the help of R. Douc, C.P. Robert, G. Fort, P. Soulier, S. Boucheron and E. Gassiat, who have variously contributed a number of chapters and chapter sections. This fascinating book offers new insights into the theory and application of HMMs, and in addition it is a useful source of reference for the wide range of topics considered.

Contents:
1. Introduction to biological modelling
2. Representation of biochemical networks
3. Probability models
4. Stochastic simulation
5. Markov processes
6. Chemical and biochemical kinetics
7. Case studies
8. Beyond the Gillespie algorithm
9. Bayesian inference and MCMC
10. Inference for stochastic kinetic models
11. Conclusions
APPENDIX: SBML Models

Readership: Final year undergraduate and graduate students of statistics, bioinformatics and systems biology

The author of this text provides the reader with a comprehensive treatise of stochastic kinetic modelling of biological networks. The latest simulation techniques and research methods are presented and illustrated with a selection of examples and figures, as well as software code in R (the open-source statistical programming environment) for the computer-based implementation of a number of the examples in the text. The book is suitable for private study but would provide an excellent course companion for a variety of graduate courses in computational biology. A basic familiarity with linear algebra and matrix theory is assumed but the author covers the necessary mathematical background for a good appreciation of stochastic kinetic modelling of biological networks in the systems biology context. Model representation is greatly assisted by the authors choice of the Systems Biology Markup Language (SBML) which is the closest to a standard that exists in the systems biology area.

Contents:
1. Introduction
2. Probabilistic background
3. Estimation for filtered counting process data
4. Nonparametric procedures for survival data
5. Additive hazards models
6. Multiplicative hazards models
7. Multiplicative-additive hazards models
8. Accelerated failure time and transformation models
9. Clustered failure time data
10. Competing risks models
11. Marked point process models

APPENDIX A: Khmaladze's Transformation
APPENDIX B: Matrix Derivatives
APPENDIX C: The Timereg Survival Package For R

This book is a welcome addition to the literature on survival analysis for several reasons. The coverage of both multiplicative and, especially, additive models with time-varying covariates is well beyond that found in other books. There is also more emphasis on model checking than in most books. The theoretical background is presented rigorously, but details are often sketched or placed in position where they do not interrupt the flow, so the book is enjoyable to read. In addition, many worked examples based on R software are presented, and an R package (timereg) for implementing all the methodology is described in an Appendix and made available via the web. An interesting final chapter discusses regression methodology for longitudinal processes which are observed at discrete times, making use of the analogy with marked point processes. This book is an important resource for anyone with an interest in survival or event history analysis.

Contents:
1. Introduction
2. Concepts and applications of stochastic ageing
3. Bathtub shaped failure rate life distributions
4. Mean residual life-concepts and applications in reliability analysis
5. Weibull related distributions
6. An introduction to discrete failure time models
7. Tests of stochastic ageing
8. Bivariate and multivariate ageing
9. Concepts and measures of dependence in reliability
10. Reliability of systems with dependent components
11. Failure time data
Readership: Reliability-analysis researchers and practitioners, graduate students in reliability or applied probability

This book provides a review of how systems age, with emphasis on the dependence properties between components of a system. It begins with the standard basic definitions of hazard rate, mean residual life, and so on, describes bathtub distributions, as well as pointing out inadequacies of the Weibull family, and includes a discussion of discrete failure-time models, tests of constant failure rate, and dependence relationships between two or more lifetime variables. The final chapter gives thirty-three sets of failure-time data which have arisen from various applications, illustrating the various different shapes of distributions which occur. Reliability and survival analysis is now a very large area, and one could not expect a single book to be comprehensive. Nonetheless, I was a little surprised to see only two entries for 'competing risks' in the index in a book partly on dependence concepts in multi-component systems.
The book will provide a useful reference work and would be good supplementary reading for a graduate course on reliability analysis.

Contents:
PART I: Theoretical Concepts
1. Stationary autoregressive-moving average (ARMA) processes
2. Nonstationary time series
3. Cointegration
PART II: Unit Root Tests
4. Testing for the order of integration
5. Further considerations
PART III: Cointegration
6. Single equation methods
7. Multiple equation methods

Readership: Final year undergraduate and graduate students of statistics and econometrics

Topics in stationary and non-stationary time series, together with their application to univariate and multivariate analyses are covered in this book. The text is divided into three main parts ? broadly covering introductory concepts of stochastic processes in econometrics, tests for trends and the order of integration, and the methodology underlying cointegration. The main focus is on approaches that are applicable within econometrics where much of the formal models, hypotheses and test statistics have been developed. Integrated, seasonally integrated, and fractionally integrated time series will be familiar to many statisticians but the concept of cointegration less so. The author explains how easily the methods and tools can be implemented in R ? the open-source statistical programming environment. Exercises are provided at the end of each chapter and give the reader an opportunity to apply the presented tests and methods to previously published data sets. The text is suitable for private study but would provide an excellent course companion to computer-based laboratory classes.

Volume I: Design of Investigations, Statistical Methods and Applications;
Volume II: Foundations of Statistical Inference, Theoretical Statistics, Time Series and Stochastic Processes.

Contents:
Volume 1:
1. Design of investigations (10 papers)
2. Statistical methods (25 papers)
3. Applications (6 papers)

Volume 2:
4. Foundations of statistical inference (8 papers)
5. Theoretical statistics (22 papers)
6. Time series (4 papers)
7. Stochastic processes (11 papers)

Professors Hand and Herzberg should be thanked for the efforts which have culminated with the publication of these two volumes. The work presents eighty-six of David Cox's papers, selected from those published up to the end of 1993. Also included is a complete list of David Cox's publications through 2004.
There can be little doubt that the papers of David Cox provide a rich vein of statistical insight and their availability, in a very convenient form, will be very much appreciated by current and future statistical researchers. And to those who remember the papers, and perhaps their own, now rather dog-eared, reprints, publication in the original format seems particularly appropriate.
The selection process, a joint endeavor of David Cox and the editors, must have been difficult. The choice of 86 papers out of 236 reflects a reasonable set of priorities but will, for many readers and perhaps inevitably, leave a personal favourite or two 'out in the cold'. A two page paper from Technometrics in 1965 was one I would have been glad to see included.
What makes this work of special value is David Cox's willingness to contribute commentaries on the papers. With characteristic modesty he describes his famous paper on Regression models and life tables as "cited a fairly large number of times although no doubt read rather less often". However the task of providing the commentaries has been taken seriously and provides valuable additional insight and perspective on the papers, as well as some personal recollections. Discussion of the genesis of the paper on An analysis of transformations, joint with George Box, provides an interesting case of the latter. However, I must resist the temptation to say more about the commentaries. Do read and enjoy them for yourself. But, don't miss the remark about the paperless office in the Preface!
I think many will be pleased to have these two volumes on their bookshelves.

Table des matières:
1. Panorama du data mining
2. Le déroulement d'une étude de data mining
3. L'exploration et la préparation des données
4. L'utilisation des données commerciales
5. Aperçu sur les techniques de data mining
6. L'analyse factorielle
7. Les réseaux de neurones
8. Les techniques de classification automatique
9. La recherche d'associations
10. Les techniques de classement et de prédiction
11. Une application du data mining: Le scoring
12. Les facteurs de succès d'un projet de data mining
13. Les logiciels de statistique et data mining
14. Le text mining
15. Le web mining

ANNEXE A: Rappels de Statistique
ANNEXE B: Data Mining, Informatique et Libertés

Ce livre de data mining et statistique décisionelle est un ouvrage de référence pour les gestionnaires de bases de données dans tous les secteurs d'activité. Le 'forage de données' est une discipline moderne qui veut extraire des informations utiles d'une grande base de données en utilisant la statistique et l'informatique. Le livre est surtout basé sur les méthodes classiques mais il y a aussi des méthodes plus modernes comme arbres de décision, réseaux de neurones, support vector machines, algorithmes génétiques, bagging et boosting, etc. Le livre contient beaucoup d'exemples et applications avec les logiciels SAS, SPSS et SPAD. Il y a aussi des annexes intéressantes sur les bases des statistiques et sur les aspects juridiques de l'informatique. Ce livre est util pour le grand public des enseignants et practiciens dans ce domaine.

Contents:
1. Introduction
2. One-factor balanced studies
3. One-factor unbalanced studies
4. Testing for equal variances
5. Complete multifactor studies
6. Incomplete multifactor studies
7. Axial mixture designs
8. Heteroscedastic data
9. Distribution-free techniques

Readership: Practitioners who want to compare means arising from a variety of contexts

1. "The analysis of means (ANOM) is a graphical procedure used to quantify differences among treatment groups in a variety of ... situations."
2. "The book is a very practical comprehensive guide to the ANOM procedure. It has widespread application and is written at a level that can be comprehended by those who do not have a background in statistics. The ANOM procedures presented in the book provide the reader a means of communicating data to management, industry regulators, and peers in simple, graphical terms. The procedures combine the power of statistics with the simplicity of basic plotting techniques." (Sheri L. Meredith, Senior Quality Engineer, Vistakon, Division of Johnson & Johnson Vision Care, Inc., Jacksonville, Florida.)
Paragraphs 1 and 2 above are taken from the back of the book. l agree with everything Ms. Meredith says, and it seems pointless for me to rephrase it. l add only that a paper by C.V. Rao "Analysis of means - a review" appears, coincidently, in the Journal of Quality Technology 37, (October 2005), p. 308.

Contents:
1. Probability theory
2. Basics of stochastic processes
3. Random point processes
4. Markov chains in discrete time
5. Markov chains in continuous time
6. Martingales
7. Brownian motion

Readership: Senior undergraduate students and graduate students in mathematical and applied sciences with an interest in applied probability.

This book is a self-contained pre-measure theory introduction to stochastic processes with emphasis on applications, to repair and replacement problems in operations research and to actuarial risk analysis. The book assumes little more than basic probability (it includes an introductory chapter on probability theory) and mathematical maturity. The material is supported with many end-of-chapter problems, making it suitable as a course text.

