Short Book Reviews On-Line 2003

UNE HISTOIRE DE LA COMPTABILITÉ NATIONALE.	A.Vanoli.
WRITING THE HISTORY OF MATHEMATICS: ITS HISTORICAL DEVELOPMENT.	J .W .Dauben and C.J. Scriba (Eds.).
STATISTICS WITH APPLICATIONS IN BIOLOGY AND GEOLOGY.	P. Blaesild and J. Granfeldt.
ORDINAL MEASUREMENT IN THE BEHAVIOURAL SCIENCES.	N. Cliff and J.A. Keats.
THE STATISTICAL ANALYSIS OF FAILURE TIME DATA, 2nd edition.	J.D. Kalbfleisch and R.L. Prentice.
STATlSTICAL MODELS AND METHODS FOR LIFETIME DATA, 2nd edition.	J.F. Lawless.
RECURRENT EVENTS DATA ANALYSIS FOR PRODUCT REPAIRS, DISEASE RECURRENCES AND OTHER APPLICATIONS.	W.B. Nelson.
MODERN MEDICAL STATISTICS. A PRACTICAL GUIDE.	B.S. Everitt.
THIELE: PIONEER IN STATISTICS.	S.L. Lauritzen.
THE STATISTICAL MIND IN A PRE-STATISTICAL ERA: THE NETHERLANDS 1750-1850.	P.M.M. Klep and I.H. Stamhuis (Eds.).
COMPONENTS OF VARIANCE.	D.R. Cox and P.J.Solomon.
PARAMETRIC AND NONPARAMETRIC INFERENCE FROM RECORD-BREAKING DATA.	S. Gulati and W.J. Padgett.
BAYESIAN METHODS: A SOCIAL AND BEHAVIORAL SCIENCES APPROACH.	J. Gill.
THE OPTIMAL DESIGN OF BLOCKED AND SPLIT-PLOT EXPERIMENTS.	P. Goos.
BLOCK DESIGNS: A RANDOMIZATION APPROACH. Volume II: Design.	T. Calinski and S. Kageyama.
A DISTRIBUTION-FREE THEORY OF NONPARAMETRIC REGRESSION.	L. Györfi, M. Kohler, A. Krzyi¿ak and H. Walk.
PRINCIPAL COMPONENT ANALYSIS, 2nd edition.	I.T. Jolliffe.
APPLIED MULTIVARIATE ANALYSIS.	N.H. Timm.
MULTILEVEL STATISTICAL MODELS, 3rd edition.	H. Goldstein.
DATA MINING USING SAS APPLICATIONS.	G. Fernandez.
UNIFIED METHODS FOR CENSORED LONGITUDINAL DATA AND CAUSALITY.	M.J. van der Laan and J.M. Robins.
THE HANDBOOK OF DATA MINING.	N. Ye.
LIKELIHOOD, BAYESIAN, AND MCMC METHODS IN QUANTITATIVE GENETICS.	D. Sorensen and D. Gianola.
STATISTICAL ANALYSIS OF GENE EXPRESSION MICROARRAY DATA.	T. Speed (Ed.).
THE ANALYSIS OF GENE EXPRESSION DATA.	G. Parmigiani, E.S. Garrett, R.A. Irizarry, and S.L. Zeger (Eds.).
AN INTRODUCTION TO THE THEORY OF POINT PROCESSES. Volume l. Elementary Theory and Methods, 2nd edition.	D.J. Daley and D. Vere-Jones.
BRANCHING PROCESSES IN BIOLOGY.	M. Kimmel and D.E. Axelrod.
STATISTICAL ANALYSIS OF SPATIAL POINT PATTERNS, 2nd edition.	PJ. Diggle.
STATISTICS ON SPATIAL MANIFOLDS.	Y. Chikuse.
MARKOV CHAINS AND INVARlANT PROBABILITIES.	O. Hernández-Lerma and J.B. Lasserre.
APPLIED DERIVATIVES: OPTIONS, FUTURES AND SWAPS.	R.J. Rendleman, Jr.. Malder.
DATA MONITORING COMMITTEES IN CLINICAL TRIALS. A PRACTICAL PERSPECTIVE.	S.S. Ellenberg. T.R. Fleming and D.L. DeMets.
META-ANALYSIS OF CONTROLLED CLINICAL TRIALS.	A. Whitehead,
STATISTICAL METHODS IN AGRICULTURE AND EXPERIMENTAL BIOLOGY, 3rd edition.	R. Mead, R.N. Curnow and A. M. Hasted.
THE OXFORD DICTIONARY OF STATISTICAL TERMS.	Y. Dodge (Ed.).
STOCHASTIC MUSINGS. Perspectives from the Pioneers of the Late 20th Century.	J. Panaretos (Ed.).
THE GREATE INVENTION OF ALGEBRA, THOMAS HARRIOT'S TREATISE ON EQUATIONS.	J.A. StedaIl.
ELEMENTARE GRUNDBEGRIFFE ElNER ALLGEMEINEREN WAHRSCHEINLICHKEITSRECHNUNG (In German).	K. Weichselberger. With the assistance of T. Augustin and A. Wallner.
STOCHASTIC CALCULUS. APPLICATIONS IN SCIENCE AND ENGINEERING.	M. Grigoriu.
ELEMENTARY PROBABILITY THEORY. With Stochastic Processes and an Introduction to Mathematical Finance,4th edition.	K.L. Chung and F. AitSahlia.
ELEMENTS OF QUEUEING THEORY. PALM MARTINGALE CALCULUS AND STOCHASTIC RECURRENCES, 2nd edition.	F. Baccelli and P. Brémaud.
RANDOM NUMBER GENERATION AND MONTE CARLO METHODS, 2nd edition.	J.E. Gentle.
STATISTICAL METHODS IN DIAGNOSTIC MEDICINE.	X.H. Zhou, N.A. Obuchowski and D.K. McClish.
DESIGN AND ANALYSIS OF QUALITY OF LIFE STUDIES IN CLINICAL TRIALS.	D.L. Fairclough.
GROWTH CURVE MODELS AND STATISTICAL DIAGNOSTICS.	J.-X. Pan and K.-T. Fang.
STATISTICAL ANALYSIS OF DESIGNED EXPERIMENTS, 2nd edition.	H. Toutenburg.
CATEGORICAL DATA ANALYSIS WITH SAS® AND SPSS APPLICATIONS.	B. Lawal.
ANALYZING CATEGORICAL DATA.	J.S. Simonoff.
MODELLING BINARY DATA, 2nd edition.	D. Collett.
SMALL AREA ESTIMATION.	J.N.K. Rao.
TESTING STATISTICAL HYPOTHESES OF EQUIVALENCE.	S. Wellek.
STATISTICAL DATA MINING AND KNOWLEDGE DISCOVERY.	H. Bozdogan (Ed.).
DESIGNING EXPERIMENTS AND ANALYZING DATA: A Model Comparison Perspective, 2nd edition.	S.E. Maxwell and H.D. Delaney.
DESIGN AND ANALYSIS OF CROSS-OVER TRIALS, 2nd edition.	B. Jones and M.G. Kenward.
STATISTICAL PROCESS CONTROL: THE DEMING PARADIGM AND BEYOND, 2nd edition.	J.R. Thompson and J. Koronacki.
DATA ANALYSIS TOOLS FOR DNA MICROARRAYS.	S. Drãghici.
MATHEMATICAL BIOLOGY. Volume I: An Introduction; Volume II: Spatial Models and Biomedical Applications,	J.D. Murray.
APPLIED BAYESIAN MODELLING.	P. Congdon.
PARTIAL IDENTIFICATION OF PROBABILITY DISTRIBUTIONS.	C.F. Manski.
APPLIED PROBABILITY AND QUEUES, 2nd edition.	S. Asmussen.
SELFSIMILAR PROCESSES.	P. Embrechts and M. Maejima.
APPLIED TIME SERIES MODELLING AND FORECASTING.	R. Harris and R. Sollis.
GENERALIZED POISSON MODELS AND THEIR APPLICATIONS IN INSURANCE AND FINANCE.	V.E. Bening and V.Y. Korolev.
MODELING FINANCIAL TIME SERIES WITH S-PLUS®.	E. Zivot and J. Wang.
FINANCIAL MARKETS IN CONTINUOUS TIME.	R.A. Dana and M. Jeanblanc.
ADVANCES IN FINANCE AND STOCHASTICS: ESSAYS IN HONOUR OF DIETER SONDERMANN.	K. Sandman and P. Schönbucher (Eds.).
A. STOCHASTIC CONTROL FRAMEWORK FOR REAL OPTIONS IN STRATEGIC VALUATION.	A. Vollert.
TWO-SCALE STOCHASTIC SYSTEMS: ASYMPTOTIC ANALYSIS AND CONTROL.	Y. Kabanov and S. Pergamenshahikou.
BAYESIAN FIELD THEORY.	J.C. Lemm.
ESTIMATING ANIMAL ABUNDANCE: CLOSED POPULATIONS.	D.L. Borchers, S.T. Buckland and W. Zucchini.
ACTIVEPI.	D.G. Kleinbaum.
ANALYSIS OF FAILURE AND SURVIVAL DATA.	P.J. Smith.
ELEMENTS OF COMPUTATIONAL STATISTICS.	J.E. Gentle.
AMPL, A MODELLING LANGUAGE FOR MATHEMATICAL PROGRAMMING.	R. Foutor, D.M. Gay and B.W. Kernighan.
SUBSET SELECTION IN REGRESSION, 2nd edition.	A.J. Miller.
APPLIED FUNCTIONAL DATA ANALYSIS METHODS AND CASE STUDIES.	J.O. Ramsay and B.W. Silverman.
PREDICTIONS IN TIME SERIES USING REGRESSION MODELS.	F. Stulajter.
REGRESSION MODELS FOR TIME SERIES ANALYSIS.	B. Kedem and K. Fokianos.
DETERMINISTIC AND STOCHASTIC TIME DELAY SYSTEMS.	E. Boukas and Z.-K. Liu.
GRAPH COLOURING AND THE PROBABILISTIC METHOD.	M. Molly and B. Reed.
GREEN, BROWN, AND PROBABILITY AND BROWNIAN MOTION ON THE LINE.	K.L. Chung.
TOOLS FOR COMPUTIONAL FINANCE.	R. Seydel.
COMPUTATIONAL FINANCIAL MATHEMATICS USING MATHEMATICA®.	S. Stojanovic.
STOCHASTIC PORTFOLIO THEORY.	E.R. Fernholz.
INTEREST-RATE MANAGEMENT.	R. Zagst.
CREDIT SCORING AND ITS APPLICATIONS.	L.C. Thomas, D.B. Edelman and J.N. Crook.
UNCERTAIN VOLATILITY MODELS – THEORY AND APPLICATION.	R. Butff.

Contenu:
Introduction
PARTIE I: Naissance
1. De I'estimation du revenu national à la construction des comptes de la nation
PARTIE II: Systèmes et harmonisation internationale
2. La comptabilité nationale française fait d'abord cavalier seul
3. Progrès de I'harmonisation internationale des cadres comptables
4. Tendance à I'unification, et problèmes comptables persistants
PARTIE III: Synthèse statistique
5. La comptabilité nationale comme synthèse statistique
PARTIE IV: Concepts et théorie économique
6. Production, valeur, bien-être - tensions autour des activités des administrations
7. Production, valeur, bien-être - comptabilité nationale et bien-être
8. Production, revenu, patrimoine
9. Valeur, volume, prix
PARTIE V: Politique
10. Utilisations et statut de la comptabilité nationale
11. Perspective générale - destin d'un grand projet
12. Postface (à la première personne du singulier)

Lectorat: Comptables nationaux, économistes, universitaires et statisticiens intéressés par des thèmes économiques
Avec cet imposant ouvrage, le lecteur retrouvera d'avantage qu'une simple description de I'histoire de la comptabilité nationale. Il y découvrira notamment une analyse des relations entre la théorie économique et la comptabilité nationale, une discussion des questions ou enjeux qui ont marqué et qui marquent toujours la discipline ainsi qu'une discussion des concepts couramment traités.
Bien que vus par un expert français et abordant spécifiquement à plusieurs reprises les développements dans ce pays, I'histoire et les thèmes y sont néanmoins présentés et analysés sous une dimension internationale, juxtaposant les développements et visions de plusieurs pays et organismes mondiaux.
Par ailleurs, ayant été lui-même un acteur important de I'histoire de la comptabilité nationale française ainsi que celle du manuel du SCN93, I'auteur est en mesure de fournir un éclairage intéressant sur la pensée économique sous-jacente et le processus de décision émanant de tels travaux.
La section ‹‹perspective›› que I'on retrouve à la fin de chaque chapitre combine, comme son nom I'indique, conclusion, point de vue et mise en relation avec d'autres sujets. Par ailleurs, avec ses références bibliographiques choisies et commentées, I'auteur s'est fort habilement éloigné des traditionnelles longues listes d'auteurs et de documents.

Contents:
Introduction
Glossary
PART I: Countries
1. France, J. Peiffer
2. Benelux, P. Bockstaele
3. Italy, U. Bottazzini
4. Switzerland, E. Neuenschwander
5. Germany, M. Folkerts, C.J. Scriba and H. Wussing
6. Scandinavia, K. Andersen
7. The British Isles, I. Grattan-Guinness
8. Russia and the USSR, S. Demidov
9. Poland, S. Domoradzki and Z. Pawlikowska-Brozek
10. Bohemian Countries, L. Novy
11. Austria, C. Binder
12. Greece, C. Phili
13. Spain, E. Ausejo and M. Hormigón
14. Portugal, L.M.R. Saraiva
15. The Americas, U. D'Ambrosio, A.R. Garciadiego, J.W. Dauben and C.G. Fraser
16. Japan, Sasaki Chikara
17. China, L. Dun and J.W. Dauben
18. India, R.C. Gupta
19. Arab Countries, Turkey and Iran, S. Brentjes
20. Postscriptum, J.W. Dauben, J. Peiffer, C.J. Scriba and Hans Wussing
PART II: Portraits and Biographies
PART III: Abbreviations, Bibliography and Index

Readership: Historians and mathematicians
This book is about the writing of the history of mathematics. With more than forty historians of mathematics from many countries contributing to the volume, the central question it addresses is how did the writing of the history of mathematics develop in each of the countries. This is of course a difficult task. However, the book does make a valuable beginning in this direction and offers the scholar a starting point for such a study. Often, writing of this kind requires an extensive knowledge of both mathematics and history. Indeed, in recent times, many prominent mathematicians such as Jean Dieudonné, A.N. Kolmogorov, Dirk Struik, B.L. van der Waerden and André Weil have contributed to such a study. On the other hand, there have been exclusive historians who focussed on the historical development of mathematics such as 0. Neugebauer, J. Hoffman and A. Youshkevich.
The book has two parts. The first part traces the writing of the history of mathematics as it developed in each country. The second part has an alphabetical listing and short biographies of the historians of mathematics. The book is a welcome addition to the meager literature dealing with this theme.

Contents:
1. Statistical analysis
2. Preliminary Investigations
3. Normal data
4. Linear normal models
5. An introduction to the power of tests and design of experiments
6. Correlation
7. The multinomial distribution
8. The Poisson distribution
9. Generalized linear models
10. Models for directional data
11. The likelihood method
12. Some nonparametric tests

Appendix A: Simulated fractile diagrams
Appendix B: The Newton-Raphson procedure

Readership: Students in biology and geology (and areas with similar problems) who need statistics
This paperback volume grew out of a course presented since 1993 to students at the University of Aarhus, Denmark by two faculty members of the Department of Mathematical Sciences. It is a very nicely written text and is very enjoyable to read; it evidently provides the statistical materials these students need in their professional lives ahead. I note the following points; these will be advantages or drawbacks, depending on the reader's personal situation.
1. There is a lot of mathematical notation because students have already had "probability theory and a course in mathematics."
2. "SAS has been chosen... to illustrate all the statistical tools described... for the simplest models the calculations must be made... on a pocket calculator as well."
3. The topics are oriented to the disciplines in the title. Chapters 9-11 would be especially tough for a student who is not well grounded in the prerequisites.
4. There are lots of examples of data, printouts and diagrams.

Contents:
1. The purpose of psychological assessment
2. What makes a variable a scale?
3. Types of assessment
4. Item scores and their addition to obtain total test scores in the case of dichotomous items
5. Item scores and their addition to obtain total test scores in the case of polytomous items
6. Dominance analysis of tests
7. Approaches to ordering things and stimuli
8. Alternatives to complete paired comparisons
9. The unfolding model
10. The application of ordinal test theory to items in tests used in cross-cultural comparisons

Appendix A: Flow chart for a program to carry out a complete item analysis of items in a test or scale using a small personal computer
Appendix B: Statistical tables

Readership: Advanced students, researchers, and practitioners concerned with psychological measurement, as well as measurement professionals
At the elementary level, most psychological data consist of dichotomous or ordinal responses. Elaborate measurement tools, based on such theories as classical test theory and item response theory, have been developed to translate these low measurement level observations into higher level values via sophisticated latent variable type models. The authors argue that such theories and tools have weaknesses, and that, as a consequence, unjustified inferences are often drawn. This book seeks to describe alternative approaches which do not require more than ordinality of the variables. The book aims to step back from the model-driven perspective which underlies almost all of modern statistics, to adopt a more elementary empiricist perspective.
Unfortunately I do not think it always achieves this aim. For example, combining multiple dichotomous items to yield a single score necessarily involves a model of how those items are related to the underlying variable. The fact that this book combines such variabIes without explicitly stating a model, but expressing it in a concealed way via an algorithm, does not mean that such a model does not exist. The latent variable models in common use may well have weaknesses, but alternatives which do not make their modelling nature explicit are likely to be subject to similar weaknesses.
The book has a number of misprints which should have been picked up by the copy editor. A sentence referring to formula 1.2 but stating 'formula 1.1' and then giving the formula p = exp(x-a) as the logistic function would not be useful for the uninitiated.
Despite all that, the book does contain stimulating material and is a welcome addition to the literature on the foundations of measurement in psychology.

Contents:
1. Introduction
2. Failure time models
3. Inference in parametric models and related topics
4. Relative Risk (Cox) regression models
5. Counting processes and asymptotic theory
6. Likelihood construction and further results
7. Rank regression and the accelerated failure time model
8. Competing risks and multistate models
9. Modeling and analysis of recurrent event data
10. Analysis of.correlated failure time data
11. Additional failure time data topics

Readership: Graduate students and research workers in biostatistics and statistics
The first edition contained a distillation and integration of the literature on failure time data up to 1980. This volume presents an equally successful integration of that material with developments in the intervening years. The style of the first edition is retained for the first four chapters. Thereafter, new inference results flowing from advances in counting-process methods and martingale convergence are brought in as are extensions of models and methods. The use of bibliographic notes has been continued, and exercises have been added at the ends of chapters. A review of the first edition [Short Book Reviews Vol 0, p. 2], my first contribution to Short Book Reviews and the first under the current editorship, stated: "This book should become a standard reference in the field." In view of the undeniable accuracy of that prediction, need I say more?

Contents:
1. Basic concepts and models
2. Observation schemes, censoring, likelihood
3. Some nonparametric and graphical procedures
4. Inference procedures for parametric models
5. Inference procedures for log-location-scale distributions
6. Parametric regression models
7. Semiparametric multiplicative hazards regression models
8. Rank-type and other semiparametric procedures for log-location-scale models
9. Multiple modes of failure
10. Goodness-of-fit tests
11. Beyond univariate survival analysis

Appendix A: Glossary of Notation and Abbreviations
Appendix B: Asymptotic Variance Formulas, Gamma Functions, and Order Statistics
Appendix C: Large-Sample Theory for Likelihood and Estimating Function Methods
Appendix D: Computational Methods and Simulations
Appendix E: Inference in Location-Scale Parameter Models
Appendix F: Martingales and Counting Processes
Appendix G: Data Sets

Primary Readership: Statisticians
This text represents an updated version of the popular text originally published by the author in 1981 [Short Book Reviews Vol. 2, p. 14]. Much of the basic structure of the original version has been retained, but several new items have been added to reflect changes in the field over the past two decades. In particular, there is a greater emphasis on graphical methods, counting processes and martingale approaches, methods for other types of censoring than right-censoring, methods for truncated data, and methods for event history data.
This excellent book will serve as either a reference or a graduate-Ievel textbook.

Contents:
1. Recurrent events data and applications
2. Population model, MCF and basic concepts
3. MCF estimates for exact age data
4. MCF confidence limits for exact age data
5. MCF estimate and limits for interval age data
6. Analysis of a mix of events
7. Comparison of samples
8. Survey of related topics

Readership: Engineers, scientists, statisticians wanting to analyze recurrent events
This short book describes methods for the analysis of recurrent events. The emphasis is on situations involving multiple individuals or units, and the book features examples drawn from industrial, engineering and medical contexts. The book focusses almost exclusively on simple robust methods for estimating mean cumulative functions (MCF'S) for numbers of events or for related variables such as costs. This is not a bad thing: the methods, most of which were developed by the author, are useful in a broad range of settings. The methodology is clearly presented and illustrated, and software implementations are identified. A short final chapter provides references to other topics, such as models for recurrent events and how to deal with co-variates.
This book will be of interest to persons who analyze recurrent events. The nontechnical explanatory style makes it accessible to those with a limited backgrond in statistics.

Contents:
1. The generalized linear model
2. Generalized linear models for longitudinal data
3. Missing values, drop-outs, compliance and intention-to-treat
4. Generalized additive models
5. Classification and regression trees
6. Survival analysis I: Cox's regression
7. Survival analysis ll: Time-dependent covariates, frailty and tree models
8. Bayesian methods and meta-analysis
9. Exact inference for categorical data
10. Finite mixture models

Appendix A: Statistical Graphics In Medical Investigations
Appendix B: Answers to Selected Exercises

Readership: Medical statistician, trialist
The aims of this practical guide are to introduce medical researchers to a number of newer methods and to fill a gap between the many good introductory books to biostatistics and the more advanced ones that cover specific topics. The style is discursive: it often reads as a one-to-one tutorial because of the many comments enclosed in brackets. The technical content is succinct while practical applications are illustrated via several examples reinforced by exercises and listings of the relevant software.
Several topics in "modem medical statistics" are covered; as the author recognizes, the choice is subjective and, in my view, not sufficiently broad, at least if medical statistics should include applications in epidemiology. Each topic is treated briefly and assumes previous knowledge of its foundations. This is particularly true for the first chapter on generalized linear models but affects other, more advanced, topics too. Indeed this book is an excellent resource for teachers hunting for examples and for applied statisticians wishing to revise previously digested material and to update it with new references. It is not suitable for newcomers as there are gaps in the justification of several statements or important material is introduced only in the discussion of the examples.
Since the book's title states it to be a practical guide, its main deficiency, however, is the lack of explicit discussion of analytical strategies. The author is more focussed on the methods than on guiding the reader in answering the scientific questions underlying the examples.

Contents:
1. Introduction to Thiele. Lauritzen
2. On the application of the method of least squares to some cases in which a combination of certain types of inhomogeneous random sources of errors gives these a 'systematic' character. Thiele
3. Time series analysis in 1880: A discussion of contributions made by T.N. Thiele. Lauritzen
4. The general theory of observations: Calculus of probability and the method of least squares. Thiele
5. T.N. Thiele's contributions to statistics. Hald
6. On the halfinvariants in the theory of observations. Thiele
7. The early history of cumulants and the Gram-Charlier series. Hald
8. Epilogue

Readership: Researchers and graduate students of statistical science and mathematics history

T.N. Thiele, Professor of Astronomy at the University of Copenhagen for thirty-two years until 1907, and founder of the Danish insurance company Hafnia, made substantial scientific advances in actuarial science and statistics. He was an original thinker, who appears to have been one of those who independently conceived concepts such as likelihood, Kalman filtering, the EM algorithm, nonparametric density estimation, residual analysis, and the method of least squares, though, of course, not appreciating their breadth and being con-strained by the limitations of his time not to be able to develop them to the modern extent.
This book presents the first English translations of three of Thiele's statistical publications: the 1880 article in which he describes the Kalman filter; the 1889 book in which he derives the notion of likelihood for binomial experiments, presents the canonical form of the normal linear model, and describes residual analysis, amongst other things; and an 1899 article in which he completes the theory of cumulants.
The translations are supplemented by biographic information about Thiele and critical assessment of his work from Steffen Lauritzen and Anders Hald. Although Sir Ronald Fisher did not appreciate Thiele's work at the time, the fourth edition of Fisher's Statistical Methods for Research Workers lists just six people as being the main contributors to statistics: Bayes, Laplace, Gauss, Karl Pearson, Student, and Thiele. Clearly, this is a book which anyone with an interest in the history of statistics will wish to read.

Contents:
1. Introduction
General Developments
2. A historical perspective on statistics and measurement in the Netherlands 1750-1850
3. An unbridgeable gap between two approaches to statistics
4. The Dutch paths to statistics 1815-1830
Theory
5. Early quantification of scientific knowledge: Nicolaas Struyk (1686-1769) as a collector of empirical data
6. The teaching of statistics in the eighteenth century at the law faculties of the Republic of the United Provinces
7. The differentiation of statistics and political economy: the teaching of Kluit and Vissering
8. Sources of information of Dutch University Statisticians after 1800
Practice
9. The Batavian statistical revolution in the Netherlands 1798-1802: Frequency, Formats, Adminstrative Success, and Political Background
10. The beginning of health statistics 1750-1870
11. Judicial statistics before 1850
12. Statistics as an instrument in the struggle against water 1700-1850

Readership: Statisticians and historians interested in the early development of statistics in the Netherlands

A multidisciplinary team of eleven academics trace the development of the Dutch 'statistical mind' in the period 1750 to 1850. Three sources are recognized: the development of mathematical statistical theory, for example: by Adolphe Quatelet and Rehuel Lobatto, including the rise of insurance statistics: staatsbeschrijving (state description), which was taught by law faculties; and (after 1795) Dutch government statistics.
A basic understanding of the complex political history of the period would be a help to the reader; this is alluded to in many chapters. Also an appreciation of the cross-fertilization of ideas between countries at that time would be useful. The book is written lucidly and is well documented.
This is primarily a monograph for specialists in the history of statistics. It is a reminder to us all that Britain and America tended in the twentieth century to overlook the early contributions from continental European countries.

Contents:
1. Key models and concepts
2. One-way balanced case
3. More general balanced arrangements
4. Unbalanced situations
5. NonnormaI problems
6. Model extensions and criticism

Readership: Applied and research statisticians, users of statisticaI methods
This book provides an excellent treatment of the statistical and scientific concepts associated with identifying and modeling sources of variation. The emphasis is on the construction of random effects models and on their use in understanding components of variability or in making inference about fixed effects in the presence of various sources of error. The first four chapters on normal/least squares modeling consider simple to quite complex models in both balanced and unbalanced situations. Chapter 5 gives a compact introduction to Poisson and binomial random effects models and to frailty models in survival analysis. Relatively elementary introductions make the material accessible to students; researchers will be attracted to discussions of recent research, excellent bibliographic notes and a number of realistic worked examples. Each chapter is introduced with a "PreambIe" which provides a brief guide and an overview. Problems at the end of each chapter make the book attractive as a course text or principal reference.

Contents :
1. Introduction
2. Preliminaries and earlier work
3. Parametric inference
4. Nonparametric inference-Genesis
5. Smooth function estimation
6. Bayesian models
7. Record models with trend

Readership: Statisticians, mathematicians, engineers
New record values in sports, finances, climate, … are of interest to most people, and for about half a century, probabilists and statisticians have taken up the challenge of modelling their behaviour. The present monograph provides results on statistical inference problems for record-breaking data. For example: how to fit a parametric or nonparametric model to such data? Or also: how to predict the next record, based on the values of the past records. The main body of the book (Chapters 4-7) is a discussion of all the known work on nonparametric inference for this type of data.
The book will be a useful reference for researchers in this area. There could also be interest from engineers working in destructive stress testing and quality control.

Contents:
1. Background and introduction
2. Likelihood inference and the generalized linear model
3. The Bayesian set up
4. The normal and Student's t models
5. The Bayesian prior
6. Assessing model quality
7. Bayesian hypothesis testing and the Bayes factor
8. Monte Carlo methods
9. Basics of Markov chain Monte Carlo
10. Bayesian hierarchical models
11. Utilitarian Markov Monte Carlo

Readership: Statisticians and graduate students in statistics; researchers in sociology, economics and psychology
Military coups in Africa, elections in Florida, rural poverty, children's social contacts; data aggregated from many sources are of interest to sociologists and behavioural scientists. Such data are also particularly amenable to a Bayesian analysis. For researchers in the social sciences with a firm grounding in statistical inference and for graduate students in statistics wishing to enter these fields, or any other for that matter, this book gives an excellent introduction to Bayesian analysis. In fact it does more than this. It could turn them into competent practitioners.
Although written in an engagingly enthusiastic and informal style, with a good discussion of the plethora of prior distributions, it is still a seriously rigorous text from a mathematical point of view. However, manipulative mathematical skill is not demanded, the emphasis being on using computational methods. Numerical integration, the Gibbs sampler, the Metropolis-Hastings algorithm and Markov Chain Monte Carlo are discussed in detail and practical advice on their effective use is given. The availability of good, free software such as R and Wins-Bugs means that the obstacles that inhibited the widespread use of numerical Bayesian methods for many years have been overcome. Throughout this text these software tools are used, both to illustrate the examples and for the exercises that follow each chapter. The exercises are graded on a scale of 1, simple, to 50 graduate thesis! Owing to constraints of space, neither Bayesian decision theory nor empirical Bayes are discussed.
The exposition is somewhat uneven; therefore the book can be used at several levels: as a simple practical primer for Bayesian analysis, as an introduction to simulation methods and, using the detailed and up-to-date references given after each chapter, as a guide to the extensive literature on each topic. This is a useful book!

Contents:
1. Introduction
2. Advanced topics in optimal design
3. Compound symmetric error structure
4. Optimal designs in the presence of random block effects
5. Optimal designs for quadratic regression on one variable and blocks of size two
6. Constrained split-plot designs
7. Optimal split-plot designs in the presence of hard-to-change factors
8. Optimal split-plot designs
9. Two-level factorial and fractional factorial designs
10. Summary and future research

Readership: Experimental scientists, applied statisticians and academic statisticians interested in optimal design
This is number 164 in the Springer Lecture Notes in Statistics Series covering the development of optimal designs in two common types of experiment, blocked experiments and split-plot experiments. The material reads much more like a thesis than a set of lecture notes, but there is quite a concise, clearly written, introduction to optimal experimental designs provided in the opening chapter. The discussion includes blocked designs, factorial designs, response surface designs and mixture designs, with an example of each design introduced for later development. Fixed and random block effect models are examined and various optimality properties of orthogonally blocked designs are discussed. Both types of design considered may be regarded as bi-randomization designs with a compound symmetric error structure. With this formulation, the information matrix can be expressed in a convenient form which allows a more uniform treatment of the development of optimal designs. The information matrices for these types of experiments are represented as sums of outer products of vectors, a representation which permits fast update of the optimality criteria in the design algorithms. The effects of adding, deleting or interchanging design points can be readily assessed in the point exchange algorithms described. In later chapters, the methodology is applied to a selection of specific design situations including constrained split-plot designs, factorial designs and designs where some of the factors are hard to change. In each case, the underlying problem of design choice is illustrated with an example, and comparisons are made with alternative designs from the research literature.

Contents:
1. Constructional approaches and methods
2. Designs with full efficiency for some contrasts
3. Designs with no full efficiency
4. Resolvable designs
5. Special designs

There are three short appendices on finite fields, finite geometries and orthogonal latin squares.

Readership: Very mathematically oriented students and research workers in the field of incomplete block designs
This volume contains a massive amount of information about the existence and construction of designs. The reader will need to be acquainted with much of the first volume [Short Book Reviews, Vol.21, p. 44], particularly with the concept of efficiency balance that is developed in Chapter 4. There are some tables giving catalogues of designs, e.g. BIB designs with v = 100, r = 10 in Table 8.2. and resolvable BIB designs in Table 9.1. However, the average user who wishes to find a design will probably not be comfortable with the mathematical requirements of the book. The authors have produced two comprehensive volumes of which they can be proud.

Contents :
1. Why is nonparametric regression important?
2. How to construct nonparametric regression estimates?
3. Lower bounds
4. Partitioning estimates
5. Kernel estimates
6. k-NN estimates
7. Splitting the sample
8. Cross-validation
9. Uniform laws of large numbers
10. Least squares estimates I: Consistency
11. Least squares estimates II: Rate of convergence
12. Least squares estimates III: Complexity regularization
13. Consistency of data-dependent partitioning estimates
14. Univariate least squares spline estimates
15. Multivariate least squares spline estimates
16. Neural networks estimates
17. Radial basis function networks
18. Orthogonal series estimates
19. Advanced techniques from empirical process theory
20. Penalized least squares estimates I: Consistency
21. Penalized least squares estimates II: Rate of convergence
22. Dimension reduction techniques
23. Strong consistency of local averaging estimates
24. Semirecursive estimates
25. Recursive estimates
26. Censored observations
27. Dependent observations

Readership: Graduate students and researchers in statistics, mathematics, computer
science, engineering
The book gives a deep and modern mathematical treatment of nonparametric regression with random design. From the table of contents it is seen that all well-known classes of estimators are dealt with. For each of them, the authors mainly prove results on consistency and on rates of convergence. The book follows the style Theorem-Proof and gives rigorous derivations of all the results. There is a useful mathematical appendix with proofs of exponential type inequalities for sums of independent variables and for sums of martingale differences. Each chapter has a section called 'Bibliographical Notes' containing references to the extensive bibliography of more than 400 items. (The authors write that a computer search for the topic of nonparametric regression resulted in 3457 items!). Each chapter ends with a number of problems and exercises, which could be used in a teaching situation.