Contents:
1. Introduction
2. The binomial model for stock options
3. The binomial model for other contracts
4. Multiperiod binomial models
5. Hedging
6. Forward and futures contracts
7. American and exotic option pricing
8. Path-dependent options
9. The Greeks
10. Dividends
11. Implied volatility trees
12. Implied binomial trees
13. Interest rate models
14. Real options

APPENDIX A: The Binomial Distribution
APPENDIX B: An Application of Linear Programming
APPENDIX C: Volatility Estimation
APPENDIX D: Existence of a Solution
APPENDIX E: Some Generalizations
APPENDIX F: Yield Curves and Splines

Readership: MBA students, undergraduate students in mathematics, statistics and economics, practitioners

This is a textbook on the mathematics of pricing and hedging financial derivatives with discrete stochastic models. It is directed towards a readership that is interested in the principles and applications of mathematical finance without having to deal with the technicalities of stochastic calculus. Therefore the book focuses on the so-called Cox-Ross-Rubinstein or binomial model, together with its various extensions. A nice feature is the very clear descriptions of many financial terms, which, on the one hand, are often missing in more mathematics-oriented books and, on the other hand, can be somewhat imprecise in textbooks aiming at the business community. A good example is the precise distinction between futures and forward contracts. Chapters 12 through 14 discuss subjects one rarely finds in other textbooks on mathematical finance: Chapter 12 deals with implied binomial trees, Chapter 13 introduces a couple of discrete-time models for interest rate curves, and Chapter 14 contains a novel discussion of real options.

Contents:
1. Introduction
2. Some basic theory of finance
3. Basic Monte Carlo methods
4. Variance reduction techniques
5. Simulating the value of options
6. Quasi-Monte Carlo multiple integration
7. Estimation and calibration
8. Sensitivity analysis, estimating derivatives, and the Greeks
9. Other methods and conclusions
Notes

Readership: Practitioners, researchers, senior undergraduate and graduate students of finance

Monte Carlo simulation provides a valuable tool in quantitative finance. The pricing of derivatives and other financial instruments is often based on complex mathematical models. It is now possible to compute the price of these instruments in practice using Monte Carlo techniques and the high computational power available. The book explains how to use Monte Carlo simulation in the valuation of financial instruments. The author focuses on particular problems where Monte Carlo simulation can play a very important role. The book provides insights and methodologies for problem formulation, model selection, calibration, simulation and analysis of the results. A set of exercises in the back of each chapter facilitates the use of the book for teaching quantitative finance. There is a useful Notes section, where the reader can find references to more comprehensive treatments of the subjects approached in each chapter of the book.

Contents:
1. Frederick Mosteller-A brief biography
2. Bibliography
3. 40 selected papers

After a truly extraordinary career, Frederick Mosteller passed away earlier this year at the age of 89. During more than fifty years as a faculty member at Harvard, he chaired four different departments, founded the Statistics Department, taught and advised countless students and faculty, wrote numerous textbooks, and conducted research in a wide range of areas. Professor Mosteller was also very active in statistical societies, government committees, the National Academy of Science, the American Academy of Arts and Sciences, and the American Association for the Advancement of Science, and held leadership positions in each. He was president of the International Statistical Institute from 1991-1993. To those that were fortunate enough to know him personally, he was a wonderful mentor and role model.
The contents of this volume were selected to illustrate Mosteller's broad range of research interests, and include papers on statistical methodology, applications in law, policy, medicine, education and ethics. For example, one paper examines a probabilistic basis for addressing causation and compensation for persons who experienced a particular disease and had previous exposure to a known causative agent. Others include such diverse topics as reporting results of clinical trials in surgical journals, methods for studying coincidences, statistics and ethics, an empirical study of the distribution of prime numbers, statistical aspects of the World Series (baseball), classroom and platform performance, and statistical methods for public policy.
By selecting such a wide range of topics, the Editors provide an excellent source of information on many interesting statistical issues, and a valuable reminder of the value of statistical thinking to a wide variety of important problems.

Contents:
Prologue
1. Uncertainty
2. Stylistic questions
3. Probability
4. Two events
5. The rules of probability
6. Bayes rule
7. Measuring uncertainty
8. Three events
9. Variation
10. Decision analysis
11. Science
12. Examples
13. Probability assessment
Epilogue
Readership: The blurb on the back of the book says this book is 'useful as a text for all students who have probability or statistics as part of their course, even at the most introductory level'. While this is true, it is also suitable for the much wider readership referred to in the preface: 'for the layman, for you, for everyone, because all of us are surrounded by uncertainty.'

This is a book about 'personal' probability, that interpretation which holds that probability is not an objective property of the world, but rather an individual's degree of belief that an event will occur. Professor Lindley, of course, is one of the founders of the modern theory, and in this book has attempted a non-technical and informal introduction to the ideas and applications. The book is wonderfully accessible, and is indeed a pleasure to read. It is full of examples, relating the technical issues under discussion to real issues in everyday life. I strongly recommend it as background reading for all students of probability and statistics (although perhaps statistics students should also read D.R. Cox's recent Principles of Statistical Inference [Short Book Reviews, Vol. 26, p. 39]).
The level of mathematics is truly minimal, and I would like to believe that all modern university students, of whatever discipline, could cope with it. I am certain that they would all benefit from reading it. And I am equally certain that lawyers, politicians, scientists and journalists (the list the author gives in his preface) would likewise benefit. Oh, and fuzzy set theorists.

Contents:
1. Preliminaries
2. Some concepts and simple applications
3. Significance tests
4. More complicated situations
5. Interpretations of uncertainty
6. Asymptotic theory
7. Further aspects of maximum likelihood
8. Additional objectives
9. Randomization-based analysis

This book grew from lectures to doctoral students at Chalmers/Gothenburg University. It retains much of the spirit and the economical style of a lecture series. A frequentist approach to inference is described, with many interesting examples and asides, and at each stage, the corresponding Bayesian techniques are presented for comparison. The book does not dwell on the formal principles of inference enunciated by Birnbaum and others in the 1960s ? the sufficiency principle, the likelihood principle, the conditionality principle ? but it is very much concerned with the frequentist-Bayesian dialogue, and the differences between inference and decision. On one level, it is a very useful and interesting introduction to statistical theory. On another level, it is a welcome personal statement by one of the foremost contributors to the foundations of inference.
In fact, in the development, principles of a prescriptive nature are kept to very few: (i) with parametric models, "a preference for sufficiency"; (ii) conditioning where it is clearly appropriate; (iii) and rejection of methods "that in repeated application would mostly give misleading answers". Such principles restrict both frequentist and Bayesian inference, but still leave ample room for differing perspectives.
There is something of a sense that formal systems of statistical inference could be regarded as testable models of reasoning under uncertainty. The validation of a particular approach is the viability of its results. "Some assurance of being somewhere near the truth takes precedence over internal consistency". Confidence intervals and tests of significance are the preferred expressions of inference at least partly because of their wide currency. The value of a much used non-Iikelihood method in a particular parametric problem may lie in its transparency.
In one way or another the book offers guidance through or signposts toward most modern developments in statistical inference. The treatments of asymptotic theory and of modified likelihood are especially to be recommended for their clarity and scope.

Contents:
1. Introduction
2. Balanced one-factor random models
3. Balanced two-factor crossed random models with interaction
4. Design of gauge R&R experiments
5. Balanced two-factor crossed random models with no interaction
6. Balanced two-factor crossed mixed models
7. Unbalanced one-and two-factor models
8. Strategies for constructing intervals with ANOVA models

APPENDIX A: Analysis of Variance
APPENDIX B: MLS and GCI Methods
APPENDIX C: Tables of F-Values

When a gauge is used by different people at different times, in different circumstances, results vary. Repeatability variation occurs when the same object is measured again and again under the same conditions. Reproducibility variations occur when the same gauge is used in changed circumstances, e.g., different operators, setups or time periods. Hence "gauge R&R". If you are still reading this review, you should know that this paperback book provides a well-thought-out explanation of the subject and will enable you to "determine whether the testing system is capable of monitoring the manufacturing process with the desired level of accuracy and precision." Recommended.

Contents:
1. Brief history of test theory and design
2. Formulating test specifications
3. Modeling test-assembly problems
4. Solving test-assembly problems
5. Models for assembling single tests
6. Models for assembling multiple tests
7. Models for assembling tests with item sets
8. Models for assembling tests measuring multiple abilities
9. Models for adaptive test assembly
10. Designing item pools for programs with fixed tests
11. Designing item pools for programs with adaptive tests
12. Epilogue

APPENDIX 1: Basic Concepts in Linear Programming
APPENDIX 2: Example of a Test-Assembly Problem in OPL Studio

Readership: Test specialists or creators
'Test' in this book refers to educational and psychological testing, a field which has advanced dramatically in recent decades. In his preface, the author says 'test theory has developed by careful modeling of response processes on test items and by using sophisticated statistical tools for estimating model parameters and evaluating model fit. In doing so, it has reached a current level of perfection that no one ever thought possible, say, two or three decades ago'. An important recent development is that of adaptive tests, which use computer programs to decide which test items a candidate should receive, so that the most accurate estimate of their ability can be obtained. The technical level is kept to a minimum, using little beyond 'high school algebra' and linear programming.
In general, the area is one which is insufficiently widely recognized and appreciated outside its own community of experts. I would like to think that this superb comprehensive synthesis of the field by one of its leading proponents will serve to broaden awareness of what can nowadays be achieved using such tools.

Contents:
Opening Thoughts
PART l: Understanding Uncertainty
1. Unbridled coincidences
2. Chaos reigns
3. Digesting life's data
PART II: Embracing Figures
4. Secrets held, secrets revealed
5. Sizing up numbers
6. A synergy between nature and number
PART III: Exploring Aesthetics
7. From precise beauty to pure chaos
8. Origami for the origamically challenged
9. A twisted turn in an amorphous universe
PART lV: Transcending Reality
10. The universe next door
11. Moving beyond the confines of our nutshell
12. In search of something still larger
Closing Thoughts

According to the book flyleaf this book 'fuses a professor's understanding of the hidden mathematical skeleton of the universe with the distorted sensibility of a stand-up comedian'.
Professionals might well then read no further! However, the book is not, of course, aimed at attracting them. It is aimed at getting across mathematical ideas in a light-hearted (and sometimes irreverent) way to those who are curious, but who fear all the technicalities.
However, there is a fine line between popularization and trivialization; therefore this is no easy task. The result of the authors' endeavours is entertaining, so that many of those in the target audience just might be enthused to find out more, assuming they get to read this.
In my opinion, though, it might well be that the authors' hope, 'that these puzzles, stories, and illustrations will stimulate great discussions and debates over dinner and cocktails,' could be somewhat over-optimistic!