Contents:
1. Introduction
2. Properties of population principal components
3. Properties of sample principal components
4. Interpreting principal components: Examples
5. Graphical representation of data using principal components
6. Choosing a subset of principal components or variables
7. Principal component analysis and factor analysis
8. Principal components in regression analysis
9. Principal components used with other multivariate techniques
10. Outlier detection, influential observations, and robust estimation
11. Rotation and interpretation of principal components
12. PCA for time series and other non-independent data
13. Principal components analysis for special types of data
14. Generalizations and adaptations of principal components analysis

Readership: Anyone beginning research in principal components analysis or related techniques or who wishes a sound understanding of such tools
This is the Bible of principal components analysis (PCA). This second edition of the book is nearly twice the length of the first. [Short Book Reviews, Vol.6, p. 45] New material includes discussion of ordination methods linked to PCA, including biplots, determining the number of components to retain, extended discussion of outlier detection, stability, and sensitivity, simplifying PCAs to aid interpretation, time series data, size/shape data, and non-linear PCA, including the Gifi system and neural networks, and other topics. As can be seen from this, the book is not a narrow discussion of PCA, but links it effectively and in an illuminating way to a wide variety of other multivariate statistical tools.
Principal component analysis is the empirical manifestation of the eigen value-decomposition of a correlation or covariance matrix. The fact that a book of nearly 500 pages can be written on this, and noting the author's comment that 'it is certain that I have missed some topics, and my coverage of others will be too brief for the taste of some readers' drives home the extent to which statistics exceeds mere mathematics.
This book is an invaluable reference work and I am pleased to have it on my shelves. My only regret is that I probably will not have time to read it from cover to cover with the attention it deserves.

Contents:
1. Introduction
2. Vectors and matrices
3. Multivariate distributions and the linear model
4. Multivariate regression models
5. Seemingly unrelated regression models
6. Multivariate random and mixed models
7. Discriminant and classification analysis
8. Principal component, canonical correlation, and exploratory factor analysis
9. Cluster analysis and multidimensional scaling
10. Structural equation models

Readership: Statisticians and students of statistics; researchers in psychology, social sciences, psychology
This text is on the analysis of structured data that, for the most part, can be assumed to have been generated by a multivariate normal distribution. The topic of each chapter would warrant a book to itself, and indeed most of them have received such attention, more than once! The author has managed to encapsulate so much in this book by giving a clear statement of each model, in matrix notation of course, and using an appropriate small set of data to fit the model. Often the matrices are spelled out in detail, either numerically or symbolically. A chapter on the necessary matrix theory, well illustrated with simple numer-ical examples, provides the basic tools.
To fit the regression type models two powerful procedures in SASTM Version 8, namely Proc GLM and Proc Mixed, are used. In the detailed discussion of the output pertinent references to the literature are given, some of which would be helpful for data that depart from the underlying assumptions. Emphasis is on testing and achieving fixed significance levels. In spite of much adverse criticism in the literature, it was surprising to see how often the SAS "Type III" sum of squares for unbalanced data was recommended.
Other aspects of multivariate analysis: scaling and clustering, classification, factor analysis are also illustrated with SAS procedures and the final chapter gives a very clear introduction to structural equation models.
This book is more than an up-to date textbook on multivariate analysis. It could enable SAS users to take full and informed advantage of the many options offered in the SAS procedures. For non-SAS users, the clear statement of the models should enable them to fit and interpret them with other software.

Contents:
1. An introduction to multilevel models
2. The basic two-level model
3. Three-level models and more complex hierarchical structures
4. Multilevel models for discrete response data
5. Models for repeated measures data
6. Multivariate multileveI data
7. Multilevel factor analysis and structural equation models
8. Nonlinear multilevel models
9. Multilevel modelling in sample surveys
10. Multilevel event history models
11. Cross-classified data structures
12. Multiple membership models
13. Measurement errors in multilevel models
14. Missing data in multilevel models
15. Software for multilevel modelling, resources and further developments

Readership: Graduate students in statistics, education, social researchers and other areas
This is the third edition of this leading text providing a comprehensive treatment of modern multilevel techniques. These are statistical models which properly allow for the fact that the individual observations may have arisen from a hierarchical structure. For example, in educational research individual observations on pupils will almost certainly have been grouped into classes which were grouped into schools, and so on.
The first edition of the book appeared in 1987 [Short Book Reviews, Vol. 8, p. 25] and the second edition in 1995. The volume originally had 98 pages, grew to 178 in the second edition, and now, in a major restructuring, has 253 pages, and includes a substantial amount of new material, covering such topics as MCMC tools, bootstrapping, missing data, measurement errors, and multivariate models. The book is rich in real examples, which provide motivation for the ideas. One of the features that I found attractive was the separation of details of estimation methods into appendices of the relevant chapters: readers can obtain a good grasp of the ideas without being swamped by technical detail.
The continued growth of the book matches the continued growth of interest in and application of the ideas. A glance at a recent volume of almost any major statistics journal will indicate how important the ideas have become. This book provides the best general introduction to this important area of statistics.

Contents:
1. Data mining: A gentle introduction
2. Preparing data for data mining
3. Exploratory data analysis
4. Unsupervised learning methods
5. Supervised learning methods: Prediction
6. Supervised learning methods: Classification
7. Emerging technologies in data mining

Readership: Data analysts using SAS for data mining, graduate students in business, natural and the social sciences, and 'any SAS users who want to impress their supervisors'
This is an introduction to data mining which also enables readers to download and use SAS data mining macros. It is very much aimed at the users of this particular package, and does not describe the mathematical basis of the methods. This is entirely appropriate for most users of data mining tools, though one always has the nagging doubt that lack of proper understanding of what they are doing could lead to expensive mistakes.
As a statistician, I have to say that I found the apparent lack of appreciation of the statistical basis and history of the methods disappointing. Thus, on page 3, we find 'data mining algorithms embody techniques that have existed for at least 10 years but have only recently been implemented as mature, reliable, understandable tools that consistently outperform older methods'. The truth is that most of the tools described in the book are statistical methods which have been around for a very long time indeed (such as regression analysis, Gauss, 1811?, certainly Galton), linear discriminant analysis (Fisher, 1936), factor analysis, principal components analysis (Hotelling, 1933), cluster analysis, logistic regression, and CHAID (Kass, 1980, based on AID; Morgan and Sonquist, 1963). Of course details of the implementation algorithms have changed over the years, but the truth is that there have existed algorithms which were 'mature, reliable, understandable' for many decades (indeed, even within the SAS statistical packages which existed prior to the launch of its data mining tools), so it is not clear what 'older methods' they consistently outperform.
Looking at the detailed contents list, I found very little which a statistician would not recognize as a standard statistical tool. Such tools do exist in data mining, through its inheritance from the computer science community, but they are mainly concerned with visualization, association analysis, and so on. Association analysis (in the form of its particular application in market basket analysis) is dealt with in just two pages towards the end of the book.
I would have liked to see more discussion of the iterative and cyclic nature of the data mining process (mentioned on page 11), as well as of data quality issues. Having said that, I can see that this book will be useful for the analyst, new to the area, who is suddenly tasked by his organization with applying data mining tools to the organization's data.

Contents
1. Introduction
2. General methodology
3. Monotone censored data
4. Cross-sectional data and right-censored data combined
5. Multivariate right-censored multivariate data
6. Unified approach for causal interference and censored data

Readership:Mathematical statisticians
This book by two major research workers in the field addresses in generality important problems involving multivariate longitudinal data subject to missingness, random coarsening and possibly involving high-dimensional time-varying covariates. The primary motivation comes from biomedical applications although there are no doubt othter fields where such data arise. The approach is semi-parametric throughout and hinges on developments arising from a general estimating function approach due to J.M. Robins and A. Rotnitzky (1992) and on recent theoretical work concerning the large-sample efficiency of certain semi-parametric procedures. After seven pages listing the notation used, the introductory chapter of nearly one hundred pages introduces some key examples and sets out the basic ideas, developed further in the later chapters. There are a number of subject-matter examples of the resulting analyses although these are not readily understandable apart from the surrounding text.
This reviewer is in a difficult position. A music critic wrote once of Bach's B minor mass that he did not at all like or enjoy the work but he could appreciate that it was a great masterpiece. I could not at all understand this book. Unfortunately for me, I do not have the requisite theoretical basis that, the author's best efforts not withstanding, seems essential for such understanding. But I am reasonably confident that it is an important book dealing with important problems. Therefore, experts in modern semi-parametric theory should certainly read the book. Those with an interest focussed more on applications and able to draw together a reading group with appropriate expertise are very likely to profit greatly from a sustained study of the book.

Robins, J.M. and Rotnitzky, A. (1992). Recovery and adjustment for dependent censoring using surrogate markers. In AIDS Epidemiology: Methodical Issues. Boston: Birkhäuser.

Contents:
Part I: Methodologies of Data Mining
1. Decision trees
2. Association rules
3. Artificial neural network models for data mining
4. Statistical analysis of normal and abnormal data
5. Bayesian data analysis
6. Hidden Markov models and sequential data mining
7. Strategies and methods for prediction
8. Principal component and factor analysis
9. Psychometric methods of latent variable modelling
10. Scalable clustering
11. Time series similarity and indexing
12. Nonlinear time series analysis
13. Distributed data mining
Part II: Management of Data Mining
14. Data collection, preparation, quality and visualization
15. Data storage and management
16. Feature extraction, selection, and construction
17. Performance analysis and evaluation
18. Security and privacy
19. Emerging standards and interfaces
Part III: Applications of Data Mining
20. Mining human performance data
21. Mining text data
22. Mining geospatial data
23. Mining science and engineering data
24. Mining data in bioinformatics
25. Mining customer relationship management (CRM) data
26. Mining computer and network security data
27. Mining image data
28. Mining manufacturing quality data

Readership: Developers of data mining methods and tools, and those who want to use data mining methods
There are now several 'Handbooks of data mining' or similar compendia. One's suspicion is that they are produced for marketing reasons rather than to satisfy a detected need.
Perhaps inevitably, such books are vulnerable to (at least) three potential criticisms. The first is that their choice of content or emphasjs may not match one's own. The second is that the different chapters, written by different authors may show confusing duplication. And the third is that the different chapters may not be very well integrated. So first, for example, simple linear models receive relatively little discussion here (my first potential criticism) despite the fact that they play a very important role indeed in commercial data mining because investigations of customer behaviour can explore vast numbers of such models in reasonable time. Secondly, in the present volume, factor analysis (Chapter 8) and psychometric latent variable models (Chapter 9) are really of a kind, so it is not clear why two chapters are needed (my second potential criticism). And thirdly, methods for prediction (Chapter 7) include decision trees and neural networks (Chapter 3); therefore it is not clear why the first two have been singled out for extensive discussion and other topics have not (my third potential criticism).
However, having raised these obvious potential weaknesses of any such coIlection, I have to say that this is an impressive volume. While, again inevitably, some contributions are deeper than others, in general the level of quality is good. This book will serve as a useful resource for anyone new to data mining, and for anyone wishing to discover what potential tools are available, as well as what might be achieved through the use of those tools. This is a good 'data-mining handbook' to have on one's shelves.

Contents:
PART I: Review of Probability and Distribution Theory
1. Probability and random variables
2. Functions of random variables
PART II: Methods of Inference
3. An introduction to likelihood inference
4. Further topics in likelihood inference
5. An introduction to Bayesian inference
6. Bayesian analysis of linear models
7. The prior distribution and Bayesian analysis
8. Bayesian assessment of hypotheses and models
9. Approximate methods of inference: The EM algorithm
PART III: Markov Chain Monte Carlo Methods
10. An overview of discrete Markov chains
11. Markov chain Monte Carlo
12. Analysis of MCMC samples
PART IV:Applications in Quantitative Genetics
13. Gaussian and thick-tailed linear models
14. Analyses involving ordered categorical traits
15. Bayesian analysis of longitudinal data
16. Segregation and the QTL analysis

Readership: Graduate students and researchers in agriculture, biology and medicine with masters-level training in statistical concepts including calculus, linear algebra (matrices from a statistical perspective), probability and statistical inferences and mixed effects linear models
This monograph covers in great detail the intricate algebraic steps of Markov chain Monte Carlo methods applied to quantitative genetics with a Bayesian perspective. It is meant for scientists, particularly in animal genetics, who want to fully understand these theoretical developments with an eye to extending methods in forefront research on theory or computation. Unfortunately, there are few figures or data calculations to ground the well-presented ideas. The first two parts, Chapters 1-9, provide a fairly complete overview of statistical methods, with attention to Bayesian analysis in Chapters 5-8. Part 3 reviews MCMC methods with some nice algebraic intuition. The last part addresses some interesting twists on quantitative genetics, including phenotypes based on thresholding or from longitudinal studies. Quantitative trait loci are only briefly addressed in the last chapter of this part.

Contents:
1. Model-based analysis of oligonucleotide arrays and issues in cDNA microarray analysis by C. Li, G.C. Tseng and W.H. Wong
2. Design and analysis of comparative microarray experiments by Y.H. Yang and T. Speed
3. Classification in microarray experiments by S. Dudoit and J. Fridlyand
4. Clustering microarray data by H. Chipman, T.J. Hastie and R. Tibshirani

Readership: Scientists analyzing microarray data
The great advance associated with gene expression microarray data began with the invention of a revolutionary new measurement technology. This was allowed by the development of statistical methodology to elicit and exploit the latent information where the dimension of the response is orders of magnitude greater than the number of units measured. This book is a milestone, documenting major significant advances in the statistical methodology. The four chapters, though independent, share common foci with issues of design, robustness and the freely available associated software. The statistical ideas are introduced succinctly. The book is especially valuable for research scientists in the field seeking an understanding of the related statistical developments.

Contents:
1. The analysis of gene expression data: An overview of methods and software by G. Parmigiani, E.S. Garrett, R.A. Irizarry, and S.L. Zeger
2. Visualization and annotation of genomic experiments by R. Gentleman and V. Carey
3. Bioconductor R packages for exploratory analysis and normalization of cDNA microarray data by S. Dudoit and J.Y.H. Yang
4. An R package for analyses of affymetrix oligonucleotide arrays by R.A. Irizarry, L. Gautier and L.M. Cope
5. DNA-chip analyzer (dChip) by C. Li and W.H. Wong
6. Expression profiler by J. Vilo, M. Kopushesky, P. Kemmeren, U. Sarkans and A. Brazma
7. An S-PLUS library for the analysis and visualization of differential expression by J.K. Lee and M. O.'Connell
8. DRAGON and DRAGON View: Methods for the annotation, analysis, and visualization of large-scale gene expression data by C.M.L.S. Bouton, G. Henry, C. Colantuoni and J. Pevsner
9. SNOMAD: Biologist–friendly web tools for the standardization and normalization of microarray data by C. Colantuoni, G. Henry, C.M.L.S. Bouton, S.L. Zeger and J. Pevsner
10. Microarray analysis using the microarray explorer
by P.F. Lemkin, G.C. Thornwall and J. Evans
11. Parametric empirical Bayes methods for microarrays by M.A. Newton and C. Kendziorski
12. SAM Thresholding and false discovery rates for detecting differential gene expression in DNA microarrays by J.D. Storey and R. Tibshirani
13. Adaptive gene picking with microarray data:
Detecting important low abundance signals
by Y. Lin, S.T. Nadler, H. Lan, A.D. Attie and B.S. Yandell
14. MAANOVA: A software package for the analysis of spotted cDNA microarray experiments by H. Wu, M.K. Kerr, X. Cui and G.A. Churchill
15. GeneClust by K.-A. Do, B. Broom and S. Wen
16. POE: Statistical methods for qualitative analysis of gene expression by E.S. Garret and G. Parmigiani
17. Baysian decomposition by M.F. Ochs
18. Baysian clustering of gene expression dynamics
by P. Sebatiani, M. Ramoni and I.S. Kohane
19. Relevance networks: A first step toward finding genetic regulatory networks within microarray data by A.J. Butte and I.S. Kohane

Readership: Microarray data analysts with master's-Ievel training in computer science, biostatistics, or informatics
This useful book about methods and software for microarray data analysis is written by the developers. The goal of the book is to provide guidance to practitioners in the techniques of analysis and programs. The book combines the advantages of both a tutorial in methods and catalogue for software. Procedures are described and their implementations illustrated. All of the software described is freely available to academic users.
The Table of Contents indicates the encyclopedic coverage, spanning a very broad spectrum of methods. The collection is introduced with an informative overview. While the authors acknowledge the formative stage of current work, they have provided a useful guide to the here and now.