Contents:
1. Let there be light
2. The origins of data
3. A lever to move the world
4. Big Brother's eyes
5. Modern data science
6. In data we trust
7. Deception and dishonesty with data
Epilogue: The beginning

Readership: Anyone interested in the history of scientific thought and data generation

The first three chapters in this book provide a history of science and the scientific method, and the history of the collection of data. Chapter 4 tells of the alarming situation of big brother's eye watching one. Chapter 5 discusses the analysis of data, whereas Chapter 6 discusses the trustworthiness of the data and sometimes the bad quality of the data and the problem of missing data. Is bad data better than no data at all? Chapter 7 shows the possibilities of lying with data. In the epilogue, Professor Hand notes "Data, in fact, form the very basis of modern civilization. The old and familiar aphorism says that 'knowledge is power', but that knowledge is built from data, and it is that knowledge, constructed from the basic building blocks of data, which has enabled our civilization to progress to its current state."
The book is very absorbing and informative. The text is enlivened by anecdotes such as that of Sir Isaac Newton, who lectured to an empty lecture theatre to fulfil his teaching obligations. An analogy with Jeremy Bentham's panopticon, used originally as a surveillance device at prisons, is that everyone is now being watched by an extended panopticon, with closed circuit TV cameras, enhancing one's safety, perhaps.
Professor Hand writes of data and information generation. Medical records are a necessity, but is it an infringement of one's privacy when records of one's purchases are used by many.
The book provides a very good read, either to read all the way through or to dabble into; it shows how one must be constantly vigilant. Although Professor Hand writes about the collection and analysis of data, many will be grateful that there are no equations and only eleven figures! One missing item is a discussion of the flaws in the now ubiquitous self-selected surveys.
John Dryden's quotation "Errors, like straws, upon the surface flow. He who would search for pearls must dive below." summarizes this interesting volume. And Professor Alison Wolf is correct in her pre-publication review when she says: "A really excellent book, full of interest whether you are a statistician, a student or a lay person aware of living in a world awash with data."

Contents:
1. Introduction
2. Statistical graphics
3. Scaling up graphics
4. Interacting with graphics
5. Multivariate categorical data-mosaic plots
6. Rotating plots
7. Multivariate continuous data-parallel coordinates
8. Networks
9. Trees
10. Graphics of a large dataset

Readership: Anyone interested in statistical computing and exploratory data analysis

This fascinating book looks at the question of visualizing large datasets from many different perspectives. Different authors are responsible for different chapters and this approach works well in giving the reader alternative viewpoints of the same problem. Interestingly the authors have cleverly chosen a definition of 'large dataset'. Essentially they focus on datasets with the order of a million cases. As the authors point out there are now many examples of much larger datasets but by limiting to ones that can be loaded in their entirety in standard statistical software they end up with a book that has great utility to the practitioner rather than just the theorist. Another very attractive feature of the book is the many colour plates, showing clearly what can now routinely be seen on the computer screen. The interactive nature of data analysis with large datasets is hard to reproduce in a book but the authors make an excellent attempt to do just this.

Contents:
1. Apples or oranges? Selection of the control group
2. Who was left out? Exclusions, refusals, and drop-outs
3. Mountain or molehill? The clinical importance of the results
4. What do the other witnesses say? Corroborating evidence
5. Do the pieces fit together? Systematic overviews and meta-analyses
6. What do all these numbers mean?
7. Where is the evidence? Searching for information

Readership: Health care professionals and statisticians who want to communicate with them

The author indicates that this book is written for "any health care professional who is making the effort to read and evaluate medical publications". He also says he "did not write the book to teach you how to conduct good research" and that the things he is proud of in the book are "Extensive use of real world examples", "Focus on statistical issues", "Avoidance of formulas and technical language" and "Presentation of counterpoints". The book is based on experience since 1997 in communicating how journal articles should be read and clearly benefits from an extensive electronic resource built up by the author.
The book succeeds very well in providing an interesting and informative introduction to the proper assessment of published medical research. The use of stories and anecdotes, some of long-standing use in this area, is effectively employed. One can raise minor quibbles such as in the mention of "number needed to harm" on page 56 before it is actually defined on page 125, or what appears to be the selective reporting of studies that made comparisons of observational and randomized studies of the same topic when presenting the counterpoint view that "Randomized trials are overrated." And, I think the word "studies" could have replaced "trials" in the title. Overall, however, this book provides an excellent way into assessment of the medical literature and covers the topics that should be covered. And while it may not be definitive, it is certainly eminently readable.

Contents:
1. Introduction
2. General framework
3. Power: Conditional, unconditional, and predictive
4. Historical monitoring boundaries
5. Spending functions
6. Practical survival monitoring
7. Inference following a group-sequential trial
8. Options when Brownian motion does not hold
9. Monitoring for safety
10. Bayesian monitoring
11. Adaptive sample size methods
12. Topics not covered

This book provides a detailed account of statistical methods for the interim monitoring of randomized comparative clinical trials. Traditional methods are covered in depth, and the text also includes brief discussions of safety monitoring and Bayesian monitoring, though monitoring using repeated confidence intervals and adaptive monitoring are not included as topics. A summary of available group sequential software is provided in an appendix. The extensive practical experience of the authors is reflected in the presentation of much of the material. This book would provide a valuable source of information for statisticians wishing to learn more about issues and methods for the interim monitoring of clinical trials.

Contents:
1. What is a randomized controlled trial?
2. Bias
3. How many patients do I need?
4. Methods of allocation
5. Assessment, blinding and placebos
6. Analysis of results
7. Further analysis: Binary and survival data
8. Monitoring accumulating data
9. Subgroups and multiple outcomes
10. Protocols and protocol deviations
11. Some special designs: Crossovers, equivalence, and clusters
12. Meta-analyses of clinical trials

Over recent decades randomized controlled clinical trials have become established as the method used to assess new medical treatments if any claims of the efficacy of a treatment are to find widespread acceptance. The author has revised and updated his first edition of this readable book by incorporating additional material pertaining to methods for balancing treatment allocations: together with the inclusion of a new chapter dealing with the analyses of clinical trials with binary outcomes and survival times. This revision continues to provide an introduction to the statistical methodology that underpins the randomized controlled trial and I recommend its inclusion in the reading list of any statistician wishing to attain a well-rounded education in the application of statistics. Administrative aspects of running a trial still receive little emphasis in the text but there are many excellent books on clinical trial methodology that may be consulted. The text assumes no underlying medical background, and is well written and easy to read.

Contents:
1. Randomization
2. The jacknife
3. The bootstrap
4. Monte Carlo methods
5. Some general considerations
6. One- and two-sample tests
7. Analysis of variance
8. Regression analysis
9. Distance matrices and spatial data
10. Other analyses on spatial data
11. Time series
12. Multivariate data
13. Survival and growth data
14. Nonstandard situations
15. Bayesian methods
16. Final comments

Readership: Postgraduate students and researchers of biometry, biostatistics, medicine and statistics
This is the third edition of the author's well-known, and popular, text on computer-intensive methods. In the second edition, a new chapter had been included on bootstrapping and the word bootstrap had been added to the title. However, is a third edition really warranted? The text has been extensively updated with the results of recent research but the emphasis remains firmly on biological applications. Substantial changes have been made to the chapters on regression and time series analyses. There is much to interest the general reader of statistical methodology and in particular, those interested in controversial topics for which statistical insight may be enlightening. The text is easy to read and contains many references which will facilitate private study, research and motivate further scientific investigation. I spent an interesting morning tracing references on superposed epoch analysis - a technique that appears to have many potential uses and is certainly worthy of wider use!

Contents:
1. Introduction
2. Inference for parametric models and imputation approaches
3. Nonparametric maximum likelihood estimation
4. Comparison of survival functions
5. Regression analysis of current status data
6. Regression analysis of Case II interval-censored data
7. Analysis of bivariate interval-censored data
8. Analysis of doubly censored data
9. Analysis of panel count data
10. Other topics

Interval-censored data, where one does not observe the exact times of events of interest, for example HIV serocoversion, but only that they occur in certain intervals defined in this case by patient visits to the physician, occur commonly in biostatistical practice. When, as is often the situation, these intervals are irregularly spaced and differ among patients, the usual methods for the analysis of complete data or right-censored survival data do not apply. This is the first book to summarize the substantial body of recent work in this field. It has something of the flavour of a lengthy review article. The reader is referred to the cited papers for detailed derivations and proofs, especially regarding the asymptotic theory, which is nonstandard. The book includes a number of applications, with full sets of data provided. Algorithms for fitting the various models presented are described carefully in the text and there is a short section discussing available statistical software, but detailed programs are not given. A frequentist approach is adopted throughout except for one short section on Bayesian methods. Those unfamiliar with concepts such as multiple imputation, sieve estimation, the EM algorithm and kernel smoothing may need to do some additional background reading on these topics. Although standard survival analysis methodology is reviewed briefly in the early chapters, the book will be of most benefit to those who have already mastered these techniques.

Contents:
Introduction
1. Self-similar fragmentation chains
2. Random partitions
3. Exchangeable fragmentations
4. Exchangeable coalescents
5. Asymptotic regimes in stochastic coalescence

Fragmentation and coagulation are natural phenomena that are observed in many sciences, including astrophysics, polymer chemistry, population genetics, biology and computer science, and which arise at a great variety of scales. This monograph develops in a formal manner mathematical models which may be used in situations where either phenomenon occurs randomly and repeatedly over time. It is very clearly written, as well as being beautifully produced, but is primarily aimed at readers with a very firm background in probability theory. The development is through a 'theorem/proof' style of presentation, which is very clearly signposted, and I am of the firm view that the book will succeed in making recent advances in random fragmentation and coagulation processes accessible to its intended readership. It is elegant and authoritative.

Contents:
PART I: Introduction
1. Introduction
2. Examples
PART II: Seeing Data - The Process
3. Interfaces and environments
4. Tools and techniques
PART III: Seeing Data - Objects
5. Seeing frequency data
6. Seeing univariate data
7. Seeing bivariate data
8. Seeing multivariate data
9. Seeing missing values

Readership: Novices, without a strong mathematical or statistical background, who are afraid of mathematics, who judge their mathematical skills to be inadequate, who have had negative experiences with statistics or mathematics, or have not recently exercised their skills. Practitioners, who are actively using statistical or data analytic methods, including teachers. Developers, who wish to develop new visual statistics methods

Forrest Young has been a leading light in developing dynamic interactive graphics for decades. I have been impressed and inspired by seeing descriptions of his methods at conferences, and so am delighted to see this book appear. As well as representing 'the major publication of [his] entire research career, research that has been the focus of [his] professional life for the last 40 years', this book also represents a collaboration with two co-authors who have each also made highly significant contributions to statistical graphics. I think a sentence from the preface nicely summarizes the aim of the book. It says 'our presentation emphasizes a paradigm for understanding data that is visual, intuitive, geometric, and active, rather than one that relies on convoluted logic, heavy mathematics, systems of algebraic equations, and passive acceptance of results.' Of course, one needs to recognize that there are places for different approaches to data analysis. Tools for helping one to gain an understanding of a set of data may well be different from tools for objectively inferring the information content of a set of data and communicating that to a second party. Eclecticism and inclusion are important in modern data analysis. The authors appreciate this.
Above all, however, I think this book illustrates well how dramatically the discipline of data analysis has been advanced by the computer. 'Dynamic, interactive graphics' is a meaningless concept without such machines. Dynamic, interactive graphics provide an excellent example of how the computer has extended the range of our ability to probe data, to understand data, and to think about data.
I cannot finish without observing that there is something curiously anomalous about the idea of a book (necessarily static and non-interactive) being used as a medium through which to describe dynamic interactive methods. The authors also recognize this, and stress that learning about such things necessarily requires abandoning the book at times, and actually sitting in front of a computer to try them out for oneself.

Contents:
PART I: Background
1. Motivation
2. Quantification with different perspectives
3. Historical overview
4. Conceptual preliminaries
5. Technical preliminaries
PART II: Analysis of Incidence Data
6. Contingency tables
7. Multiple choice data
8. Scoring data
9. Forced classification of incidence data
PART III: Analysis of Dominance Data
10. Paired comparison data
11. Rank-order data
12. Successive categories data
PART IV: Beyond the Basics
13. Further topics of interest
14. Further perspectives

Readership: Students in the social and biological sciences and researchers in such fields as marketing research, education, health and medical sciences, psychology, sociology, biology, ecology, agriculture, economics, political science, criminology, archaeology, geology and geography

I have taken the readership Iist from the preface of the book, and it invites two remarks. The first is that the length of the list is entirely justified. Researchers in almost every area could and should make use of the ideas and tools described in this book. The second is that the list does not include statisticians. It should. In fact, perhaps it is something of an implicit indictment of modern statistics and modern statistical teaching that this was not the obvious first item for the author to include in the list. Data analytic methods as widely used as those described in this book should certainly be taught in statistics courses. The book describes methods for quantifying categorical data, which are, of course, ubiquitous. Such methods have gone under various names, having been developed in parallel in a variety of areas - including those listed. The fact that this parallel development occurred again indicates the importance of the ideas: many people, in many different disciplines, have need of such tools.
This book provides a comprehensive overview of such methods; this is to be expected since Shizuhiko Nishisato is one of the seminal thinkers in the area. It includes historical perspectives and also nicely integrates the different developments and ties different strands together. It requires only intermediate mathematics, and only occasionally uses matrix algebra. I think it would serve as a rewarding book around which to base a course.

Contents:
1. An introduction to R
2. Simple inference
3. Conditional inference
4. Analysis of variance
5. Multiple linear regression
6. Logistic regression and generalised linear models
7. Density estimation
8. Recursive partitioning
9. Survival analysis
10. Analysing longitudinal data I
11. Analysing longitudinal data II
12. Meta-analysis
13. Principle component analysis
14. Multidimensional scaling
15. Cluster analysis

Readership: Graduate and advanced undergraduate students in statistics, statistical consultants

The R package, which has become somewhat ubiquitous in academic departments, is designed as part of the GNU project as freeware similar in appearance to S and hence S-plus. This book, using analyses of real sets of data, takes the reader through many of the standard forms of statistical methodology using R. The description of each technique is necessarily brief, but for a user with sound statistical background, this book is a very valuable reference. The sets of data used contain some old friends and in general are well chosen for illustrating the 'classroom demonstration' of a topic. The book is particularly good at highlighting the graphical capabilities of the language. Note that downloading the associated R packages and vignettes requires at least R version 2.2.1 and will not work on even relatively new systems with older versions. This might cause frustration if you are keen to get started working with the book.

Contents:
1. Introduction to semi-parametric models
2. Hilbert space for random vectors
3. The geometry of influence functions
4. Semi-parametric models
5. Other examples of semi-parametric models
6. Models and methods for missing data
7. Missing and coarsening at random for semi-parametric models
8. The nuisance, tangent, space and its orthogonal complement
9. Augmented inverse probability weighted complete-case estimators
10. Improving efficiency and double robustness with coarsened data
11. Locally efficient estimators for coarsened-data parametric models
12. Approximate methods for gaining efficiency
13. Double-robust estimator of the average causal treatment effect
14. Multiple imputation: A frequentist perspective

This book describes theory underlying semi-parametric methods that treat nuisance parameters as infinite-dimensional. The first five chapters review the theory for complete data, from a geometric perspective. The other chapters describe applications to missing (coarsened) data, including chapters on causal inference and multiple imputation. Mathematical demands are quite high, though the book presents key ideas and avoids proofs and technical details. As a Bayesian I am admittedly somewhat skeptical about the general approach, which is asymptotic and frequentist. For example covariates that I would view as ancillary and fixed are assigned a nonparametric model; the estimator (5.62) for a randomized clinical trial involves the probability of treatment allocation, an ancillary quantity of doubtful relevance to inference once sample counts are known. However, this approach has its devotees. Since much of the work in this area is very technical, it is most welcome to have a self-contained clearly written account by a highly-regarded author. The application to missing data is also clearly of great interest.

Contents:
PART I: Statistical Background for Nonparametric Statistics and Functional Data
1. Introduction to functional nonparametric statistics
2. Some functional datasets and associated statistical problematics
3. What is a well-adapted space for functional data?
4. Local weighting for functional variables
PART II: Nonparametric Prediction from Functional Data
5. Functional nonparametric prediction methodologies
6. Some detected asymptotics
7. Computational issues
PART III: Nonparametric Classification of Functional Data
8. Functional nonparametric supervised classification
9. Functional nonparametric unsupervised classification
PART IV: Nonparametric Methods for Dependent Functional Data
10. Mixing, nonparametric and functional statistics
11. Some selected asymptotics
12. Application to continuous time processes prediction
PART V: Conclusions
13. Small ball probabilities and semi-metrics
14. Some perspectives

The book gives the theoretical aspects as well as the practical implementation of the nonparametric modelling of functional data (like curves or images). The first part deals with the general statistical methodology for such data while the other parts focus on more advanced aspects: functional exploratory variables, classification of functional data, and dependent samples from functional data. The practical implementation is done in R and S-PLUS and is available from a companion website. This is certainly a very valuable book for anyone interested in this new methodology.

Contents:
1. Preliminaries
2. Discrete uniform distribution
3. Binomial distribution
4. Hypergeometric distribution
5. Negative binomial distribution
6. Logarithmic series distribution
7. Uniform distribution
8. Normal distribution
9. Chi-square distribution
10. F-distribution
11. Student's t-distribution
12. Exponential distribution
13. Beta distribution
14. Noncentral chi-squared distribution
15. Noncentral t-distribution
16. Laplace distribution
17. Lognormal distribution
18. Pareto distribution
19. Extreme value distribution
20. Cauchy distribution
21. Rayleigh distribution
22. Bivariate normal distribution
23. Sign test and confidence interval for the median
24. Wilcoxon signed-rank test
25. Wilcoxon rank-sum test
26. Nonparametric tolerance interval
27. Tolerance factors for a multivariate normal population
28. Distribution of the sample multiple correlation coefficient

A reference book containing commonly used facts about the standard distributions used in undergraduate statistics, the book also contains a CD with the software StatCalc the use of which is illustrated in the text.

Contents:
1. Probability and distributions
2. Moments and generating functions
3. Multivariate random variables
4. Sampling distributions
5. Notations of convergence
6. Sufficiency, completeness and ancillarity
7. Point estimation
8. Tests of hypotheses
9. Confidence intervals
10. Bayesian methods
11. Likelihood ratio and other tests
12. Large sample methods
13. Abbreviations, historical notes and tables

This book aims to be the basis of a single semester course at the graduate level that covers both probability and statistics. It follows a very classical approach to mathematical statistics and tries to aim somewhere between the level of Hogg and Craig (1995), to which it is very similar, and the much more comprehensive Schervish (1995). The book is unfortunately somewhat marred by poor presentation in language, typesetting and graphics.

References:
R.V. Hogg and A.T. Craig, (1995). Introduction to Mathematical Statistics, MacMillan.
M.J. Schervish, (1995). Theory of Statistics, Springer-Verlag.

Contents:
1. Generalities on probability
2. Some random sets in statistics
3. Finite random sets
4. Random sets and related uncertainty measures
5. Random closed sets
6. The Choquet integral
7. Choquet weak convergence
8. Some aspects of statistical inference with coarse data

Readership: Researchers and students in probability and statistics, mathematics, computer science

The book starts from situations in statistics and related fields where sets appear naturally as 'observations'. The theory to analyze such data is random set theory and it is based on ideas of Choquet, Kendall and Mathéron. The presentation of the material is from the ground up and there is an Appendix with basics from probability theory. The text is a solid introduction to this modern topic and brings together various results from the literature. The examples and the exercises make the book interesting for use in teaching.

Contents:
1. Statistical preliminaries
2. Bayesian inference and decision theory
3. Utility, proof and Bayesian robustness
4. Large sample methods
5. Choice of proofs for low-dimensional parameters
6. Hypothesis testing and model selection
7. Bayesian computations
8. Some common problems in inference
9. High-dimensional problems
10. Some applications

APPENDIX A: Common Statistical Densities
APPENDIX B: Bimbaum's Theorem on Likelihood
Principle
APPENDIX C: Coherence
APPENDIX D: Microarray
APPENDIX E: Bayes Sufficiency

This text provides a unique blend of theory, methods and applications that is suitable for a one-semester course in Bayesian analysis. Theoretical topics are supplemented by simulation and the elicitation of subjective priors is discussed, as well as the limitations of objective priors. The last chapter presents three methodological applications - hierarchical Bayesian modelling of spatial data, Bayesian nonparametric regression using wavelets and the estimation of an unknown regression function using Dirichlet multinomial allocation. The text contains a number of exercises at the end of each chapter but further problems are available from a dedicated website. A prerequisite knowledge of basic statistics, undergraduate calculus and linear algebra would be helpful before embarking upon a course of study with this text. However, the authors have indicated sections which are either highly technical or specialised and these may be omitted on a first reading of the text.