Contents:
Principal notation
Concordance of statements from the first edition
1. Early history
2. Basic properties of the Poisson process
3. Simple results for stationary point processes on the line
4. Renewal processes
5. Finite point processes
6. Models constructed by conditioning: Cox, cluster and marked point processes
7. Conditional intensities and likelihoods
8. Second-order properties of stationary point processes

Appendix 1: A review of some basic concepts of topology and measure theory
Appendix 2: Measures on metric spaces
Appendix 3: Conditional expectations, stopping times and martingales

Readership: Probabilists, applied probabilists, statisticians interested in methodological developments connected with point processes
The first edition of this book by two major research workers in the field speedily established itself as an authoritative account of an important and rapidly developing subject. In this substantially revised and expanded second edition, the authors have wisely decided to divide the book into two parts leaving some of the very technical material, including spatial point processes (perhaps not necessarily so technical!) to a second volume. The last three chapters of the present volume contain material largely not in the first edition, although there are substantial changes also to the earlier chapters. Workers in this field with a copy of the first edition will need this new volume too.
After a very wide-ranging first chapter, the book settles into a carefully rigorous formulation, supplemented by much valuable material in the form of exercises and complements. These form an especially important part of the book for those with more applied interests and perhaps particularly for those more concerned with developing new models for specific scientific situations than relying more on computer simulation than mathematical analysis for the derivation of properties.
It is an interesting question as to what extent there is an ultimately unavoidable clash between the needs of general rigorous formulation and those of intuitive motivation, typically based on specific simple examples. The authors have made a brave attempt to bridge this chasm. Readers will have to judge for themselves how successful they have been. A key issue arises near the start with a mathematically natural but non-intuitive and unconstructive definition of the Poisson process. The tensions get less as the discussion proceeds and the advantages of a general formulation are clearer.
The book is likely to establish itself quickly as a major contribution to the field.

Contents:
1. Guide to applications, or how to read this book
2. Motivating examples and other preliminaries
3. Biological background
4. The Galton-Watson Process
5. The age-dependent process: The Markov case
6. The Bellman-Harris process
7. Multitype processes
8. Branching processes with infinitely many types

Appendix A: Multivariate probability generating functions
Appendix B: Probability distributions for the Bellman-Harris process
Appendix C: General Processes
Appendix D: Glossaries

Readership: Mathematicians, statisticians, biologists
The theory of branching processes is an intensively developing area of mathematics, particularly stochastic processes, with many important applications in biology, medicine, physics and other natural sciences. This book contains both the mathematical background, the rigorous description of the most important classes of branching processes and the biological background with many examples of very interesting and successful applications of branching processes to biological and medical problems, in particular to cell cycle models, chemotherapy, evolution theory, gene amplification, mutations and other genetic models. The book will be very interesting and useful for mathematicians, statisticians and biologists as well, and especially for researchers developing mathematical methods in biology, medicine and other natural sciences.

Contents:
1. Introduction
2. Preliminary testing
3. Statistic methods for sparsely sampled patterns
4. Spatial point processes
5. Models
6. Model-fitting using summary descriptions
7. Model-fitting using likelihood-based methods
8. Non-parametric methods
9. Point process methods in spatial epidemiology

Readership: Statisticians, biostatistitians, epidemiologists, ecologists, numerate biologists
The statistical analysis of the location of a set of irregularly distributed points in space, i.e. spatial point patterns, arises in many branches of science from ecology, microbiology to epidemiology. The questions of interest depend on the context and range from simple tests of randomness to elaborate modelling of the spatial intensity function. This clear and synthetic text is aimed deliberately at an applied minded audience and the numerous illustrations help understanding of the techniques. It gives an up-to-date account of methods of analysis and of the recent (more mathematically based) developments in this area. The introduction to the underlying theory and to the different types of spatial point processes is didactic. This new edition has additional chapters on likelihood-based methods for estimation of point processes and on the application of these techniques to model spatial patterns of disease risk. This book is a key reference text for any researcher interested in spatial patterns.

Contents:
1. The special manifolds and related multivariate topics
2. Distributions on the special manifolds
3. Decompositions of the special manifolds
4. Distributional problems in the decomposition theorems and the sampling theory
5. The inference on the parameters of the matrix Langevin distributions
6. Large sample asymptotics theorems in connection with tests for uniformity
7. Asymptotic theorems for concentrated matrix Langevin distributions
8. High dimensional asymptotic theorems
9. Procrustes analysis on the special manifolds
10. Density estimation on the special manifolds
11. Measures of orthogonal association on the special manifolds

Appendix A:.Invariant polynomials with matrix arguments
Appendix B: Generalized Hermite and Laguerre polynomials with matrix arguments
Appendix C: Edgeworth and saddle-point expansions for random matrices

Most statisticians are trained in dealing with standard sample spaces such as Euclidean Rm. For them, there is a wealth of other sample spaces to discover through this book describing different type of manifolds. Such sample spaces arise primarily from directional data regularly occurring in Earth (or Geological) Sciences, Astrophysics, Medicine, Biology, Meteorology, and many other fields.
The two primary objects discussed in this book in great detail are the Stiefel manifold and Grassmann manifold. The first can be thought as the set of m x k matrices X such that X'X = Ik, where Ik is the k x k identity matrix. In the special case k = 1, the observations can be regarded as vectors on the unit sphere. The Grassmann manifold is representing the set of projectors on all k-planes in Rm.
The book is designed as a reference book for both theoretical and applied statisticians. It includes a very accessible introduction into the algebraic background, describes invariant measures on related manifolds as well as normal and other related distributions on them, including those fulfilling the role of the exponential distribution. For such distributions, the basic statistical techniques are introduced, including the maximum likelihood, Fisher's scoring, large sample asymptotic theorems in connection with the tests for uniformity, and non-parametric density estimation estimation on special manifolds.
More advanced topics include high-dimensional asymptotic expansions of distributions and a modification of the minimum distance method of fitting a distribution to the data. Invariant and generalized Hermite and Laguerre polynomials with a matrix arguments and saddle-point expansions are overviewed in the Appendices.
I highly recommend this carefully presented reference book to all those interested in extending their statistical reach into a well cultivated and promising landscape of statistics on manifolds.

Contents:
1. Preliminaries
PART I: Markov Chains and Ergodicity
2. Markov chains and ergodic theorems
3. Countable Markov chains
4. Harris Markov chains
5. Markov chains in metric spaces
6. Classification of Markov chains via occupation measures
PART II: Further Ergodicity Properties
7. Feller Markov chains
8. The Poisson equation
9. Strong and uniform ergodicity
PART III: Existence and Approximation of Invariant Probability Measures.
10. Existence of invariant probability measures
11. Existence and uniqueness of fixed point for Markov operators
12. Approximation procedures for invariant probability measures

Readership: Probabilists
This is a research monograph on Markov chains. The basic topic, invariant probability distributions for Markov chains, is a familiar one to anyone who has taken an undergraduate course on homogeneous, discrete time and state space Markov chains.
In that setting, if P is the transition matrix of the Markov chain, then an invariant distribution x is a solution of the equation x=ðP. Central questions are: when does an invariant distribution exist, when is it unique, and when is the chain ergodic (so that the unique invariant distribution is the equilibrium distribution)? These distributions are closely linked to the long-term proportions of time spent in (occupying) each state. and thus the convergence of the expected occupation distribution, or the pathwise occupation distribution, as the length of the time interval increases, are important issues.
As can be seen from the Table of Contents this monograph considers similar questions in a wider setting in which the state space is a general measurable space. A key concern relates to the various types of convergence of the corresponding occupation measures. Although the topics are relatively abstract the approach is well structured and each section is clearly motivated, so that the book will serve a valuable reference role for those needing these results.

Contents:
1. An introduction to option markets
2. Put-call parity and other pricing restrictions
3. An introduction to the binomial option pricing model
4. Advanced binomial option pricing
5. Practical issues associated with binomial and Black-Scholes-based option replication
6. The Black-Scholes model: Using and interpreting the 'Greeks''
7. Options arbitrage
8. Option investing from a risk-return perspective
9. Advanced option replication: Creating the most cost-effective replicating portfolio
10. The use of exchange-traded options in asset allocation
11. Pricing interest rate-dependent financial claims with option features
12. Introduction to futures, forward, and swap markets
13. Futures pricing
14. Hedging with futures
15. Interest rate futures
16. Swap markets

Readership: Academic (final-year undergraduate and postgraduate students of statistics, economics, finance, business); industry (investment banking,insurance)
The book gives a non-technical account of the field to a target audience typified by MBA students. The core topics are covered, together with material reflecting the author's research. This subject area has been treated in a number of recent books using sophisticated mathematics, involving stochastic calculus, etc.
In this book, such technicalities are avoided: the assumption is that the reader will not be mathematically advanced, e.g. in Section 6.2. a largely verbal description is given of the basic ideas of ordinary calculus, explaining the important difference between partial and 'total' derivatives.
The calculations involved in the subject are all very clearly explained, mostly using simple arithmetic illustrations rather than algebraic formulae. Likewise, the jargon is explained very clearly, making this a book that the mathematical finance specialist will find useful reading.
There are exercises at the end of each chapter, though answers are not included. Also, all but one of the chapters have a concluding summary and references for further study. The presentation is enhanced by many tables and figures, and enlivened by the use of boxes containing important messages for the reader.

Contents:
1. Introduction
2. Responsibilities of the data monitoring committee and motivating illustrations
3. Composition of a data monitoring committee
4. Independence of the data monitoring committee: Avoiding conflicts of interest
5. Confidentiality issues relating to the data monitoring committee
6. Data monitoring committee meetings
7. Data monitoring committee interaction with other trial components or related groups
8. Statistical, philosophical and ethical issues in data monitoring
9. Determining when a data monitoring committee is needed
10. Regulatory considerations for the operation of data monitoring committees
APPENDIX A: The Data Monitoring Committee Chapter

Readership: Medical statisticians and managers co-ordinating clinical trials, and other professionals involved in regulatory affairs

Recently there has been considerable highly publicized criticism of the early termination of high profile clinical studies on the basis of analyses of available data prior to the studies having run their full course. Such events exemplify the need for strictly controlled data monitoring procedures and careful assessment of decisions based on available data, both during interim analyses and on completion of the study. This text discusses the practical aspects of setting up a data monitoring committee for a clinical trial. The responsibilities and composition of the data monitoring committee are considered in detail, as is the way in which the committee carries out these responsibilities. The authors use their extensive experience in this area to explore the role of the data monitoring committee at all stages of a clinical trial, from reviewing the study protocol to ensuring the quality of the data collection and conduct of the study throughout. They point out that the committee, which should be formed from knowledgeable and experienced medical researchers without any obvious conflicts of interest, should be responsible for the assessment of safety and efficacy data and be fully involved in any decisions relating to early termination as a result of planned interim analyses.
The book discusses the reasons for having a data monitoring committee, the ethical issues related to data monitoring, the relationship between the data monitoring committee and the regulatory authority (expressed in terms of the FDA but similar relationships would hold with the MCA in the UK). There are illustrative examples used throughout the book to elucidate the issues discussed. Each chapter has an extensive list of references and a list of key points covered. The book concludes with an appendix giving the structure of a Data Monitoring Committee Charter. This provides an excellent starting point in the form of a template for anyone involved in setting up a data monitoring committee for a clinical trial.

Contents:
1. Introduction
2. Protocol development
3. Estimating the treatment difference in an individual trial
4. Combining estimates of a treatment difference across trials
5. Meta-analysis using individual patient data
6. Dealing with heterogeneity
7. Presentation and interpretation of results
8. Selection bias
9. Dealing with non-standard sets of data
10. Inclusion of trials with different designs
11. A Bayesian approach to meta-analysis
12. Sequential methods for meta-analysis
APPENDIX: Methods of Estimation and Hypothesis Testing

Meta-analysis, the determination of overall estimates and conclusions through the integration of the results of a collection of individual studies, has developed into a scientific discipline in its own right. These days, more and more meta-analyses are being reported in the medical literature, and this timely book uses a unified approach to bring together the various techniques involved by describing in detail how to plan, carry out and report the findings of a meta-analysis applied to a series of randomized controlled clinical trials.
The similarity of meta-analyses and multi-centre trials is used to identify a continuum across which the same statistical analyses may be applied. In this way, this book presents the various approaches to meta-analysis in a general framework, paying particular attention to the similarities and differences between the statistical methods employed. The meta-analysis methods are described in detail and there are copious examples to illustrate the techniques throughout the text. Most analyses are carried out using both fixed and random effects to compare and contrast the results for different underlying assumptions. The methods described involve the derivation of estimates by combining simple summary statistics or by accumulating individual patient data from many different trials. Heterogeneity between the individual trial estimates is discussed and a strategy for dealing with this heterogeneity is proposed. Methods of presenting and reporting meta-analyses are given, and guidance is provided for judging the reliability of the results. Important factors such as selection and publication bias are addressed, and special topics including the use of sequential methods and Bayesian methods within a meta-analysis are considered.
The text is a comprehensive treatment of the wealth of meta-analysis methods developed over the last two decades, and is to be highly recommended as essential reading for medical statisticians involved in the analysis of clinical trials.

Contents:
1. Introduction
2. Probability and distributions
3. Estimation and hypothesis testing
4. A simple experiment
5. Control of random variation by blocking
6. Particular questions about treatments
7. More on factorial treatment structure
8. The assumptions behind the analysis
9. Studying linear relationships
10. More complex relationships
11. Linear models
12. Nonlinear models
13. The analysis of proportions
14. Models and distributions for frequency data
15. Making and analyzing several experimental measurements
16. Analyzing and summarizing many measurements
17. Choosing the most appropriate experimental design
18. Sampling finite populations

When I reviewed the first edition of this book twenty years ago [Short Book Reviews, Vol. 3, p.31], I was very complimentary about its clear style, use of suitable numerical real-life examples and its common sense approach. All this holds for the third edition. New material on factorial structures, linear models and multivariate methods is welcome, but the authors wisely counsel experimentalists to consult a statistician for anything other than straightforward experiments. The emphasis is on encouraging experimental scientists to understand the basic principles, and to look at their experiments and the output from computer packages critically. There is little in the way of formal mathematics but this is not a "cook book" of statistical methods. Reasons are given for all the calculations and the reader is led gently through the various analyses. This must be a very useful reference book for experimentalists wanting to understand classical statistical methods, rather than simply feed their data into a package.

Contents:
A α - error
α - resolvability
Abbe, Ernst
.
.
.
Zonal sampling
zone of indifference
zone of preference

Although the title has changed, this volume, The Oxford Dictionary of Statistical Terms edited by Y. Dodge and an editorial board consisting of D.R. Cox, D. Commenges, A. Davison, P. Solomon and S. Wilson, is the sixth edition of the dictionary published for the International Statistical Institute since 1957. The first edition was published in 1957 and edited by M.G. KendaIl and W.R. Buckland, as were the second, 1960; the third, 1971 and the fourth, 1982; the fifth was edited in 1990 by F.H.C. Marriott.
The present edition took as its starting point the fifth edition deleting 265 terms and adding 640 new terms, and rewriting the definitions of many others. The sixth edition also differs from the previous editions being more like an encyclopaedia by including an extensive and useful list of references comprising sixty pages. The page style has changed from two columns, as usual in dictionaries, to one. I prefer the former style for ease of searching; all the same, this new edition is a recommended necessity for every statistician's book shelf.

Contents:
1. Sample ordering for effective statistical inference with particular reference to environmental issues by V. Barnett
2. A unified statistical approach to some measurement problems in the social sciences by D.J. Bartholomew
3. Some remarks on statistical aspects of econometrics by D.R. Cox
4. The statistical century by B. Efron
5. From association to causation: Some remarks on the history of statistics by D. Freedman
6. Scanning a lattice for a particular pattern by J. Gani
7. Mixtures everywhere by D. Karlis and E. Xekalaki
8. New paradigms (models) for probability sampling by L. Kish
9. Limit distributions of uncorrelated but dependent distributions on the unit square by S. Kotz and N.L. Johnson
10. Latent variable models with covariates by I. Moustaki
11. Some new elliptical distributions by S. Nadarajah and S. Kotz
12. Extreme value index estimators and smoothing alternatives: A critical review by J. Panaretos and Z. Tsourti
13. On convex sets of multivariate distributions and their extreme points by C. Rao, B.M. Rao and D.N. Shandhag
14. The lifespan of a renewal by J. Teugels
15. Maximum likelihood estimates of genetic effects by W. Urfer and K. Emrich
16. A predictive model evaluation and selection approach - the correlated Gamma ratio distribution by E. Xekalaki, J. Panaretos and S. Psarakis
17. Convergence rate estimates in functional limit theorems by V.M. Zolotarev

Readership: Researchers, professionals and students interested in the history and development of statistics, probability and related areas

This book includes contributions from a selection of eminent authors connected in some way to the Department of Statistics of the Athens University of Economics and Business, to celebrate the thirteen years of existence of that department.
Collections are always difficult to review because one is unlikely to be interested in or appreciate all of the contributions uniformly. The present book is, however, a pleasant surprise because it contains discussions of so many important topics. I will not attempt to comment on all of them, simply noting that I certainly found a very high proportion of the contributions extremely illuminating. They range from D.J. Bartholomew's review of his seminal work on social measurement, an area which he has unified and put on a sound statistical basis, to more specialized contributions such as that by V. Zolotarev on convergence rates. On the way, they cover topics such as causality, an area which is currently attracting considerable attention in the statistical community, discussions of new families of distributions, the relationship between statistics and economics, and a retrospective on statistics in the twentieth century. I recommend the book to any new graduate student of statistics, wishing to get an idea of the breadth of the discipline, as seen by the masters.

Contents:
Introduction
I. The treatise on equations
II. Harriot's algebra after 1621
III. Harriot's reputation and influence
Treatise on Equations
Operations of arithmetic in letters
Treatise on equations

Appendix: Correlations between Harriot 's Manuscripts and the Texts of Viète, Warner and Torporley

Thomas Harriot (c. 1560-1621) studied several branches of mathematical sciences but never published his studies. After his death, over 4000 manuscript sheets were found. This book collects a subset of these sheets, those connected with solutions of equations. Harriot seems to have been the first to introduce the modern notation ">" and "<". Stedall has done a good job in organizing Harriot's notes pertaining to algebra. This will be a useful book for historians of mathematics.

Contents:
1. Intervallwahrscheinlichkeit
2. Total determinierte Wahrscheinlichkeit
3. Partiell determinierte Wahrscheinlichkeit
4. Endliche Stichprobenräume

This book is the result of several years of work done in the research group directed by K. Weichselberger at Ludwig-Maximilians Universität Munich. The goal of the project is the creation of an axiomatic theory of interval probabilities at the same time useful for practical applications, mathematically rigorous and remaining in close linkage with classical probabilities. The book under review is an exposition of the fundamental concepts. In order to understand the ideas, the reader must know classical probability theory on the level of a typical introductory course, which puts this text within reach of a wide public interested in probability. It does not contain any exercises and is not intended to be used as a textbook. Some material of a more technical nature is collected in an appendix.
There are many reasons for generalizing traditional probabilities and several attempts have been made in this direction. Closest to the book under discussion is the concept of imprecise probabilities as explained in Walley (1991) and Kusnetsov (1991). The first chapter of the book does a good job in reviewing this history and arguing in favour of specifying probabilities of events not by a real number, as in P(E), but rather by an interval as in [L(E),U(E)]. Building on Kolmogorov's axioms for probabilities, Weichselberger calls a pair of lower and upper bounds L(),U() reasonable, if there exists at least one P() satisfying Kolmogorov's axioms and lying inside the interval, that is: L(E)<=P(E)<=U(E) for all events E. If, in addition, the infimum over all such P() is equal to L() and the supremum over all such P() is equal to U(), then the interval probability is called feasible.
In the remainder of the book, the consequences of this definition are explored, for example in making decisions based on a feasible interval probability and in comparing different interval probabilities defined on the same sample space. A
major difficulty in working with interval probabilities is the increased complexity. The third chapter of the book addresses this problem by considering partially specified interval probabilities and their completions. The last chapter is devoted to finite sample spaces and practical computations.
The author has taken great care in preparing as complete an account of the foundations of interval probabilities as possible. As a consequence, the text is somewhat unwieldy, with results and examples similar in spirit to those found in an introductory text on measure theory. For researchers interested in the foundations of statistics or in novel uses of statistical reasoning, this book offers a fine introduction to the axiomatics of interval probabilities. Anyone looking for substantial statistical applications will have to wait for the two further volumes in planning.

AdditionaI Iiterature:
Kuznetsov, V.F. (1991) Interval Statistical Models. Moscow: Radio i Svyaz.
Walley, P. (1991) Statistical Reasoning with Imprecise Probabilities. London: Chapman and Hall. [Short Book Reviews Vol.11, p. 21]

Contents:
1. Introduction
2. Probability theory
3. Stochastic processes
4. Itô's formula and stochastic differential equations
5. Monte Carlo simulation
6. Deterministic systems and input
7. Deterministic systems and stochastic input
8. Stochastic systems and deterministic input
9. Stochastic systems and input

Readership: Researchers and graduate students in science and engineering interested in the modern theory and applications of stochastic calculus

I was hoping to be able to write that here finally is a book on modern stochastic calculus without an application to finance. Alas, on page 536, the Black-Scholes formuIa appears. The emphasis, however, lies almost entirely on theory and applications relevant for engineering. As such, this book is a novel, refreshing and a must read for any quantitavely-oriented engineer who wants to know how the theory and applications of stochastic calculus evolved from one of its cradles in electrical engineering. As the author rightly states: "The book is intended to serve as a bridge between heuristic arguments used at times in the applied sciences and the very rich mathematical literature that is largely inaccessible to many applied scientists. Mathematicians will find interesting unresolved technical problems currently solved by heuristic assumptions." The whole current list of modern tools from stochastic calculus appears: from semi-martingales to stochastic differential equations, from Itô to Stratanovich, from boundary value problems to Monte Carlo simulation and numerics. Numerous examples and exercises are included. The technical background needed to be able to work through the text is non-trivial; the rewards for those being able to do so will be considerable. I highly recommend this book for the hardworking engineer wanting to link up with the old Wong-Zakai tradition of advanced stochastic modelling in engineering. The only negative remark is with respect to the far too short index; less than four pages for over a seven-hundred-page book while covering so many topics.

Contents:
1. Set
2. Probability
3. Counting
4. Random variables
5. Conditioning and independence
6. Mean, variance and transforms
7. Poisson and normal distributions
8. From random walks to Markov chains
9. Mean-variance pricing model
10. Option pricing theory
Readership: Undergraduate students with basic calculus

The justification for a fourth edition [Previous editions 1974, 1975, 1979] of this well-known and widely-used text in elementary probability is the addition of two chapters on introductory mathematical finance, contributed by the second author. With just nineteen pages devoted to option-pricing theory the coverage is of necessity very limited. Nevertheless, the new material ties in well with the earlier parts of the book, maintaining the same level of sophistication.
In spite of the original edition of the book being nearly thirty years old, the text still has a role to play in first and second year undergraduate probability courses. It provides an excellent foundation to more advanced courses in the subject. However, the opening paragraph of Chapter 1 betrays the era in which the book was first written. It discusses "the set of girls in his (KLC's son's) class" and goes on to list the names: "Nancy, Florence, Sally, Judy, Ann, Barbara,..." First, this conjures up that brief flirtation, around the 1970's, of primary school mathematical education with set-theoretic ideas and secondly, one wonders how many classes in the schools today could replicate that list of names. Fortunately, we have become a more multi-cultural society over the intervening years.

Contents:
1. The Palm calculus of point processes
2. Stationarity and coupling
3. Formulas
4. Stochastic ordering of queues

Readership: Probabilists and engineering students/researchers interested in the modelling of queueing systems

This classic text now appears in a third version: first published as Springer Lecture Notes in Statistics in 1987 [Short Book Reviews, Vol.7, p. 46], then in mono-graph format in 1994 [Short Book Reviews, Vol.15, p. 8] and finally this second edition which goes more in the direction of a textbook. Compared with the first edition, numerous sections have been reworked or added. In particular, Chapter 2 on stationarity and coupling has been expanded. A further novelty is the addition of exercises together with soIutions which vary from very fundamental must do's to more research oriented ones, the latter either practically oriented or of a more methodological nature. No doubt this textbook will further convince the queueing modeller of the essential importance of point process and martingale technology. Besides providing an elegant and broad theoretical foundation, the general results obtained allow for straightforward explicit calculations as is amply illustrated in Chapter 3 where, using Palm theory, various classical results (like Little's formula) are derived. The theory presented is non-trivial; however, those who master it will be in the possession of a powerful tool with considerable potential for applied work. As such, the
situation is akin to the fundamental knowledge of stochastic calculus in so many applications in economics and engineering. I take pleasure in recommending this text very highly.

Contents:
1. Simulating random numbers from a uniform distribution
2. Quality of random numbers
3. Quasirandom numbers
4. Transformations of uniform deviates: General methods
5. Simulating random numbers from specific distributions
6. Generation of random samples, permutations, and stochastic processes
7. Monte Carlo methods
8. Software for random number generation
9. Monte Carlo studies in statistics

Appendix A: Notation and Definitions
Appendix B: Solutions and Hints for Selected Exercises

Readership: Anyone knowledgeable of "Mathematical Statistics" with some computer literacy

This is a new edition of the author's 1998 text on simulation [Short Book Reviews Vol.19, p. 9]. The discussion on basic Monte Carlo topics: uniform and non-uniform random number generation is extended and updated, and there is some information about simulation software. There is at Ieast a short discussion of each of several newer topics, including quasi-random sequences, parallel random number generation, Markov Chain Monte Carlo, as well as brief discussions of Computational Statistics, Physics, and Finance. This second edition goes well beyond the first, though only five years separate the two.

Contents:
Part I: Introduction
Part II: Basic Concepts and Methods
1. Measures of diagnostic accuracy
2. The design of diagnostic accuracy studies.
3. Estimation and hypothesis testing in a single sample
4. Comparing the accuracy of two diagnostic tests
5. Sample size calculation
6. Issues in meta-analysis for diagnostic tests
Part III: Advanced Methods
7. Regression and independence for ROC data
8. Analysis of correlated ROC data
9. Methods for correcting verification bias
10. Methods for correcting imperfect standard bias
11. Statistical methods for meta-analysis

Readership: Clinicians who are interested in conducting diagnostic studies, statisticians conducting statistical research in diagnostic medicine

Contents:
1. Introduction
2. Study design and protocol development
3. Models for longitudinal studies
4. Missing data
5. Analytic methods for ignorable missing data
6. Simple imputation
7. Multiple imputation
8. Pattern mixture models
9. Random-effects mixture, shared-parameter, and selection models
10. Summary measures
11. Multiple endpoints
12. Design: Analysis plans
APPENDIX I: Abbreviations
APPENDIX II: Notation
APPENDIX III: Formal Definitions for Missing Data

Quality of life is, appropriately, playing an increasingly important role in clinical trials for the evaluation and comparison of treatments for chronic diseases such as cancer and AIDS. This text is based largely on the author's extensive experience in assessing quality of life in cancer clinical trials, and as a result, addresses many of the important issues that need to be considered in the design and analysis of quality of life end-points. Missing data are considered in detail, as are issues in the design of trials that evaluate quality of life. This text is a welcome and valuable addition to the literature on this important topic, and is suitable for persons with a basic understanding of statistical methods and clinical trials.

Contents:
1. Introduction
2. Generalized least square estimation
3. Maximum likelihood estimation
4. Discordant outlier and influential observation
5. Likelihood-based local influence
6. Bayesian local influence
APPENDIX: Data Sets Used in this Book

Models are discussed for data variously described as growth curves, longitudinal data, or multilevel data. The text supplements the growing number of references on techniques for longitudinal data by focusing on diagnostics for outliers and influential observations. The mathematical level is advanced, so that researchers and statisticians with a more applied background will probably find the material inaccessible. However, within this constraint the book is well written and does contain a goodly number of real-data applications.

Contents:
1. Introduction
2. Comparison of two samples
3. The linear regression model
4. Single-factor experiments with fixed and random effects
5. More restrictive designs
6. Multifactor experiments
7. Models for categorical response variables
8. Repeated measures models
9. Cross-over design
10. Statistical analysis of incomplete data
APPENDIX A: Matrix Algebra
APPENDIX B: Theoretical Proofs
APPENDIX C: Distributions and Tables

Readership: Statisticians and non-statisticians who want to learn about classical design of experiments

This book is mostly concerned with the mathematical detail of the topics in the contents. There are a few sets of data, to illustrate the material; on these, SAS, S-PLUS or SPSS is used for analysis. However, the exercises at the ends of the chapters are either theoretical or clearly made-up small examples. This would be an excellent book for mathematics students who take a course in statistics, or graduate statistics students, but is unlikely to find its way to the shelf of the industrial statistician.

Contents:
1. Introduction
2. Probability models
3. One-way classification
4. Models for 2 x 2 contingency tables
5. The general I x J contingency table
6. Log-linear models for contingency tables
7. Strategies for log-linear model selection
8. Models for binary responses
9. Logit and multinomial response models
10. Models in ordinal contingency tables
11. Analysis of doubly classified data
12. Analysis of repeated measures data

Readership: Undergraduate students of biostatistics, statistics, epidemiology, psychology, sociology, political science and education

This book offers an easy-to-read coverage of the uses and interpretation of long-established and recent statistical methods appropriate for dealing with categorical data, which consist of counts rather than measurements. The methods are presented in the context of the two software packages SAS/ STAT® (version 8e) and SPSS (version 11).
There are numerous examples and exercises interspersed throughout the text, and the SAS software codes for implementing a variety of problems are presented either in the text or in the appendices that accompany the text in a CD ROM. Most of the examples require intensive use of PC-based statistical software but the text is suitable for private study.
The mathematical prerequisites for using this book are modest, a knowledge of calculus and a basic undergraduate course in statistical methods. 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. Irritatingly for such a potentially useful student text, the first edition contains a number of misprints and missing references but these will no doubt be corrected in a second edition.

Contents:
1. Introduction
2. Gaussian-based data analyis
3. Gaussian-based model building
4. Categorical data and goodness-of-fit
5. Regression models for count data
6. Analyzing two-way tables
7. Tables with more structure
8. Multidimensional contingency tables
9. Regression models for binary data
10. Regression models for multiple category response data

Readership: Students of statistics, social science, management, economics, engineering and public administration

This book grew out of notes prepared by the author for classes in categorical data analysis. The presentation is fresh and compelling to read. Regression ideas are used to motivate the modelling presented. The book focuses on applying methods to real problems; many of these will be novel to readers of statistics texts (for example unprovoked shark attacks, beluga whale births and Challenger space shuttle damage). The book is divided into three parts: the first reviews statistical prerequisites; the second, the analysis of binary data and extensions. The initial chapters make it possible for the book to stand-alone more effectively for personal instruction and as a course companion. All chapters end with a section providing references to books or articles for the inquiring reader.
Sets of data and computer code are available at a Web site devoted to the text. More than two hundred exercises are provided, including many based on recent subject-area scientific literature.

Contents:
1. Introduction
2. Statistical inference for binary data
3. Models for binary and binomial data
4. Bioassay and some other applications
5. Model checking
6. Overdispersion
7. Modelling data from epidemiological studies
8. Mixed models for binary data
9. Exact methods
10. Some additlonal topics
11. Computer software for modelling binary data

Appendix A: Values of Logit(p) and Probit(p)
Appendix B: Some Derivations
Appendix C: Additional Data Sets

Readership: Statisticians and students of statistics, researchers in biological and medical sciences

The first edition of this book [Short Book Reviews, Vol. 12, p. 5] with its lucid explanations and apposite examples did much to promote an understanding of logistic regression in statistical and research communities. Over the intervening eleven years there have been many extensions to the theory. The wide-spread availability of sophisticated software makes their implementation easy both for professional statisticians and for research workers in other fields. Understanding the full implications, limitations, and possible pitfalls of new methodology is not so easy. This book gives the reader much insight and practical guidance. The author has a gift for explaining quite complex issues, by means of a few key formulae and well-chosen simple numerical examples.
The first seven chapters are principally a revision of those of the first edition slightly rearranged and updated with some new material. Details of the GLIM or SAS computer code used to generate the results have now been omitted. Instead, in Chapter 11, a brief review of the procedures for the analysis of binary data offered in widely-available statistical packages is given, together with some useful comments on their idiosyncrasies.
Chapter 8 deals with mixed models, an area in which there has been much deveIopment since the first edition of this book. The approach used is that of maximizing the marginal likelihood after integrating out the random effects; the method used is, interalia, by SAS. Estimation methods that avoid integration are not discussed, nor are any methods based on generalized estimating equations. Good examples of increasing complexity show how the mixed model can be used to analyze longitudinal or multilevel data. Chapter 9 on exact methods begins with a derivation of Fisher's exact test for the 2 by 2 tables and proceeds to develop the conditional likelihood approach for matched data. A useful discussion on testing and estimation follows. Chapter 10 briefly discusses modelling ordered categorical and multivariate responses, rates, time series and measurement errors. There is no mention of fully Bayesian methods.
This second edition retains all the good qualities of the first. It presents an accessible, up-to-date, non-mathematical account of modelling binary data. Its value rests in the practical insights it offers. For those who have a well-thumbed copy of the first edition, be assured that there is sufficient new material here for you to consider replacing it! For others, it is highly recommended!

Contents:
1. Introduction
2. Direct domain estimation
3. Traditional demographic methods
4. Indirect domain estimation
5. Small area modeIs
6. Empirical best linear unbiased prediction: Theory
7. EBLUP: Basic modeIs
8. EBLUP: Extensions
9. Empirical Bayes (EB) method
10. Hierarchical Bayes (HB) method

In survey methods terminology, a small area is a subpopulation with so few sample points that its attributes cannot be estimated from sample information with adequate precision, without "borrowing strength" from data for other subpopulations. The small area may be a geographical area such as a state or county, or a small group defined demographically. The book is a systematic and economical account of the development of the subject by one of its foremost contributors. There are numerous illustrations, drawn from some of the major applications of small area theory: estimation of median incomes. poverty counts or unemployment rates for small jurisdictions; estimation of crop yields for counties; analysis of census undercoverage; mortality and disease mapping from surveys or administrative data. The exploration of generalized linear mixed models and their uses, for example in empirical best linear unbiased prediction estimates, is self-contained. There is careful emphasis on the importance of the formulation of goals, the validation of models and the accurate estimation of uncertainty. The book is intended as a research monograph, and can be read for an overview or for more intensive study. Numerical methods and implementation software are discussed in some detail. An attractive feature is the collection of proofs of the basic results at the end of each chapter.