Contents:
1. Stochastic simulation
2. Bayesian inference
3. Approximate methods of inference
4. Markov chains
5. Gibbs sampling
6. Metropolis-Hastings algorithms
7. Further topics in MCMC

Readership: Postgraduate students and researchers of biostatistics and statistics, engineers and economists

This is the second edition of one of the most comprehensive and readable texts on stochastic simulation using the technique of Markov Chain Monte Carlo. The mathematics pre-requisites of a basic undergraduate course on probability and statistics should be sufficient to understand the material contained in the book. Almost a decade has elapsed since the publication of the first edition and this second edition has been extensively updated to include the recent literature. New sections on spatial models and model adequacy have now been included; together with the incorporation of more illustrative material. Many of the computer codes, written in R and WinBUGS, used in the examples and exercises in the text are available for download from the Web. This enhances the utility of the book, both as a reference text for researchers and as an accompanying text on modern Bayesian computation and Bayesian inference courses for students.

Contents:
1. Measures
2. Integration
3. Ly spaces
4. Differentiation
5. Product measures, convolutions, and transforms
6. Probability spaces
7. Independence
8. Laws of large numbers
9. Convergence in distribution
10. Characteristic functions
11. Central limit theorems
12. Conditional expectation and conditional probability
13. Discrete parameter martingales
14. Markov chains and MCMC
15. Stochastic processes
16. Limit theorems for dependent processes
17. The bootstrap
18. Branching processes

Readership: Anyone with senior undergraduate level or graduate level training in mathematics or probability

This book arose from, and provides the material for at least two graduate level courses, a one-semester course in measure followed by a one-semester course in advanced probability theory. In addition, there are interesting and non-standard topics that are not usually included in a first course in measure-theoretic probability including Markov Chains and MCMC, the bootstrap, limit theorems for martingales and mixing sequences, Brownian motion and Markov processes. The material is well-supported with many end-of-chapter problems.

Contents:
PART I: Environmental Processes
1. First encounters
2. Case study
3. Uncertainty
4. Measurement
5. Modeling
PART II: Space-Time Modeling
6. Covariances
7. Spatial prediction: Classical approaches
8. Bayesian kriging
9. Hierarchical Bayesian kriging
PART III: Design and Risk Assessment
10. Multivariate modeling
11. Environmental network design
12. Extremes
PART IV: Implementation
13. Risk assessment
14. R tutorial

The authors are experts in environmental space-time processes and cover in this book a wealth of methodology for dealing with data from this field. There are some technical sections containing the mathematics and statistics. The others are non-technical ones on environmental topics. There is a special Chapter 14 with a tutorial on the proposed methods, using the R software. It is certainly a very useful book for researchers and consultants in this challenging field.

Contents:
1. Introduction
2. The components of a time series
3. The Cholette-Dagum regression based benchmarking method - The additive model
4. Covariance matrices for benchmarking and reconciliation methods
5. The Cholette-Dagum regression based benchmarking method - The multiplicative model
6. The Denton method and its variants
7. Temporal distribution, interpolation and extrapolation
8. Signal extraction and benchmarking
9. Calendarization
10. A unified regression-based framework for signal extraction, benchmarking and interpolation
11. Reconciliation and balancing systems of time series
12. Reconciling one-way classified systems of time series
13. Reconciling the marginal totals of two-way classified systems of series
14. Reconciling two-way classified systems of series

APPENDIX A: Extended Gauss-Markov Theorem
APPENDIX B: An Alternative Solution for the Cholette-Dagum Model for Binding Benchmarks
APPENDIX C: Formulas for Some Recurring Matrix Products
APPENDIX D: Seasonal Regressors
APPENDIX E: Trading-Day Regressors

Readership: Econometricians, time series analysts, statistical researchers in statistical agencies

This book is an attempt at bringing together the specialized research in topics related to benchmarking and reconciliation. The first chapter gives an introduction to several topics such as benchmarking, interpolation, temporal distribution, signal extraction, calendarization and reconciliation. Chapter 2 discusses time series decomposition models while Chapters 3 to 6 and 8 detail benchmarking. Later chapters discuss topics indicated in the contents. Each chapter is somewhat self-contained and contains examples to illustrate the methodology. As indicated before, it is mainly oriented to researchers and hence the book assumes a good statistical background such as the knowledge of advanced Linear Models and Autoregressive Integrated Moving Average (ARIMA) time series models. It is an excellent reference book for people working in this area. To facilitate comparison among methods an example of real data, the Canada Total Retail Trade series, is followed throughout the book. It also gives several appendices to describe some theoretical results and an extensive list of references for readers to follow up.

Contents:
1. Introduction and overview
2. Fundamentals of factorial designs
3. Two-level fractional factorial designs
4. Fractional factorial designs: General case
5. Designs with maximum estimation capacity
6. Minimum aberration designs for mixed factorials
7. Block designs for symmetrical factorials
8. Fractional factorial split-plot designs
9. Robust parameter design

Minimum aberration (shortened to MA in this text) is a central feature in this excellent monograph, and the authors' stated intentions of "providing a comprehensive account of the modern theory of factorial designs with the MA approach at its core" (p. 4) are fully achieved. The book would be an excellent text for advanced courses in experimental design and for use as a reference text. There are many tables of designs throughout. However, these designs are necessarily listed in abbreviated form. Selecting a specific design for practical use requires mastery of the notation.

Contents:
1. General incomplete block design
2. Balanced incomplete block designs
3. Construction of balanced incomplete block designs
4. Partially balanced incomplete block designs
5. Construction of partially balanced incomplete block designs
6. More block designs and blocking structures
7. Two-level factorial designs
8. Confounding in 2n factorial designs
9. Partial confounding in 2n factorial designs
10. Designs with factors at three levels
11. General symmetrical factorial design
12. Confounding in asymmetrical factorial designs
13. Fractional factorial designs
14. Main effect plans
15. Supersaturated designs
16. Search designs
17. Robust-design experiments
18. Lattice designs
19. Crossover designs

In 1952 Oscar Kempthorne's book gave a lucid and comprehensive account of the theory of the design of experiments as it was then. For many years that book was a standard reference on the subject. Sadly it was allowed to go out of print. In 1994 the present authors published Volume 1 of a massive revision of the older book. That first volume [Short Book Reviews, Vol. 14, p. 44] dealt with the more basic aspects of the subject. Now, Oscar Kempthorne having in the meantime died, Klaus Hinkelmann has brought the rewriting to a conclusion with the volume under review. Kempthorne's original has in total roughly trebled in length.
The result is a massively impressive work of scholarship concentrating on the traditional core of the subject. For the main such categories of design, the combinatorics of construction are developed and linear-model-based analyses set out in detail and often illustrated with empirical examples, not discussed in depth in subject-matter terms, and with SAS programs. There is a cautious chapter on Taguchi-methods, but so far as I could see from my sampling of the text and from the index, there is no mention of central composite designs and similar arrangements. The theory of optimal design is outlined and reservations over its practicality set out. The emphasis is strongly on designs of complex structure and on traditional methods of error control by blocking rather than by, for instance, spatial or temporal modelling.
The two volumes form an encyclopaedic account of the chosen aspects of that challenging topic, the statistics of experimental design.

Contents:
1. Classical likelihood theory
2. Generalized linear models
3. Quasi-likelihood
4. Extended likelihood functions
5. Normal linear mixed models
6. Hierarchical GLMs
7. HGLMs with structured dispersion
8. Correlated random effects for HGLMs
9. Smoothing
10. Random-effect models for survival data
11. Double HGLMs
12. Further topics

Readership: Statisticians, scientists with good background in statistical theory and modelling

Professors Y. Lee and J.A. Nelder proposed a new approach to handling random effects in generalized linear models in a 1996 discussion paper in the Journal of the Royal Statistical Society Series B [Vol 58, pp. 619-678]. Since then, they have published a substantial body of literature extending their hierarchical "h-likelihood" and extended quasi-likelihood approach to inference. This book provides a comprehensive summary of their work. However, it is much more than that, and even statisticians who do not agree with their approach to inference will find much here of interest. By picking various sections mainly from the first half of this book, some instructors might find this to be a useful text for a course on generalized linear models. There are no end-of-chapter exercises, but there are many ideas that will be useful for a student to mull over, including an overview of likelihood theory and ways of defining and using likelihood functions.

Contents:
1. Introduction
2. Important concepts
3. Linear regression and attenuation
4. Regression calibration
5. Simulation extrapolation
6. Instrumental variables
7. Score function methods
8. Likelihood and quasilikelihood
9. Bayesian methods
10. Hypothesis testing
11. Longitudinal data and mixed models
12. Nonparametric estimation
13. Semiparametric regression
14. Survival data
15. Response variable error

Readership: Applied statisticians, scientists and engineers who wish to fit structural models to their data

This is the second edition of a research Ievel monograph, first published over ten years ago, about modeling with predictors that are subject to measurement error, a situation which arises in practice much more often than usually admitted. The text describes a variety of approaches to handling such data and illustrates the models and methods with numerous examples. The early chapters set the scene with a clear description of the problem through many examples, a discussion of the different types of error and the distinction between functional and structural models. These two types of models form the basis of the second and third parts of the book respectively, with the final part devoted to more specialized material including generalized Iinear structure with an unknown link function, hypothesis testing and non-parametric regression. The material depends very heavily on an appreciation of likelihood methods, and this edition has been expanded by the inclusion of much more detailed sections, even completely new chapters, on Bayesian MCMC techniques, longitudinal data and mixed models, score functions and survival analysis. The end result is an up-to-date rigorous treatmeat of the general ideas and methods of estimation and inference in difficult problems involving non-linear measurement error models.