Contents:
1. Introduction
2. Methods for one-sided equivalence problems
3. General approaches to the construction of tests for equivalence in the general sense
4. Equivalence tests for selected one-parameter problems
5. Equivalence tests for designs with paired observations
6. Equivalence tests for two unrelated samples
7. Multisample tests for equivalence
8. Tests for establishing goodness of fit
9. The assessment of bioequivalence

Appendix A: Basic Theoretical Results
Appendix B: List of Special Computer Programs
Appendix C: Frequently Used Special Symbols and Abbreviations

Is a new generic version of a drug as good as the original? We are typically taught in class either to "reject" a null hypothesis or to "not reject" it. What if we want to "prove the null hypothesis", that is, to be able to assert that it is "true"? We need this book. A first version appeared in German in 1994. Out of roughly one hundred and thirty-five references, thirty-three are newer than 1994, indicating good updating of the material.
Courses of action are indicated for how readers "primarily interested in practical applications" and for those "particularly interested in theory and mathematical background" should approach the book. The main difference is that the former should forget Appendix A and most of Chapter 3. Examples come from the author's field of application, medical research. This is a short, specialist volume in the traditional Chapman and Hall style. It will serve its intended audience well.

Contents:
1. The role of Bayesian and frequentist multivariate modeling in statistical data mining
2. Intelligent statistical data mining with information complexity and genetic algorithms
3. Econometric and statistical data mining, prediction and policy making
4. Data mining strategies for the detection of chemical warfare agents
5. Disclosure limitation methods based on bounds for large contingency tables with applications to disability
6. Partial membership models with application to disability survey data
7. Automated scoring of polygraph data
8. Missing value algorithms in decision trees
9. Unsupervised learning from incomplete data using a mixture model approach
10. Improving the performance of radial basis function (RBF) classification using information criteria
11. Use of kernel based techniques for sensor validation in nuclear power plants
12. Data mining and traditional regression
13. An extended sliced inverse regression
14. Using genetic algorithms to improve the group method of data handling in time series prediction
15. Data mining for monitoring plant devices using GMDH and pattern classification
16. Statistical modeling and data mining to identify consumer preference
17. Testing for structural change over time of brand attribute perceptions in market segments
18. Kernel PCA for feature extraction with information complexity
19. Global principal component analysis for dimensionality reduction in distributed data mining
20. A new metric for categorical data
21. Ordinal logistic modeling using ICOMP as a goodness-of-fit criterion
22. Comparing latent class factor analysis with the traditional approach in data mining
23. On cluster effects in mining complex econometric data
24. Neural network-based data mining techniques for steel making
25. Solving data: A clustering problem as a string search problem
26. Behaviour-based recommender systems as value-added services for scientific libraries
27. GTP (General Text Parser) software for text mining
28. Implication intensity: From the basic statistical definition to the entropic version
29. Use of a secondary splitting criterion in classification forest construction
30. A method integrating self-organizing maps to predict the probability of barrier removal
31. Cluster analysis of imputed financial data using an augmentation-based algorithm
32. Data mining in federal agencies
33. STING: Evaluation of scientific and technological innovation progress
34. The semantic conference organizer

Readership: Those interested in the latest developments in data mining and related areas

This book contains a collection of selected representative papers of the thematic areas covered in the C. Warren Neel International Conference on Statistical Data Mining and Knowledge Discovery, June 2002. Some of the papers present broad overviews of particular areas, while others describe more detailed research projects. Many of the contributors are recognized international experts in their areas. I liked the mixture of methodological and applications papers.
The volume has been very nicely produced, and I am confident that anyone with an interest in this area would find it contains something of value.

Contents:
Part I: Conceptual Bases of Experimental Design and Analysis
1. The logic of experimental design
2. Introduction to the Fisher tradition
Part II: Model Comparisons for Between-Subjects Designs
3. Introduction to model comparisons: One-way between-subjects designs
4. Individual comparisons of means
5. Testing several contrasts: The multiple comparison problem
6. Trend analysis
7. Two-way between-subjects factorial designs
8. Higher-order between-subjects factorial designs
9. Designs with covariates: ANCOVA and blocking
10. Designs with random or nested factors
Part III: Model Comparisons for Designs Involving Within-Subject Factors
11. One-way within subject designs: Univariate approach
12. Higher-order designs with within-subjects factors
13. One-way within-subjects: Multivariate approach
14. Higher-order designs with within-subjects factors: Multivariate approach
Part IV: Alternative Analysis Strategies
15. An introduction to multilevel models for within-subjects-designs
16. An introduction to multilevel hierarchical mixed models: Nested designs

Appendix A: Statistical Tables
Appendix B: Part 1. Linear Models: The Relation Between ANOVA and Regression
Part 2. A Brief Primer of Principles of Formulating and Comparing Models
Appendix C: Notes
Appendix D: Solutions to Selected Exercises

The authors give a comprehensive presentation of traditional statistical methods, aimed at experimentalists mainly in psychology. The mathematical level is low and the exposition is lucid and very detailed. The title is a little misleading as this book is not about how to design an experiment (which should happen in advance) but more about choosing which type of standard model to use after the data have been collected, and how to analyze it. Although the book is long there is hardly any mention of residuals or the checking of model assumptions.
This title is a second edition [Short Book Reviews, Vol. 10, p. 26] and has new features including an increased use of statistical packages and graphical representation of data, the general principles of formulating models, a discussion on statistical reasoning and a CD with SPSS and SAS data sets and tutorials reviewing basic statistics. There are ample examples and exercises.

Contents:
1. Introduction
2. The 2x2 cross-over trial
3. Higher-order designs for two treatments
4. Designs for three or more treatments
5. Analysis of continuous data
6. Analysis of categorical data
7. Bioequivalence trials

This new edition of a popular work [Original 1989; Short Book Reviews, Vol.10, p. 4] maintains its valuable mixture of theory and practice. It combines a thorough review of the literature with a practical common-sense approach. The authors have "tried to maintain a practical perspective and to avoid those topics that are of largely academic interest". For me, this is the real attraction of the book. The authors stress the need to be aware of the realities of the situation in which an experiment is being done; for example, in Chapter 2, they include advice to use "wash-out periods of adequate length" and note that this "requires a good working knowledge of the treatment effects… based on prior knowledge of the drugs under study or ones known to have a similar action".
The early chapters on design of trials have been extended to cover more recent developments in computer search algorithms. The practical examples have now been largely gathered into Chapters 5 and 6, where a variety of examples are analyzed in some detail. In this edition, the analyses of data are more tightly tied to statistical packages, especially SAS and WinBUGS. This will make it easier for readers to try and compare different analyses for themselves.
The sections on analysis of categorical and count data have been largely rewritten to reflect developments in those areas in recent years, and a new chapter on bioequivalence has been added. The book will remain a valuable reference for medical researchers and statisticians involved in cross-over trials.

Contents:
1. Statistical process control: A brief overview
2. Acceptance-rejection SPC
3. The development of mean and standard deviation control charts
4. Sequential approaches
5. Exploratory techniques for preliminary analysis
6. Optimization approaches
7. Multivariate approaches

Appendix A: A Brief Introduction to Linear Algebra
Appendix B: A Brief Introduction to Stochastics
Appendix C: Statistical Tables

Readership: Numerate practitioners of quality improvement, statisticians in industry, and advanced undergraduate or graduate students of statistics and industrial engineering

This is the second edition of a book that this reviewer enjoyed a decade ago, published at that time under the title Statistical Process Control for Quality Improvement [Short Book Reviews Vol.13, p. 18]. The book has increased in Iength by about twenty-five per cent with the addition of examples aimed at chief executive officers and service industries, as well as new discussions of management and planning tools, bootstrapping and Bayesian approaches for exploratory analysis. Coverage is very broad compared with the book's origin as a short course for workers, foremen and managers in companies in the United States of America and Poland. In fact statistical process control
topics occupy barely half the book. The authors are still full of passion, lamenting the unconstructive sociology of Quality Assurance as opposed to continuous improvement, and pointing out that "Deming was an optimizer, not a policeman". The authors' model-based approach to the Deming paradigm, and the necessary statistical underpinnings of their tools of analysis, demands greater mathematical maturity than is normally expected of readers of books about statistical process control. This book serves best those who are already fairly well read in statistics, but who seek some grounding in quality improvement, as well as numerate practitioners who wish to deepen their knowledge. It couId be a daunting read for beginners.

Contents:
1. Introduction
2. Microarrays
3. Image processing
4. Elements of statistics
5. Statistical hypothesis testing
6. Classical approaches to data analysis
7. Analysis of variance
8. Experimental design
9. Multiple comparisons
10. Analysis and visualization tools
11. Cluster analysis
12. Data pre-processing and normalization
13. Methods for selecting differentially regulated genes
14. Functional analysis and biological interpretation of microarray data
15. Focused microarrays – comparison and selection
16. Commercial applications
17. The road ahead

Readership: Microarray users with little understanding of statistics, and even readers with a shaky mathematical foundation, who will be working with more experienced collaborators

This is a wonderful book for researchers anaIyzing microarrays who know almost nothing of mathematics and statistics. The reader is.introduced to microarrays, the preprocessing of images, elementary statistics and related software. The application of some commercial packages is illustrated. The scope is exceedingly large. Even logarithms are explained. However, as a consequence, the depth of the discussion is limited. Although the important issues of experimental design are introduced, the discussion is technical rather than fundamental. For example, dye-swapped experiments are advocated initially so that an interaction sum of squares can be computed. While many scientists will find this book useful, those analyzing
microarrays will benefit from more advanced presentations.

Volume I:
1. Continuous population models for single species
2. Discrete population models for a single species
3. Models for interacting populations
4. Temperature-dependent sex determination (TSD)
5. Modelling the dynamics of marital interaction: Divorce prediction and marriage repair
6. Reaction kinetics
7. Biological oscillators and switches
8. BZ oscillating reactions
9. Perturbed and couples oscillators and black holes
10. Dynamics of infectious diseases
11. Reaction diffusion, chemotaxis, and nonlocal mechanisms
12. Oscillator-generated wave phenomena
13. Biological waves: Single-species models
14. Use and abuse of fractals

Appendix A: Phase Plane Analysis
Appendix B: Routh-Hurwitz Conditions, Jury Conditions, Descartes' Rule of Signs, and Exact Solutions of a Cubic

Volume II:
1. Multi-species waves and practical applications
2. Spatial pattern formation with reaction diffusion systems
3. Animal coat patterns and other practical applications of reaction diffusion mechanisms
4. Pattern formation on growing domains: Alligators and snakes
5. Bacterial patterns and chemotaxis
6. Mechanical theory for generating pattern and form in development
7. Evolution, morphogenetic laws, developmental constraints and teratologies
8. A mechanical theory of vascular network formation
9. Epidermal wound healing
10. Dermal wound healing
11. Growth and control of brain tumours
12. Neural models of pattern formation
13. Geographic spread and control of epidemics
14. Wolf territoriality, wolf-deer interaction and survival

This book, a classical text in mathematical biology, cleverly combines mathematical tools with subject area sciences. The multi-layer way of material presentation makes the book useful for different types of reader including graduate-level students, bioscientists with a reasonable knowledge of calculus, mathematicians, physicists, etc. I like the main emphasis of the book on conversion of "an understanding of the underlying mechanisms into predictive science". The author considered a wealth of continuous deterministic models which are applied to numerous areas in biology. The third edition [First edition, Short Book Reviews, Vol. 10, p. 9] significantly increases the scope. Somehow this increase makes the text less elegant – a distinctive feature of previous editions.
In contradiction to my previous statement, I would like to note that discrete and stochastic models could contribute to a better understanding and comparison of mathematical tools used in modern biology. In spite of the last comment, I consider these volumes the best in their class; it is an enjoyable reading and I recommend it to anyone with serious interest in mathematical modelling.

Contents:
1. The basis for, and advantages of Bayesian model estimation via repeated sampling
2. Hierarchical mixture models
3. Regression models
4. Analysis of multi-level data
5. Models for time series
6. Analysis of panel data
7. Models for spatial outcomes and geographical association
8. Structural equation and latent variables models
9. Survival and event history models
10. Modelling and establishing causal relationships: Epidemiological methods and models

Readership: Applied statisticians; quantitative researchers in physical, biological, medical and social sciences

A companion volume to the author's previous work [Bayesian Statistical Modelling, 2001, Short Book Reviews, Vol. 21, p. 46], this book provides a further exposition on the advantages of Bayesian modelling and demonstrates the feasibility of Bayesian methods even in complex problems.
Each of the ten chapters, which concentrates respectively on the key models in a range of familiar experimental situations, has numerous examples, an extensive reference list and exercises of varying levels of difficulty.
The book has a contemporary feel, with recent devel
opments in financial time series modelling and epidemiology being included. The emphasis is definitely on applications, with implementation in WinBugs.

Contents:
1. Missing outcomes
2. Instrumental variables
3. Conditional prediction with missing data
4. Contaminated outcomes
5. Regressions, short and long
6. Response-based sampling
7. Analysis of treatment response
8. Monotone treatment response
9. Monotone instrumental variables
10. The mixing problem

Readership: Postgraduate students and researchers in statistics and econometrics

The coverage of the book can be well summarized by the following quotations from its Introduction:
This book presents the main elements of my research on partial identification of probability distributions,
which has its roots in my research on nonparametric regression analysis with missing outcome data; The approach to inference that runs throughout the book is deliberately conservative and thoroughly nonparametric;
The Law of Decreasing Credibility: The credibility of inference decreases with the strength of assumptions made.

I found the material very pertinent, departing, as it does, from the usual parametric approach in which the conclusions depend rather critically on the probability model adopted. Given a chance, it will make the traditionalist, like me, stop and think and, perhaps, try to mend their ways a little.
The main part of each chapter is written in textbook style, but fairly formally and rigorously, with Propositions and Corollaries, etc. However, the level is fairly modest, requiring only a familiarity with probability, random variables and their distributions. At the end of each chapter appear "Complements", giving examples and extensions, and "Endnotes", giving literature sources and discussion.
The prototypical situation is where one has an outcome y that has been observed under condition z=1, y not being observable when z=0. The unconditional distribution of y is then given by
p(y) = p(y|z=1)p(z=1) + p(y|z=0)p(z=0),
and it is this that one wishes to determine. Assuming that p(z=1), and therefore p(z=0), is known, we can only put bounds on p(y) depending on the size of p(y|z=0). This core problem is elaborated and extended in the rest of the book to cover cases where there are 'reasonable' assumptions about p(y|z=0), covariates, missing or contaminated data, and other special circumstances.

Contents:
Part A: Simple Markovian Models
I. Markov chains
II. Markov jump processes
III. Queueing theory at the Markovian level
IV. Queueing networks and insensitivity
Part B: Some General Tools and Methods
V. Renewal theory
VI. Regenerative processes
VII. Further topics in renewal theory and regenerative processes
VIII. Random walks
IX. Lévy processes, reflection and duality
Part C: Special Models and Methods
X. Steady-state properties of GI/G/1
XI. Markov additive models
XII. Many-server queues
XIII. Exponential change of measure
XIV. Dams, inventories and insurance risk

Readership: Researchers and graduate students in industrial engineering, probability theory, operations research and queueing theory

This book, which focuses mainly on queueing theory and the basic structures and mathematical tools needed to study such models, will be a valuable resource to all those interested in applied probability and stochastic modelling. It provides a clear and careful unified treatment of traditional queueing theory, with some applications to inventory and insurance risk theory. Moreover, this updated second edition, with approximately one hundred more pages than the first edition published in 1987 [Short Book Reviews, Vol. 7, p. 46], includes a very nice treatment of queueing networks, matrix-analytic methods, Palm probability, as well as other more recent advances to the discipline. The material is self-contained, but it is technical and a solid foundation in probability theory is beneficial to prospective readers. Researchers and graduate students interested in these fields will no doubt want to acquire this book.

Contents:
1. Introduction
2. Some historical background
3. Selfsimilar processes with stationary increments
4. Fractional Brownian motion
5. Selfsimilar processes with independent increments
6. Sample path properties of selfsimilar stable processes with stationary increments
7. Simulation of selfsimilar processes
8. Statistical estimation
9 .Extensions

Self-similar processes first came to widespread public awareness some twenty years ago through the strikingly beautiful pictures of Benoit Mandelbrot. The study of such processes (which have properties that are invariant under transformations of spatial or temporal scale) goes back further, at least to the mid-twentieth century. However, in the last twenty-five years, the pace of research has speeded up as their importance for applications in areas such as mathematical finance has become increasingly apparent. Theoretical developments have largely arisen in two rather distinct fields – probability and statistical physics – and the literature is quite disparate, especially when applications are included. Thus this book, which aims to draw together the main results, is extremely welcome. It will be an invaluable starting point for those coming into the subject, perhaps in the context of a particular application, as well as a very handy reference for experts. The authors say that their "text should be viewed as intermediate lecture notes trying to bridge the gap between the various existing developments" until a definitive text is written. Anyone setting out to produce such a definitive account will have an excellent starting point.
At barely one hundred pages, this monograph is short and concise, with an extensive set of references, very many of
these having appeared in the last ten years. While the majority of the book is concerned with theoretical properties of various classes of self-similar processes, the substantial chapters on simulation and on statistical estimation will be particularly valuable for those involved in applications.