Contents:
1. Introduction and overview
2. Population, sample, and sampling design
3. Sampling algorithms
4. Simple random sampling
5. Unequal probability exponential designs
6. The splitting method
7. More on unequal probability designs
8. Balanced sampling
9. An example of the cube method

This book gives a comprehensive overview of sampling designs and algorithms for implementing them. It starts with a succinct, mathematically rigorous, introduction to designs, defined as discrete multivariate probability distributions (on {0,1}N for sampling with replacement and on NN for sampling without replacement), and their properties. It then gives a systematic description of six basic classes of algorithms (enumerative, martingale, sequential, draw by draw, eliminatory, and rejective) that together cover most known methods. These methods are explored in more detail, first for simple random sampling and then for general exponential designs.
A whole chapter is devoted to the splitting method, originally developed by DeVille and the author. This is a way of generating new algorithms but, perhaps more importantly, it also provides an integrated framework that includes almost all existing methods and hence can be used to make comparisons among different methods. There are a few important algorithms, such as systematic sampling and the Sampford method, that are not special cases of the splitting method and not exponential designs and these are covered in the following chapter. Finally there is a chapter on algorithms for implementing balanced sampling.
The mathematical level required to follow some of the treatment will seem rather formidable for many survey samplers. However the effort will be worthwhile for anyone interested in the topics covered. As well as containing much new material, the book succeeds in providing a systematic development for what often seems a collection of isolated and ad hoc methods.

Contents:
1. Sampling methods
2. Weighting
3. Statistical effects of sampling and weighting
4. Significance testing
5. Measuring relationships between variables

APPENDIX A:
Review of General Terminology
APPENDIX B:
Further Reading
APPENDIX C:
Summary Tables for Several Common Distributions
APPENDIX D:
Chapter 2 Mathematical Proofs
APPENDIX E:
Chapter 3 Mathematical Proofs
APPENDIX F:
Chapter 4 Mathematical Proofs
APPENDIX G:
Chapter 5 Mathematical Proofs
APPENDIX H:
Statistical Tables

To avoid the risk of the contents list being misleading, I should remark that Chapters 1 to 5 of the book are the first 226 of the book's 266 pages.
Both authors of this book are experienced market and social researchers. Their aim is to provide an understanding of the practical aspects of sampling, beyond straightforward simple random samples and without delving deeply into mathematics. It is not intended to provide comprehensive coverage of all social survey issues, but to provide a reference source for the principles of sampling and evaluation of sample-based data, with emphasis on the practicalities of real value. The authors comment that professional statisticians working in the assessment of sampling and non-sampling survey error may find the discussion too superficial, but that it is primarily aimed at practising researchers 'who may have only basic formal statistical training'.
In their preface, the authors remark 'we have also included a short review of some of the more common analytical tools used with survey data. If we seem slightly less than enthusiastic in our endorsement of these it is because experience has taught us to be cautious in applying sophisticated mathematical procedures to data that may be considerably less sophisticated.' This is a sentiment I heartily endorse.
I think the authors have succeeded in their aims: this book will be valuable for analysts who are not survey sampling professionals or who have not been formally trained in the technical background, but who nevertheless find themselves having to analyse survey data.

Contents:
1. Introduction
2. European contingent claims
3. American contingent claims
4. Standard American options
5. Barrier and capped options
6. Options on multiple assets
7. Occupation time derivatives
8. Numerical methods

This research monograph gives an excellent summary of recent research on the American option problem. The book is very specialized and suitable for an advanced graduate course.
The subject matter at this level is unavoidably technical but the author maintains a good balance between rigor and intuition. The book concentrates on equity options rather than interest rate options, on pricing rather than hedging. Professor Detemple may not dig far but he digs deep.

Contents:
1. Introduction
2. An introduction to branch-and-bound methods for partitioning
3. Minimum-diameter partitioning
4. Minimum within-cluster sums of dissimilarites partitioning
5. Minimum within-cluster sums of squares partitioning
6. Multiobjective partitioning
7. Introduction to the branch-and-bound paradigm for seriation
8. Seriation - maximization of a dominance index
9. Seriation - maximization of gradient indices
10. Seriation - unidimensional scaling
11. Seriation - multiobjective seriation
12. Introduction to branch-and-bound methods for variable selection
13. Variable selection for cluster analysis
14. Variable selection for regression analysis

APPENDIX A: General Branch-and-Bound Algorithm for Partitioning
APPENDIX B: General Branch-and-Bound Algorithm Using Forward Branching for Optimal Seriation Procedures
APPENDIX C: Size Distributions

Readership: Data analysts, statistical psychologists, applied statisticians interested in cluster analysis and grouping, seriation and variable selection procedures

This book both summarizes and illustrates the methods that are currently available for applying branch-and-bound methods to certain data analysis situations.
Branch-and-bound is presented as an alternative to integer programming where a partial enumeration technique is needed for solving discrete optimization problems involving combinatorial choices, e.g. ordering or grouping objects, or selecting a subset of objects.
After a general introduction, the book is split into three parts, dealing with important applications in grouping, seriation and variable selection. Each part has its own introductory chapter.
The mathematical basis is clearly explained, pseudo-code for the algorithms is given where needed, very simple cases are used to demonstrate the methodology, and larger examples are also presented. The available software is discussed and information on obtaining it via the internet is provided. Care is taken to discuss the limitations as well as the advantages of the branch-and-bound approach.

Contents:
1. Discrete distributions
2. One-sided univariate analysis
3. Two-sided univariate analysis
4. Computing fundamental
5. Elements of conditional analysis
6. Two 2 x 2 tables
7. Assessing inference
8. Several 2 x 2 tables: I
9. Several 2 x 2 tables: II
10. The 2 x K table
11. Polynomial algorithms: I
12. Polynomial algorithms: II
13. Multinomial models
14. Matched and dependent data
15. Reflections on exactness

Readership: Researchers in biology, medicine, public health, psychology, sociology, law and economics

This book introduces the statistical theory, analysis methods and computational techniques for exact analysis of discrete data. Each chapter starts with its specific aims and has relevant and poignant examples to elucidate the theory. A long list of exercises that the student can attempt, is at the end of each chapter; however, there are no solutions given. A comprehensive reference list is at the end of the book. The book is not an introductory text but rather one that pulls together the latest research and techniques used in this field of study. Three of the chapters deal with computational issues, which is a pleasant change from the theoretical approach seen in most text books of this type.
This book would be useful to anyone who has completed a basic/intermediate course in statistics and who has an interest in this area but those who had completed a course in biostatics would find this relevant to their area of study.

Contents:
PART I: Fundamentals of MDS
1. The four purposes of multidimensional scaling
2. Constructing MDS representations
3. MDS models and measures-of-fit
4. Three applications of MDS
5. MDS and facet theory
6. How to obtain proximities
PART II: MDS Models and Solving MDS Problems
7. Matrix algebra for MDS
8. A majorization algorithm for solving MDS
9. Metric and nonmetric MDS
10. Confirmatory MDS
11. MDS fit measures, their relations and some algorithms
12. Classical scaling
13. Special solutions, degeneracies and local minima
PART III: Unfolding
14. Unfolding
15. Avoiding trivial solutions in unfolding
16. Special unfolding models
PART IV: MDS Geometry as a Substantive Model
17. MDS as a psychological model
18. Scalar products and Euclidean distances
19. Euclidean embeddings
PART V: MDS and Related Methods
20. Procrustes procedures
21. Three-way Procrustean models
22. Three-way MDS models
23. Modelling asymmetric data
24. Methods related to MDS
PART VI: Appendices
A. Computer programs for MDS
B. Notation

This is an updated and expanded version of the first edition [Short Book Reviews, Vol. 17, p. 25], and it has grown from 471 to 614 pages. Some 60 per cent of this increase is accounted for by addition of two extra chapters (15 and 23) and exercises at the end of each chapter, the remainder by an updating of the existing text and the addition of more recent computer programs to the survey in the Appendix. In my review of the first edition, I highlighted the comprehensiveness of the text and its appeal to beginner and expert alike, but commented that the last chapter was perhaps the least successful one. The new chapter 23 has now filled out the latter material substantially, the updating of the remaining chapters has reinforced the previous strengths, and the exercises at the end of each chapter are an attractive feature. I can recommend the book enthusiastically.

Contents:
PART I: Survey Error Theory
1. Scientific survey research
2. Survey error
3. Survey modes
PART II: Response Accuracy Issues
4. Measurement error due to interviewers
5. Measurement error due to respondents, I
6. Measurement error due to respondents, II
7. Nonresponse error at the item level
PART III: Respondent Selection Issues
8. Nonresponse error at the unit level
9. Coverage error
10. Sampling error
PART IV: Survey Administration Issues
11. Postsurvey error
12. Mode differences
13. Comparability effects
PART V: Total Survey Error
14. Ethics in surveys
15. Survey errors

The author is professor of political science and director of the Center for Survey Research at Ohio State University. The book is a very useful guide to current practices in the design and conduct of surveys. From the title one might expect a new unified framework for survey errors, with a quantitative treatment of the trade-offs between sampling and non-sampling errors. The author does not attempt this, but takes the total survey error approach to mean a systematic consideration of all sources of error, their impact on results and the costs and benefits of reducing the various components. The main message: there are ways to reduce non-sampling error, with hard work, careful planning and adequate resources. It is left to the judgement of the research team how far to go in trying to maximize response rates, for example.
The book is well written. Much of the material is standard, but there are also many new insights and observations. I found the sections on measurement error particularly interesting. Although there are very few formulae in the book, the author describes statistical issues, such as adjusting for non-response, with clarity and accuracy.

Contents:
1. Introduction to longitudinal data
2. Plots
3. Simple analyses
4. Critiques of simple analyses
5. The multivariate normal linear model
6. Tools and concepts
7. Specifying covariates
8. Modeling the covariance matrix
9. Random effects models
10. Residuals and case diagnostics
11. Discrete longitudinal data
12. Missing data
13. Analyzing two longitudinal variables
14. Further reading

Readership: Final year students in a biostatistics or statistics master's degree programme. Doctoral students in biostatistics or statistics, applied researchers, and quantitative doctoral students in disciplines such as medicine, public health, public policy, psychology, political science, biology, sociology and education

There are now quite a few books on longitudinal data and repeated measures analysis (including two co-authored by myself), so what is different about this one? Many of the others are monographs, whereas this being a textbook rather than a monograph it contains appropriate exercises for students. The mathematical level is slightly less advanced than several of the other books. There is an extensive discussion of graphical methods. The discipline also moves on, so that this more recent book can take advantage of new developments and new ways of presenting older material.
The author asserts that 'texts on longitudinal data from the 1980s and even 1990s [that is, texts from this period, not data from this period] are already out of date, usually concentrating on generalizations of analysis of variance rather than on generalizations of regression. The techniques they cover are often archaic'. While it is true that some early books concentrated on univariate and multivariate ANOVA approaches, many other books have taken the regression approach. Furthermore, to describe the ANOVA approach as out of date skips over the fact that such approaches are often well suited to the kinds of problems which arise in certain areas - for example, designed experiments in psychology.
I am also uneasy with the author's assertion that 'the value of longitudinal data analysis to the student will be much greater than the value of MANOVA or multivariate regression'. That, of course, depends on the sorts of applications and problems which the student will work on.
Despite these reservations, I do like the author's evident enthusiasm for his subject matter: 'I often think of this course as a "money course." Take this course, earn a living,' he says. The book is clearly written and well presented. The author's accumulated experience in presenting the material comes over. On balance, this is one of the books which anyone about to teach a practical course in longitudinal data analysis should consider adopting as the course text.

Contents:
PART I: Missing Data
1. Prologue
2. Describing incompleteness
3. Single imputation and related methods
4. Multiple imputation
5. Case studies
PART II: Small-Area Estimation
6. Introduction
7. Models for small areas
8. Using auxiliary information
9. Using small-area estimators
10. Case studies
PART III: Combining Estimators
11. Model selection

This book is largely based on Longford's published work on the two main topics of the book: missing data and small area estimation. Case studies presented in the book will be useful to the reader. Part I of the book gives an introduction to imputation methods for missing data and advocates the use of multiple imputation. Chapter 5 presents five case studies, including the UK Labour Force Survey. Issues related to "proper" multiple imputation in the context of complex survey samples and unplanned domains (small areas) are not adequately covered in the book. Also, under multiple imputation, confidence intervals are based on the t-distribution, unlike the normal intervals for some single imputation and fractional imputation methods, leading to intervals with larger average length especially for a small number of imputations for each item. Confidence interval issues are not covered in the book. Part II of the book deals with the important problem of small-area estimation. Chapter 10 presents three case studies of small area estimation, including the UK Labour Force Survey. Chapter 7 is a brief introduction to point estimation based on explicit small area linking models, and the reader is referred to Rao's (2003) book for "a comprehensive treatment of the problem from a committed model-based perspective". Model-based inference methods are criticized on the grounds that such inferences ignore the uncertainty associated with the model selection (Chapter 11). Longford claims that he has developed "an approach that relies on a "good" model much less than model-based methods do". However, this approach is similar to traditional composite estimation based on a weighted combination of an unbiased but unreliable direct estimator and a biased synthetic estimator that has small variance. Implicit models are used to determine the "optimal" weights, but the resulting composite estimators are very similar to those based on explicit models. A difference between the two approaches was noted in Section 8.2, p. 213, but the composite estimator here failed to make use of a more efficient direct estimator (sample regression estimator) unlike the model-based approach. The model-based approach can handle complex cases in a systematic manner and permits model validation, as demonstrated in Rao's (2003) book. Mean squared error (MSE) estimation under the model-based approach is typically unconditional, but MSE estimators conditional on the small area means have also been proposed in the literature. The latter estimators avoid the use of linking models, but can be highly unstable.

Contents:
1. Introduction
2. Simple random sampling
3. Sampling with unequal probabilities
4. Stratification
5. Multi-stage sampling
6. Calibration with an auxiliary variable
7. Calibration with several auxiliary variables
8. Variance estimation
9. Treatment of non-response

The exercises presented in the book were used as educational material at the École Nationale de la Statistique et de l'Analyse de l'Information (ENSAI). Each chapter begins with the essential theory required to solve the problems, and then the problems are presented. Each problem is followed directly by a solution. Most of the exercises are designed to elucidate the principles of classical sampling theory. They range from applications of very basic mean and variance calculations to fairly challenging theoretical questions, for example (Exercise 3.24) the determination of conditions under which the Hájek ratio is less efficient than the Horvitz-Thompson estimator. A super-population model is introduced in Chapter 5 for a problem on variance and list order, and another is mentioned in Chapter 9 for dealing with non-response. However, for the most part, models are not used, even in the discussion of regression. The basic problems would be suitable for assigning to senior undergraduates and beginning graduate students. The more advanced problems would be useful for training of practitioners of survey science.

Contents:
The General Theory of Causality
Basics
Fuzzy logic
Dialectical logic - Negation of fuzzy logic
Discrete distributions
Continuous distributions
Conditions
The Special Theory of Causality
Causality and statistics
The law of nature

This book puts forward a treatment of causality, which appears to be based on an idea of co-occurrence of events. The chapter on Basics introduces notions of probability, Bernoulli trials, and Boolean algebra. The chapter on Causality and statistics is the one in which the author elaborates on the nature of cause and effect. There are many misprints.

Contents:
1. Probability and random variables
2. Introduction to stochastic processes
3. Markov chains
4. Markov chains in continuous time
5. Diffusions
6. Hints and solutions for starred exercises and problems

Readership: Students of statistics, mathematics, finance and operational research taking an undergraduate second course in probability

This book introduces the main ideas, applications and methods of stochastic process modelling and problem-solving. The aim to do this simply and concisely means that full proofs and most general statements of results are not always offered, and measure theory is avoided altogether. Nevertheless, it goes beyond the random walks, renewals and Markov chains to introduce, albeit briefly, simulation (Metropolis, Gibbs), diffusions and stochastic integrals.
Its concision means that it will most easily be appreciated by those already happy with mathematical notation and methods. I found myself entertained by the novelty of many of the examples, exercises and problems, despite my familiarity with most of the books cited for further reading.

Contents:
1. Introduction
2. Estimating the CDF and statistical fundamentals
3. The bootstrap and the jackknife
4. Smoothing: General concepts
5. Nonparametric regression
6. Density estimation
7. Normal means and minimax theory
8. Nonparametric inference using orthogonal functions
9. Wavelets and other adaptive methods
10. Other topics
Bibliography (195 references)

Readership: MS or Ph.D. statistics and computer science students; researchers wanting the basic concepts

This book brings to mind an old music hall joke:
DOCTOR: "Have you had this before?"
PATIENT: "Yes, doctor."
DOCTOR: (Triumphantly) "Well you've got it again, lad."
Last time, the "disease" involved getting "All of Statistics" into 442 pages; see Short Book Reviews, Vol. 24, p. 25. Now we have All of Nonparametric Statistics in only 268 pages, five of which are blank! Again, the writing is excellent and the author is to be congratulated on the clarity achieved. l wonder also whether the book will be adopted by instructors, who will have to demonstrate a wide-ranging versatility to add the right details. l hope so; the book is excellent.

Contents:
1. Introduction
2. Decision theory
3. Bayesian methods
4. Hypothesis testing
5. Special models
6. Sufficiency and completeness
7. Two-sided tests and conditional inference
8. Likelihood theory
9. Higher-order theory
10. Predictive inference
11. Bootstrap methods

Based on material used for teaching students at Cambridge University, this book gives a clear and comprehensive account of the basic elements of statistical theory. It should make a good text for an advanced course on statistical inference. As well as traditional topics in decision theory, estimation and hypothesis testing, it treats Markov chain Monte Carlo methods (briefly), conditional inference and modified likelihood, predictive inference, and the bootstrap. Students will find it informative and challenging.

Contents:
1. Divergence measures: Definition and properties
2. Entropy as a measure of diversity: Sampling distributions
3. Goodness-of-fit: Simple null hypothesis
4. Optimality of phi-divergence test statistics in goodness-of-fit
5. Minimum phi-divergence estimators
6. Goodness-of-fit: Composite null hypothesis
7. Testing loglinear models using phi-divergence test statistics
8. Phi-divergence measures in contingency tables
9. Testing in general populations

There are a number of measures of divergence between distributions. Describing them properly requires a very mathematically well-written book, which the author here provides. Exercises appear in all nine chapters; each exercise section is followed by a corresponding answers section. Excellent! This book is a fine course text, and is beautifully produced. There are about four hundred references. Recommended.

Contents:
1. Foundations
2. Limit theorems
3. Characterization
4. Interacting particles
5. Arithmetical functions
6. Miscellaneous results

Readership: Researchers in probability, in theoretical physics and in number theory

Products of random variables arise frequently enough that one would expect there to be a reasonable number of books on properties of such quantities. It is surprising that only now is there available a book in English devoted to the subject. Lithuanians however did not have to wait quite so long; a book by G. Bareikis and J. Siaulys on the subject was published in 1998 but has yet to be translated. The current book, according to the authors, has little overlap with the Lithuanian account.
A reflex reaction to a book on products might be: why bother when you can just take logs and reduce it to a problem on sums of variables. This is of course legitimate for positive random variables but not for variables that can take both positive and negative values. Such variables are handled via a generalization of the Mellin transform (a close relative of the characteristic function for log|X|), essentially an ordered pair of Mellin transforms, one for the positive part and one for the negative part. The question of whether for sequences of products, the limiting variable is positive or negative introduces some interesting combinatorial issues.
The theory of products finds application in a broad spectrum of problems: in number theory, in interacting particle systems and in characterization problems. The account is clear and accessible.

Contents:
1. An introduction to SAS
2. Describing and summarizing data
3. Basic inference
4. Scatterplots, correlation, simple regression, and smoothing
5. Analysis of variance and covariance
6. Multiple regression
7. Logistic regression
8. The generalized linear model
9. Generalized additive modeIs
10. Nonlinear regression modeIs
11. The analysis of longitudinal data I
12. The analysis of longitudinal data II: Models for normal response variables
13. The analysis of longitudinal data III: Non-normal responses
14. Survival analysis
15. Analysing multivariate data: Principal components and cluster analysis

Readership: Students of statistics, medical statisticians and applied researchers in medicine.

This book offers an easy-to-read coverage of the uses and interpretation of basic statistical methods as applied to medical data, covering both the general linear model (including analysis of variance, covariance and regression) and the generalized additive model. The methods are presented in the context of the software package SAS (Statistical Analysis System) version 9.1 for Windows and the text proceeds through the analysis of a variety of sets of data (each accessible from the Web), interspersed throughout the text with illustrative SAS code (also available from the Web). The main features of the basic system, known as BASE SAS, are presented in the excellent introductory first chapter, and the authors concentrate on the SAS/STAT and SAS/GRAPH modules throughout the remainder of the text.
The mathematical prerequisites for using this book are minimal (appreciation of mathematical formulae and graphs) and calculus is not required. Some familiarity with using a computer is necessary in order to gain the most benefit from the text, and some previous experience of using a statistical software package would be advantageous. The text is suitable for private study.