Contents:
1. Introduction and overview
2. Short- and long-run models
3. Testing for unit roots
4. Co-integration in single equations
5. Co-integration in multivariate systems
6. Modelling the short-run multivariate system
7. Panel data models and cointegration
8. Modelling and forecasting financial time series

Appendix: Co-integration Analysis Using the Johansen Technique: A Practitioner's Guide to PcGive 10.1

Readership: Academic (final-year undergraduate and postgraduate students Economics, Finance, Business, Industry (banking, insurance))

The book gives a non-technical introduction to methods for non-stationary data. Mathematical details are downplayed, many either relegated to boxes or referred to literature sources. For example, in Chapter 1 we read `It can be shown that
1/(I-rL) = (l+rL+r^2L^2+r^3L^3…)`.
This gives an indication of the mathematical level expected of a typical reader who will therefore need to take many of the mathematical results on trust.
The first topic covered, after the introductory Chap-ter 1, is that of short-run and long-run models, and the associated distinction between stationary and non-stationary series. The following chapter is devoted to testing for unit roots, concentrating on the Dickey-Fuller test and its relatives. Co-integration is explored in Chapters 5 and 6, in univariate and multivariate systems, respectively. The final three chapters cover more advanced material in an accessible way.
The flavour of the book is Economics rather than Mathematics. However, there are plenty of equations, and not just in the boxes; the equations are conveniently numbered, though the chapter sections are not. The emphasis is on fitting data and analysis and examples are given throughout.

Contents:
1. Basic notions of probability theory
2. Poisson processes
3. Convergence of superpositions of independent stochastic processes
4. Compound Poisson distributions
5. Classical risk processes
6. Doubly stochastic Poisson processes (Cox processes)
7. Compound Cox processes with zero mean
8. Modeling evolution of stock prices by compound Cox processes
9. Compound Cox processes with nonzero mean
10. Functional limit theorems for compound Cox processes
11. Generalized risk processes
12. Statistical inference concerning the parameters of risk processes

This book presents a comprehensive survey on (compound) Cox processes. In addition to well-known classical results, it also contains new results by the authors such as asymptotic approximations and convergence criteria. The applied problems considered in this book are mostly insurance related. Generalized risk processes and functionals thereof are examined in great detail. On the other hand, the applied finance problems are confined to logarithmic stock price returns coupled with stochastic intensities of trade. Even though the book is rather technical at some stages, it reads well. It will be of great value to anyone who is confronted with practical problems in insurance and finance.

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 values
6. Time series regression modeling
7. Univariate GARCH modeling
8. Long memory time series modeling
9. Rolling analysis of time series
10. Systems of regression equations
11. Vector autoregressive models for mulitivariate 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

Readership: The intended audience (cf. page v) comprises practitioners, researchers and students in empirical finance and financial econometrics. Basic familiarity with S-PLUS® is required, as is a basic background in mathematical statistics and time series

With Modeling Financial Time Series with S-PLUS®, Zivot and Wang deliver an impressive tour de force covering many relevant topics in modern financial econometrics. As the table of contents outlines, the book includes anything from modern time series methods (unit roots, cointegration, long-memory modeling) to recent advances in risk management (extreme value analysis), multivariate data analysis as applied to portfolio management, yield-curve modeling to two detailed chapters on the already classic univariate and multivariate GARCH-type volatility models. The topics are generally introduced in a succinct manner with brief formal discussions complemented by numerous references to the literature that are provided on a per-chapter basis. In general, the emphasis is on actual applications and examples using the S-PLUS® FinMetrics package, exemplifying the various available functions using included example datasets.
However, the book is really just a manual to the software package FinMetrics – and in fact includes a large (5.8mb) pdf file with the FinMetrics software. Users with access
to both S-PLUS® and FinMetric will find it very helpful. But those without the software will be left wondering how much more useful the book could have been if only the authors had followed the lead of Venables and Ripley (2002, 4th edition; 2nd
edition, 1994, Short Book Reviews, Vol.15, p. 24) and provided code for both implementations of the S language, S-PLUS® as well as R.

Contents:
1. The discrete case
2. Dynamic models in discrete time
3. The Black-Scholes formula
4. Portfolios optimizing wealth and consumption
5. The yield curve
6. Equilibrium in financial markets in discrete time
7. Equilibrium of financial markets in continuous time. the complete markets case
8. Incomplete markets
9. Exotic options

The objective of this book is to develop the continuous time theory of the valuation of asset prices and the theory of the equilibrium of financial markets. Four chapters are devoted to the valuation of asset prices and three to "equilibrium theory" including the theory of optimal portfolio and consumption choice. This is a high-level but well-written summary of the modem essentials of mathematical finance, including excellent chapters on the yield curve, pricing interest rate products, exotic options and incomplete markets.

Contents:
1. Coherent risk measures on general probability spaces
2. Robust preferences and convex measure of risk
3. Long head-runs and long match patterns
4. Factor pricing in multidate security markets
5. Option pricing for co-integrated assets
6. Incomplete diversification and asset pricing
7. Hedging of contingent claims under transaction costs
8. Risk management for derivatives in illiquid markets: A simulation study
9. A simple model of liquidity effects
10. Estimation in models of the instantaneous short term interest rate by use of a dynamic Bayesian algorithm
11. Arbitrage-free interpolation in models of market observable interest rates
12. The fair premium of an equity-linked life and pension insurance
13. On Bermudan options
14. A Barrier version of the Russian option
15. Laplace transforms and suprema of stochastic processes
16. Solving the Poisson disorder problem

This is a collection of papers in Finance and Stochastics in honour of D. Sondermann by leading researchers in Mathematical Finance. The papers are divided roughly into five categories: Risk Management in Incomplete Markets, Portfolio Theory, Market Imperfections (for example the effects of illiquidity and transactions costs), Interest Rate Modelling. and the pricing of Exotic Options. This volume of research is in honour of "a creative researcher and the editor of a leading journal, who has helped shape the subject of mathematical finance".

Contents:
1. Overview
2. Introduction to real options
3. Real options and stochastic control
4. Valuing real options in a stochastic control framework.
5. Extensions: Competition and time delay effects
6. Case Study: Flexibility in the manufacturing industry
7. Conclusions and extensions

Over the last decade real options theory has rapidly grown and been recognized as a helpful tool for valuing managerial fIexibility/decisions. Based on a stochastic control framework, the book applies the decomposition of decision situations involving real options into interactions generalized timing and switching options to the huge field of real option models. Corroborated by convincing graphical representation of real option interactions as well as a solid but sometimes clumsy mathematical formulation, the book provides a numerical and analytical approach to solve complex option interaction models. From the theoretical point of view the book opens an interesting new view on real options, but – like most books in this field – with regard to the practical point of view the book may not convince and seems therefore of little use for practitoners facing real life business decisions.

Contents:
Introduction
0. Warm-up
1. Toolbox: Moment bounds for solutions of stable SDE's
2. Tikhonov theory for SDE's
3. Large deviations
4. Uniform expansions for two-scale systems
5. Two-scale optimal control problems
6. Applications

Appendix: Stochastic equations, exponential bounds, measurable selection, Hellinger processes,
Hausdorff metrics, compact sets in the space of probability measures

Readership: Anybody with an interest in stochastic differential equations including statisticians, mathematicians, control and telecommunications engineers

Two time-scale problems arise in many problems of practical interest. The essential ingredient in these problems is a decomposition of the dynamics into "fast" and "slow" modes. Examples include the use of stochastic approximation procedures of the Robbins-Monro type and when one filters a fast process observing a slow one. This book deals, inter alia with "averaging" methods where the coefficients of the Iimiting slow dynamics are obtained by averaging the fast dynamics. Indeed, the idea of "averaging" has found wide-spread application in related problems. The book is uncompromisingly mathematical but clearly and elegantly written.

Contents:
1. Introduction
2. Bayesian framework
3. Gaussian prior factors
4. Parameterizing likelihoods: Variational methods
5. Parameterizing priors: Hyperparameters
6. Mixtures of Gaussian prior factors
7. Bayesian inverse quantum theory (BIQT)
8. Summary

Appendix A: A Priori Information and a Posteriori Control
Appendix B: Probability, Free Energy, Energy, Information, Entropy, and Temperature
Appendix C: Iteration Procedures: Learning

Readership: Physicists and other specialists in the rapidly growing number or fields that make use of Bayesian methods

This book uses the connection between nonparametric Bayesian approaches and statistical field theories to explore methods for incorporating a priori information in nonparametric approaches, and applies nonparametric Bayesian methods to physics problems. In particular, it seeks to provide a toolbox to describe a priori information in nonparametric models. The nonparametric models considered include Gaussian processes, mixtures of Gaussian processes, non-quadratic potentials, and also models specified via hyperparameters, hyperfields, and auxiliary fields.
Occasional statements in the book jarred a little. Conditioning is regarded as implying causality on page 10, and the distinction between the meanings of the word 'independent' in consecutive paragraphs, though explained, is surely confusing and seems unnecessary ('exogenous' might have been a viable alternative). Apart from such minor quibbles, the book is rich and clearly written, and will amply reward the efforts of either physicists or statisticians interested in issues involving both disciplines.

Contents:
1. Introduction
2. Using likelihood for estimation
3. Building blocks
4. Plot sampling
5. Removal, catch-effort and change-in-ratio
6. Simple mark-recapture
7. Distance sampling
8. Nearest neighbour and point-to-nearest-object
9. Further building blocks
10. Spatial/temporal models with certain detection
11. Dealing with heterogeneity
12. Integrated models
13. Dynamic and open population models
14. Which method?
APPENDIX A: Notation and Glossary
APPENDIX B: Statistical Formulation for Observation Models
APPENDIX C: The Asymptotic Variance of MLEs
APPENDIX D: State Models for Mark-Recapture and Removal Methods

Readership: Statisticians, quantitative biologists and ecologists interested in studying animal populations

This book introduces and surveys the various methods used to assess wildlife populations. It seeks to provide a unified approach based on likelihood – nearly all the methods given are formulated as special cases of a few general likelihood functions. The necessary theory of likelihood and maximum likelihood is provided in the second chapter. Both simple and more advanced models are described. The important problem of heterogeneity (differences in detectability of catchability) and how to handle it is ad-dressed in the latter part of the book. The penultimate chap-ter provides a brief introduction to handling open populations. The final chapter gives useful information on the types of situation for which individual methods are most appropriate.
This is a useful book. It provides a comprehensive, readable survey which should be of value to any-one who wants an introduction to the currently available methodology for estimating animal numbers. The book is well written. Most chapters start with one-line key ideas be-hind the methods described and end with a short summary and some exercises (some solutions can be found on the web). Relevant free software is also available on the web in the form of an R Library of simulation and estimation functions for many of the advanced methods – these are designed for learning and teaching purposes, rather than for use in practice.

Contents:
1. Introduction
2. Epidemiologic research: An overview
3. Epidemiologic study designs
4. Measures of disease frequency
5. Measures of effect
6. Measures of potential impact
7. Validity and general considerations
8. Selection bias
9. Information bias
10. Confounding
11. Confounding involving several risk factors
12. Statistical inferences about effect measures
13. Control extraneous factors
14. Stratified analysis
15. Matching

Readership: Students in an introductory epidemiology course based on statistical methods

This is a most impressive disc. One quickly brings up a first page screen and each click on a "right arrow" turns the page (with an appropriate sound effect!). Index, glossary, marks (bookmarks) and contents tabs can be opened and maintained at the same time. Icons in the text can be clicked upon to bring up quizzes and miniature "movies" describing calculation procedures. One can quickly move to any point in the text. A "complete software program called Data Desk" is provided for student calculations. Will such CD's replace books? I doubt it, but the system might work very well for some students, particularly as a self-study aid, and especially with a co-operative course instructor.

Contents:
1. Survival distributions
2. Hazard models
3. Reliability of systems
4. Data plots
5. Censoring and lifetables
6. The product-limit estimator
7. Parametric survival models under censoring
8. Fitting parametric regression models
9. Cox proportional hazards
10. Linear regression with censored data
11. Buckley-James diagnostics and applications

Readership: Senior undergraduate or graduate students of statistics, professional statisticians, engineers, other numerate scientists

Most students beginning post graduate work in statistics soon find that there is a large gap between the comfortable explanations given in undergraduate texts and the very abbreviated messages offered in the statistical literature. It is this gap that the author sets out to fill. However, in his approach he does much more than fill in mathematical details. Ways in which to tackle a proof are suggested before it is given; mathematical results are translated into numerical quantities and the extensive sets of exercises at the end of each chapter are designed to further sharpen the intuition.
The complication of censoring is introduced only after the properties of survival and hazard functions have been firmly established. This leads on naturally to regression models and Cox's proportional hazard model. The computational aspects of these models is not neglected, but again the emphasis is on understanding how the numerical methods work, rather than on how to specify the code.
The final two chapters on least squares regression with censored data, offer something to a readership beyond the beginning graduate student. This detailed exposition, particularly of the Bucklay-James method which can be used when the proportional hazards model is not applicable, is sure to stimulate further research in this area.
Among the books on survival analysis currently available at the graduate level, this book is unique. The insights it freely gives to a fascinating and important branch of statistics can often only otherwise be obtained by much time and effort.

Contents:
Part I: Methods of Computational Statistics
Introduction to Part I
1. Preliminaries
2. Monte Carlo methods for statistical inference
3. Randomization and data partitioning
4. Bootstrap methods
5. Tools for identification of structure in data
6. Estimation of functions
7. Graphical methods in computational statistics
Part II: Exploring Data Density and Structure
Introduction to Part II
8. Estimation of probability density functions using parametric models
9. Nonparametric estimation of probability density functions
10. Structure in data
11. Statistical models in dependencies
APPENDIX A: Monte Carlo Studies in Statistics
APPENDIX B: Software for Random Number Generation
APPENDIX C: Notation and Definitions
APPENDIX D: Solutions and Hints for Selected Exercises

Readership: Students of modern statistics at advanced undergraduate or graduate level

The computer revolution has also revolutionized statistics. New tools, inconceivable prior to the development of modern computing power, have been developed, and these have led to theoretical progress in statistical theory and concepts. Much of this work goes under the name of 'computational statistics', different from 'statistical computing', which is about computational methods for statisticians. This book describes many of the exciting, even revolutionary, developments in computational statistics which have been made over the last two or three decades.
The book has a rather mainstream statistical feel to it: it gives excellent discussions of topics such as boot-strap methods, density function estimation, and multivariate tools such as principal components, clustering and projection pursuit. But it would have benefited from more discussion of various other classes of modern problems and models. I have in mind such areas as Bayesian belief net-works, high dimensional problems, neural networks, and very large sets of data, all issues or tools which have arisen as a consequence of the computer revolution. Of course, to some extent, these are areas which have been explored in data analytic disciplines other than statistics, such as machine learning and data mining, but I agree with the author's statement in his preface: 'I take the view that any method of analyzing data is a statistical method', even if it is sometimes called by some other name.
The book assumes an advanced undergraduate or early graduate level of understanding of mathematical statistics. It does not assume familiarity with any particular software system, but uses fragments from various languages, mainly S-plus. It would provide an excellent grounding for someone beginning to work in this area, but would be usefully supplemented with further reading on other computational statistical tools.

Contents:
1. Production Models: Maximising profits
2. Diet and other input models minimising costs
3. Transportation and assignment models
4. Building larger models
5. Simple sets and indexing
6. Compound sets and indexing
7. Parameters and expressions
8. Linear programs: Variables objectives and constraints
9. Specifying data
10. Database access
11. Modelling commands
12. Display commands
13. Command scripts
14. Interactions with solvers
15. Network linear programs
16. Columnwise formulations
17. Piecewise-linear programs
18. Nonlinear programs
19. Complementary problems
20. Integer linear programs
APPENDIX: AMPL Reference Manual

This text is the manual for AMPL: A Mathematical Programming Language. AMPL is an algebraic modelling language that enables the user to create a computer read-able specification of an optimisation model such as a linear program. From the AMPL website, www.ampl.com, it is possible to download a size-limited version of AMPL. There are other commercially available algebraic modelling systems. AMPL was one of the first sophisticated algebraic modelling systems; it has been developing since 1985. Its command line interface is a reflection of its origins. The authors eschew the use of traditional mathematical notation using AMPLese to explain their example models. The effect is that the user is not encouraged to use a language, mathematics, to communicate his model either to col-leagues who are non-AMPL users or to any algebraic modelling system except AMPL.