Contents:
1. Introduction and history of clinical trials
2. The basis of statistical reasoning in medicine
3. Probability tools for monitoring rules
4. Issues and institutions in path analysis
5. Group sequential analysis procedures
6. Looking forward: Conditional power
7. Safety and futility
8. Bayesian statistical monitoring

APPENDICES: Boundary Values for Normal Mean
Conditional Brownian Motion
Boundary Values and Conditional Power
Supporting Bayesian Computations
Standard Normal Probabilities

Readership: Statisticians in clinical research, clinical researchers, biostatisticians

This book is aimed at helping clinical researchers, who have little or no quantitative background and who have difficulty in communicating with biostatisticians or experienced trial methodologists. The author states "If you know nothing about clinical trials, then this book is for you". I would suggest that at least an elementary understanding of clinical trials and knowledge of basic statistics would be useful to the reader before reading this text. The book does not have to be read in order but delved into as required. For those with sufficient experience Chapter 1 could be easily skipped.
The author has a wealth of experience in this area and this is demonstrated throughout the text with relevant poignant examples. Each chapter has a comprehensive reference list and a set of problems for the reader to attempt, however, there are no solutions provided.
A book to recommend for students/researchers who need to be able to apply correct use and monitor procedures for clinical investigations.

Contents:
Section 1: Introduction/Overview
1. Monitoring committees: Why and how
2. Lessons learned
3. FDA and clinical trial data monitoring committees
Section 2: General Benefit (includes 11 case studies)
Section 3: General Harm (includes 9 case studies)
Section 4: Special Issues (includes 9 case studies)

APPENDIX 1: Data Monitoring Committee Members
APPENDIX 2: Case Study Acronym Key (Title)

Readership: Students and practitioners of clinical trials, and members of data-monitoring committees

This edited book gives a brief introduction to data-monitoring committees for clinical trials, and then provides twenty-nine case studies of trials where monitoring committees were used. The case studies were selected to provide examples of the types of challenges faced by data-monitoring committees during their deliberations. It has been my experience that many persons asked to serve on data-monitoring committees have scientific or clinical expertise relevant to the trial, but have little or no understanding of the purpose of the data-monitoring committee and challenges that can arise during the monitoring of a trial. Such individuals would greatly benefit from reading this book prior to serving on a data-monitoring committee, as would anyone wishing to gain some insight into the responsibilities and challenges facing data-monitoring committees.

Contents:
1. Publication bias in meta-analysis
PART A: Publication Bias in Context
2. Publication bias: Recognising the problem, understanding its origins and scope, and preventing harm
3. Preventing publication bias: Registries and prospective meta-analysis
4. Grey literature and systematic reviews
PART B: Statistical Methods for Assessing Publication Bias
5. The funnel plot
6. Regression methods to detect publication and other bias in meta-analysis
7. Failsafe N or file-drawer number
8. The trim and fill method
9. Selection method approaches
10. Evidence concerning the consequences of publication and related biases
11. Software for publication bias
PART C: Advanced and Emerging Approaches
12. Bias in meta-analysis induced by incompletely reported studies
13. Assessing the evolution of effect sizes over time
14. Do systematic reviews based on individual patient data offer a means of circumventing biases associated with trial publications?
15. Differentiating biases from genuine heterogeneity: Distinguishing artifactual from substantive effects
16. Beyond conventional publication bias: Other determinants of data suppression
APPENDIX A: Data Sets
APPENDIX B: Annotated Bibliography

Readership: Researchers or graduate students conducting systematic reviews or meta-analyses

Publication bias is the term for the selection process in which studies yielding positive results are more likely to be published than studies yielding negative results, so that an overall analysis of published results may draw biased conclusions. In the first chapter of the book, the editors point out that there are, in fact, many potential similar information suppression mechanisms, and perhaps a better term would be dissemination bias.
This book is a valuable resource for anyone concerned with meta-analysis. It presents an up-to-date accessible and comprehensive overview of the issues. It discusses strategies for preventing such bias from arising in the first place, tools for detecting when such bias has occurred, and methods for adjusting estimates for such bias.
Edited collections can often be rather heterogeneous in style and quality of the contributions. This book is something of an exception, with all of the contributions being well written and giving valuable insights. The editors are to be highly commended for their work. I predict that this book will become the key text in the area. I highly recommend it to anyone undertaking or thinking of undertaking a meta-analysis, and more generally to researchers who have to contend with problems of selectivity bias.

Contents:
PART I: Introduction to the Four Themes
1. Statistics
2. Computation
3. Algebra
4. Biology
PART II: Studies on the Four Themes
5. Parametric inference
6. Polytope propagation on graphs
7. Parametric sequence alignment
8. Bounds for optimal sequence alignment
9. Inference functions
10. Geometry of Markov chains
11. Equations defining hidden Markov models
12. The EM algorithm for hidden Markov models
13. Homology mapping with Markov random fields
14. Mutagenetic tree models
15. Catalog of small trees
16. The strand symmetric model
17. Extending tree models to split networks
18. Small trees and generalized neighbor-joining
19. Tree construction using singular value decomposition
20. Application of interval methods to phylogenetics
21. Analysis of point mutations in vertebrate genomes
22. Ultra-conserved elements in vertebrate and fly genomes

Readership: Researchers in algebraic statistics and Markov modelling for computational
biology

This substantial, enthusiastically presented, and confidently written book is largely based on and around a graduate course taught by the two editors, who are in the mathematics department at the University of California, Berkeley, during the fall of 2004. The four introductory chapters were written by the editors, whilst the seventeen chapters in Part II were written up as a result of research projects undertaken by course participants. The chapter on Statistics is written to translate selected aspects of inference for discrete models into an algebraic setting, rather than the reverse. For example, although the solution to a set of (likelihood) equations can happily be quoted to an arbitrary number of significant figures if you are an algebraist, a statistician will surely wince at ten significant figures when the sample size is 40 or 49. Neither standard error nor variance are in the index. Tail-area probabilities to explain the statistical significance of results are used in one of the research chapters. The chapter on Biology explains DNA structure and demonstrates how maximum likelihood applied to hidden Markov models and continuous time Markov processes on trees has proved to be extremely useful for the biological analysis and understanding of the enormous and complex database now available for DNA. The chapters in Part II develop particular aspects of this general approach, for example a computationally oriented interval method for identifying a region enclosing the true maximum likelihood estimate, and an assessment of evolution in Drosophila utilizing descriptive statistics and probability models for phylogenetic trees.

Reviewer:
Institute	University of Toronto
Place	Toronto, Canada
Name	D.F. Andrews

Reviewer:
Institute	University of Waterloo
Place	Waterloo, Canada
Name	T.S. Wirjanto

Reviewer:
Institute	University of Waterloo
Place	Waterloo, Canada
Name	D.L. McLeish

Reviewer:
Institute	Shandong University
Place	Shandong, China
Name	S. Peng

Reviewer:
Institute	Julius Baer Investment Funds & Services Ltd.
Place	Zürich, Switzerland
Name	A.-M. Matache

Title	COMPUTATIONAL GENOME ANALYSIS. An Introduction.
Author	R.C. Deonier, S. Tavaré and M.S. Waterman.
Publisher	New York: Springer-Verlag, 2005, pp. xx + 534, US$79.95.

Reviewer:
Institute	Imperial College of Science, Technology and Medicine
Place	London, U.K.
Name	D.J. Hand

Reviewer:
Institute	University of Newcastle
Place	Newcastle, Australia
Name	G.C. Goodwin

Reviewer:
Institute	London School of Economics
Place	London, U.K.
Name	S. Powell

Reviewer:
Institute	University of Kent
Place	Canterbury, U.K.
Name	B.J.T. Morgan

Reviewer:
Institute	MRC Biostatistics Unit
Place	Cambridge, U.K.
Name	V.T. Farewell

Reviewer:
Institute	Queen's University
Place	Kingston, Canada
Name	M.R. Murty

Reviewer:
Institute	CEFAS Lowestoft Laboratory
Place	Lowestoft, U.K.
Name	C.M. O'Brien

Reviewer:
Institute	London South Bank University
Place	London, U.K.
Name	S. Starkings

Reviewer:
Institute	GlaxoSmithKline
Place	Collegeville, U.S.A.
Name	S. Leonov

Reviewer:
Institute	University of Wisconsin
Place	Madison, U.S.A.
Name	N.R. Draper

Reviewer:
Institute	University of Bristol
Place	Bristol, U.K.
Name	H. Goldstein

Reviewer:
Institute	Universiteit Hasselt
Place	Diepenbeek, Belgium
Name	N.D.C. Veraverbeke

Reviewer:
Institute	Berlin University of Technology
Place	Berlin, Germany
Name	A. Schied

Reviewer:
Institute	University of Warwick
Place	Coventry, U.K.
Name	A. Dias

Reviewer:
Institute	Harvard University
Place	Cambridge, U.S.A.
Name	S.W. Lagakos

Reviewer:
Institute	University of Rochester
Place	Rochester, U.S.A.
Name	D. Oakes

Reviewer:
Institute	University of Michigan
Place	Ann Arbor, U.S.A.
Name	R.J.A. Little

Reviewer:
Institute	Nuffield College
Place	Oxford, U.K.
Name	D.R. Cox

Reviewer:
Institute	University of Florida
Place	Gainesville, U.S.A.
Name	A. Agresti

Reviewer:
Institute	University of Southampton
Place	Southampton, U.K.
Name	P. Prescott

Reviewer:
Institute	University of Auckland
Place	Auckland, N.Z.
Name	A.J. Scott

Reviewer:
Institute	University of St Andrews
Place	St Andrews, U.K.
Name	C.D. Kemp

Reviewer:
Institute	University of Exeter
Place	Exeter, U.K.
Name	W.J. Krzanowski

Reviewer:
Institute	Carleton University
Place	Ottawa, Canada
Name	J.N.K. Rao

Reviewer:
Institute	University College London
Place	London, U.K.
Name	A.P. Dawid

Reviewer:
Institute	Macquarie University
Place	Sydney, Australia
Name	J.R. Leslie

Reviewer:
Institute	University of Manchester
Place	Manchester, U.K.
Name	P.J. Laycock