Contents:
1. Objectives
2. Least squares computations
3. Finding subsets which fit well
4. Hypothesis testing
5. When to stop?
6. Estimation of regression coefficients
7. Bayesian methods
8. Conclusions and some recommendations

The first edition was reviewed by R.M. Loynes in Short Book Reviews, Volume 10, p. 48. He concluded that it provided "a useful summary and survey of available material in this area of considerable practical importance." So does this second edition. It has in total only nine more text pages. Its main new feature is a separate sixteen page chapter devoted to Bayesian methods, which have been actively developed in the 1990's. Updating has also taken place throughout the text and in the list of references. In this area there are, unfortunately, no easy answers, but the experienced author explains the difficulties inherent in regression selection procedures with great clarity, and provides sensible recommendations wherever feasible. This is an excellent authoritative book which should be in every statistical book collection.

Contents:
1. Introduction
2. Life course data in criminology
3. Nondurable goods index
4. Bone shapes from a paleopathology study
5. Modeling reaction-time distribution
6. Zooming in on human growth
7. Time warping handwriting and weather records
8. How do bones shapes indicate arthritis?
9. Functional models for test items
10. Predicting lip acceleration from electromyography
11. The dynamics of handwriting printed characters
12. A differential equation for juggling

The book complements the same authors' book on "Functional Data Analysis" [Short Book Reviews, Vol. 18, p. 11] with a number of case studies from a wide range of subject areas. Examples are interestingly and well presented. The analysis methods are rooted in the previous book and are essentially descriptive. Most of them are very closely related to the methods that are widely used in the traditional function approximation theory.

Contents:
1. Hilbert spaces and statistics
2. Random processes and time series
3. Estimation of time series parameters
4. Predictions in time series
5. Empirical predictors

The purpose of this book is to provide a unified approach for the estimation of regression parameters when the errors are correlated and stationary. The author also considers prediction using these estimated models. Various methods of estimation, such as ordinary least squares, weighted least squares and maximum likelihood estimation are considered. The book should be considered as a collection of various methods applied to various models. The author uses the models to illustrate the methods described. There is no motivation provided, and the whole book looks a bit dry. Another annoying aspect is the number of times the author uses abbreviations, such as DOOLSE, DOWELSE, WELSE, LRM etc., some of which are new. The book cannot be considered as a text book, nor a research monograph. Despite these, the book should be useful for a person who is working in the areas of estimation and prediction. The approach in the book is very formal.

Contents:
1. Time series following generalizing linear models
2. Regression models for binary times series
3. Regression models for categorical times series
4. Regression models for count times series
5. Other models and alternative approaches
6. State space models
7. Prediction and interpolation
APPENDIX: Elements of Stationary Processes

Readership: Academic (final-year undergraduate and postgraduate students of statistics, econometrics, finance, ecology), industry (banking, insurance)

The book draws together, into a coherent whole, a variety of modern work on time series models. The central theme is the application of generalized linear models to time series. Thus, the traditional framework of a continuous response is extended to a discrete one, Chapters 2, 3 and 4 being devoted to the main examples. The covariates in the regression models are also allowed to form a process, and the necessary technical machinery for analysis comprises partial likelihood, filtrations (the increasing sequence of process histories) and the associated martingales. How-ever, the style is very accessible without much knowledge of stochastic processes. The last three chapters cover integer-value series, hidden Markov chains, switching regimes, ARCH models, state-space models and prediction for spatial data.
The algebraic details of Generalized Linear Models and iteratively reweighted least-squares are presented at some length in Chapter 1, but devotees will not complain. Also, the method of Generalized Estimating Equations is described and applied somewhat uncritically, i.e. without any reference to its pitfalls. These quibbles aside, the book gives an excellent introduction to a widely applicable field of methods. Moreover, plenty of real data examples are worked through in detail, and guidance is given on computation, using SPLUS in particular. Also, there are exercises and complements at the end of each chapter. Overall, I would say that the book does what it sets out to do very well and will be a useful reference for both practitioners and researchers.

Contents:
1. Introduction
Part I: Deterministic Control
2. Deterministic time delay systems
3. Stability and stabilizability
4. Robust stability and robust stabilizability
5. H8 control and filtering
6. Robust H8 control, filtering and guaranteed cost control
Part II: Stochastic Control
7. Stochastic time delay systems
8. Stability and stabilizability of Markov jump systems
9. Robust stability and stabilizability of jump linear uncertain systems with time delay
10. H8 control and filtering problems for Markov jump systems with time delay
11. Robust H8 and guaranteed cost control for jump linear systems with time delay
12. Nonlinear stochastic control problem
APPENDIX A: Linear Matrix Inequality and Preliminary Lemmas
APPENDIX B: Matrix Inversion Formulas
APPENDIX C: Kronecker Product
APPENDIX D: Markov Processes Probability

This book gives a comprehensive treatment of stability and stabilization of time delay systems. Part I deals with deterministic systems and Part II deals with stochastic systems. This topic has been central to recent developments in control theory. Many interesting research problems remain and hence this book will be of interest to the research community in this field. This basic tool used throughout is that of Linear Matrix Inequalities (LMI's). Software support is also provided for the examples in the book. The book is clearly written and is recommended to those having an interest in either robust control or stochastic systems, or both.

Contents:
Part I: Preliminaries
1. Colouring preliminaries
2. Probabilistic preliminaries
Part II: Basic Probabilistic Tools
3. The first moment method
4. The Lovász local lemma
5. The Chernoff bound
Part III: Vertex Partitions
6. Hadwiger's conjecture
7. A first glimpse of total colouring
8. The strong chromatic number
9. Total colouring revisited
Part IV: A Naïve Colouring Procedure
10. Talagrand's inequality and colouring sparse graphs
11. Azuma's inequality and a strengthening of Brooks' theorem
Part V: An Iterative Approach
12. Graphs with girth at least five
13. Triangle-free graphs
14. The list colouring conjecture
15. Part VI: A Structural Decomposition
16. The structural decomposition
17. ù, ? and ÷
18. Near optimal total colouring I: Sparse graphs
19. Near optimal total colouring II: General graphs
20. Part VII: Sharpening our Tools
21. Generalizations of the local lemma
22. A closer look at Talagrand's inequality
23. Part VIII: Colour Assignment via Fractional Colouring
24. Finding fractional colourings and large stable sets
25. Hard-core distributions on matchings
26. The asymptotics of edge colouring multigraphs
27. Part IX: Algorithmic Aspects
28. The method of conditional expectations
29. Algorithmic aspects of the local lemma

The probabilistic method in graph theory was initiated by Paul Erdös in 1947, when he discovered an elegant probability argument to show the existence of graphs which do not contain complete subgraphs of a certain size. Since then, the method has evolved into a flourishing subdiscipline. This book is an introduction to this powerful method. It assumes no knowledge of probability theory and reviews the main portion of the relevant theory in the early chapters. The second goal of the book is to give a unified treatment of results in the theory of graph colouring. Many problems of "scheduling" that occur in "real life" can be formulated in graph theoretical terms and thus, these questions have practical applications.
This book is well-written and brings the researcher to the frontiers of an exciting field.

Contents:
Part I: Green, Brown, and Probability
1. Green's ideas
2. Probability and potential
3. Process
4. Random time
5. Markov property
6. Brownian construct
7. The trouble with boundary
8. Return to Green
9. Strong Markov property
10. Transience
11. Last but not least
12. Least energy
Addenda
Notes
References
Some Chronology
Appendices
Part II: Brownian Motion on the Line
Generalities
1. Exit and return
2. Time and place
3. A general method
4. Drift
5. Dirichlet and Poisson problems
6. Feynman-Kac functional
Part III: Stopped Feynman-Kac Functional
1. Introduction
2. The results
3. The connections

This is a new edition of a book designed to provide a conversational approach to the interface between Green's boundary-value problem, first exit of Brownian motion and the Feynman-Kac functional. It requires a reasonable level of knowledge of differential equations, functional analysis and Markov Processes.

Contents:
1. Modeling tools for financial options
2. Generating random numbers with specified distributions
3. Numerical integration of stochastic differential equations
4. Finite differences and standard options
5. Finite-element methods
6. Pricing of exotic options

Numerous textbooks already exist on the mathematics underlying finance. Over recent years, we have been able to welcome texts concentrating more on the numerics needed to transport the fundamental theory to the end-user. This book aims very much at providing the latter. It does it in a very readable way. After a brief introduction to the fundamentals from finance, the author presents the numerical theory and algorithms needed to turn analytical formulae to numbers. As it stands, the book makes an ideal (post) graduate course on computational finance. Numerous examples and exercises make the text also very useful for self-study.

Contents:
0. Introduction
1. Cash account evolution
2. Stock price evolution
3. European style stock options
4. Stock market statistics
5. Implied volatility for European options
6. American style stock options
7. Optimal portfolio rules
8. Advanced trading strategies

Readership: Students and professionals with at least undergraduate level mathematics and probability and with ready access to Mathematica

This book is intended for "hands-on" use in a laboratory with Mathematica available and indeed is so liberally interlaced with statements in Mathematica that in places it is not easy to read. It covers most of the basic elements of financial mathematics from the perspective of symbolic computation; the generalized Black-Scholes model, option pricing for European and American options, portfolio optimization and ends with some interesting and more advanced topics in advanced trading strategies.

Contents:
1. Stochastic portfolio theory
2. Stock market behaviour and diversity
3. Functionally generated portfolios
4. Portfolios of stocks selected by rank
5. Stable models for the distribution of capital
6. Performance of functionally generated portfolios
7. Applications of stochastic portfolio theory
APPENDIX A: Evaluation of Local Times

The dominant approach used in mathematical finance is based on dynamic asset pricing, a theory which is itself based on market equilibrium and absence of arbitrage. This short book presents an alternative, descriptive theory, 'consistent with either equilibrium or disequilibrium, with either arbitrage or no arbitrage'. This theory, stochastic port-folio theory, can be used for portfolio optimization and performance analysis. The author is Chief Investment Officer of INTECH, where the ideas described in this book form the basis for the investment strategies. The book is a monograph rather than a text.

Contents:
Part I: Introduction
1. Mathematical finance background
2. Stochastic processes and martingales
3. Financial markets
Part II: Modelling and Pricing in Interest-Rate Markets
4. Interest-rate markets
5. Interest-rate derivatives
Part III: Measuring and Managing Interest-Rate Risk
6. Risk measures
7. Risk management

This book is essentially about two main topics: first of all about the mathematics of interest-rate markets, and secondly about risk management issues in such markets. The former is dealt with in a mathematical way, and by now can be found (in more or less depth) in several textbooks. The latter offers a primer of risk management issues in interest-rate markets, written in a mathematical language but with an eye for applications. This second part does not yield the full spectrum of ALM applications but helps the mathematical reader to gain a first introduction to this field. All in all, an interesting book which offers first insight into the world of true money-market risk management. By keeping content and length well balanced it will be easy to base a course on it.

Contents:
1. The history and philosophy of credit scoring
2. The practice of credit scoring
3. Economic cycles and lending and debt patterns
4. Statistical methods for building credit scorecards
5. Nonstatistical methods for scorecard development
6. Behavioural scoring models or repayment and usage behaviour
7. Measuring scorecard performance
8. Practical issues of scorecard development
9. Implementation and areas of application
10. Application of scoring in other areas of lending
11. Applications of scoring in other areas
12. New ways to build scorecards
13. International differences
14. Profit scoring, risk-based pricing, and securitization

Readership: Researchers, students and bankers who wish to understand the technical side of retail credit scoring

The term 'credit scoring' describes the models and methods used by financial organizations to predict consumer behaviour and to assist these organizations in making decisions. The concepts and tools are used across the retail finance sector, for a wide variety of products, including bank loans, mortgage, car finance, current account monitoring, and so on. It is no exaggeration to say, as the authors do in the opening sentence of their preface, that 'credit scoring is one of the most successful applications of statistical and operations research modelling in finance and banking'.
Although consumer credit scoring is a very active research area, with a growing research literature, the pedagogical literature is limited. The book by Lewis (An Introduction to Credit Scoring, Athena Press, 1992) is rather outdated, the recent book by McNab and Wynn (Principles and Practice of Consumer Credit Risk Management, CIB Publishing, 2000) does not deal with the technical aspects in any detail, and Mays' book (Handbook of Credit Scoring, Glenlake Publishing Co., 2001) is an edited collection of contributions, and not ideal as a text. The book under review fills this gap perfectly. It is a comprehensive account of the technical aspects of score-card construction. It presents the material in a clear and accessible way – at a level such that a graduate in a quantitative discipline should be able to handle without difficulty. The authors' affiliations span the continuum from leading academic researchers in the area to practicing bankers, so that they are able nicely to relate the material to actual practice.
Although there are no exercises – which would have enhanced its use for teaching – the book does include a CD containing data which can be used to construct score-cards, and would be suitable as exercise material for a course. However, with only 1226 cases on 14 variables it is very small, compared with the sizes of sets of data which do arise in the industry. It would have been more useful if a larger, and perhaps more comprehensive set of data could have been included.
In conclusion, I thoroughly recommend this book. It is ideal for a course on score-card construction, and I would certainly use it for such a course.

Contents:
1. Introduction
Part I: Computational Finance: Theory
2. Notation and basic definitions
3. Continuous time finance
4. Scenario-based evaluation and uncertainty
Part II: Algorithms for Uncertain Volatility Models
5. A lattice framework
6. Algorithms for vanilla options
7. Algorithms for barrier options
8. Algorithms for American Options
9. Exotic volatility scenarios
Part III: Object-Oriented Implementation
10. The architecture of Mtg
11. The class hierarchy of MtgLib – external
12. The class hierarchy of MtgLib – internal
13. Extensions for Monte-Carlo pricing and calibration
APPENDIX A: The Network Application MtgClt/MtgSvr
APPENDIX B: The Scripting Language MtgScript
APPENDIX C: Mathematica Extensions

Readership: Graduate students, researchers and practitioners who wish to study advanced aspects of volatility risk in portfolios of vanilla and exotic options

The assumption of constant volatility in financial models of derivative prices does not conform with empirical observation. The uncertain volatility models approach to tackling this discrepancy finds an upper bound for the value of a portfolio over volatility surfaces chosen from a given set. Uncertain volatility scenario models generalize this approach by determining some of the uncertain coefficients in such a way that an objective, called a scenario is fulfilled. The book begins with some fairly general theory, and then progresses through details of the algorithms, to implementation issues, culminating, in the second part of the book, with a software system, Mtg (for Martingale) for uncertain volatility models. The next part emphasizes structure over implementation details, and the book includes object-oriented source code for Mtg written in C++ and Java on a CD.
The book is accessible to mathematics graduates. It is not, of course, a statistics book.

Title	UNE HISTOIRE DE LA COMPTABILITÉ NATIONALE.
Author	A.Vanoli.
Publisher	Paris: La Découverte (Manuels Repères), 2002, pp. 650, €38.00.

Reviewer:
Institute	Statistique Canada
Place	Ottawa, Canada
Name	J. Delisle

Title	WRITING THE HISTORY OF MATHEMATICS: ITS HISTORICAL DEVELOPMENT.
Author	J .W .Dauben and C.J. Scriba (Eds.).
Publisher	Basel: Birkhäuser, 2002, SFr186/€ 119.63 Cloth; SFr118/€73.83 Paper.

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

Title	STATISTICS WITH APPLICATIONS IN BIOLOGY AND GEOLOGY.
Author	P. Blaesild and J. Granfeldt.
Publisher	Boca Raton, Florida: Chapman and Hall/CRC, 2003, pp. ix + 555, US$59.95 /£29.99.

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

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

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

Reviewer:
Institute	University of Waterloo
Place	Waterloo, Canada
Name	J.F. Lawless

Reviewer:
Institute	Tropical Medicine
Place	London.U.K.
Name	B.L. De Stavola

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

Reviewer:
Institute	University of St Andrews
Place	St Andrews, U.K.
Name	A.W. Kemp

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

Reviewer:
Institute	Limburgs Universitair Centrum
Place	Diepenbeek, Belgium
Name	N.D.C. Veraverbeke

Reviewer:
Institute	University of Cape Town
Place	Rondebosch, South Africa
Name	J.M. Juritz

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

Reviewer:
Institute	University of Texas
Place	Austin, U.S.A.
Name	P.W.M. John

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

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

Reviewer:
Institute	GlaxoSmithKline
Place	Harlow, U.K.
Name	V.V. Anisimov

Reviewer:
Institute	University College London
Place	London, U.K.
Name	V.S.Isham

Reviewer:
Institute	EPFL - DMAMA (Ecublens)
Place	Lausanne, Switzerland
Name	S. Morgenthaler

Reviewer:
Institute	ETH-Zürich
Place	Zürich Switzerland
Name	P.A.L. Embrechts

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

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

Reviewer:
Institute	McGill University
Place	Montreal, Canada
Name	J.O. Ramsey

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

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

Reviewer:
Institute	CSIRO Mathematical and Information Sciences
Place	Melbourne, Australia
Name	R.G. Jarrett

Reviewer:
Institute	GlaxoSmithKline,
Place	U.S.A.
Name	V.V. Fedorov

Reviewer:
Institute	ETH - Zürich
Place	Zürich, Switzerland
Name	H. Furrer

Reviewer:
Institute	Bank of America
Place	Chicago U.S.A.
Name	D. Eddelbuettel

Reviewer:
Institute	Swiss Federal Institute of Technology
Place	Zürich, Switzerland
Name	P. Schiltknecht

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