ISI - INTERNATIONAL STATISTICAL INSTITUTE


Back to SBR-index

Short Book Reviews

Reviews 2005


WEIBULL MODELS. D.N. Prabhakar Murthy, M. Xie and R. Jiang.
COMPUTATIONAL METHODS IN STATISTICS AND ECONOMETRICS. H. Tanizaki.
STATISTICS FOR MICROARRAYS. S. Wit and J. McClure.
ANALYZlNG MICROARRAY GENE EXPRESSION DATA. G.J. McLachlan, K.-A. Do and C. Ambroise.
DESIGN AND ANALYSIS OF DNA MICROARRAY INVESTIGATIONS. R.M. Simon, E.L. Korn, L.M. McShave, M.D. Radmacker, G.W. Wright and Y. Zhao.
BAYESIAN NONPARAMETRICS VIA NEURAL NETWORKS. H.K.H. Lee.
INTRODUCTION TO RANDOM TIMES AND QUANTUM RANDOMNESS. New Edition. K.L. Chung and J.-C. Zambrini.
DISTRIBUTION THEORY OF RUNS AND PATTERNS AND ITS APPLICATIONS. A FINITE MARKOV CHAIN IMBEDDING APPROACH. J.C. Fu and W.Y.W. Lou.
ELEMENTS OF THE RANDOM WALK: AN INTRODUCTION FOR ADVANCED STUDENTS AND RESEARCHERS. J. Rudnick and G. Gaspari.
WHY STOCK MARKETS CRASH. Critical Events in Complex Financial Systems. D. Sornette.
THE CONCEPTS AND PRACTICE OF MATHEMATICAL FINANCE. M.S. Joshi.
DYNAMICS OF MARKETS. ECONOPHYSICS AND FINANCE. J.L. McCauley.
HANDBOOK OF COMPUTATlONAL AND NUMERICAL METHODS IN FINANCE. S.T. Rachev (Ed.).
C++ DESIGN PATTERNS AND DERIVATIVE PRICING. M.S. Joshi.
CONVEX OPTIMIZATION. S. Boyd and L. Vandenberghe.
INTRODUCTION TO LOGISTIC SYSTEMS PLANNING AND CONTROL. G. Ghiani, G. Laporte and R. Musmanno. With a Foreword by M. Goetschalck.
THE CROSS-ENTROPY METHOD. A Unified Approach to Combinatorial Optimization, Monte-Carlo Simulation, and Machine Learning. R.Y. Rubinstein and D.P. Kroese.
THE NATURE OF SCIENTIFIC EVIDENCE: STATISTICAL, PHILOSOPHICAL, AND EMPIRICAL CONSIDERATIONS. M.L. Taper and S.R. Lele (Eds.).
R.L. MOORE, MATHEMATICIAN AND TEACHER. J. Parker.
?: A BIOGRAPHY OF THE WORLD'S MOST MYSTERIOUS NUMBER. A. Posamentier and I. Lehmann. With an Afterword by H.A. Hauptman.
A TOUR THROUGH MATHEMATICAL LOGIC. R.S. Wolf.
DATA ANALYSIS AND GRAPHICS USING R: AN EXAMPLE-BASED APPROACH. J. Maindonald and J. Braun.
DATA ANALYSIS OF ASYMMETRIC STRUCTURES. Advanced Approaches to Computational Statistics. T. Saito and H. Yadohisa.
MATHEMATICAL LABORATORIES FOR MATHEMATICAL STATISTICS: EMPHASIZING SIMULATION AND COMPUTER INTENSIVE METHODS. J.A. Baglivo.
PROBABILITY AND COMPUTING: RANDOMIZED ALGORITHMS AND PROBABILISTIC ANALYSIS. M. Mitzenmacher and E. Upfal.
STATISTICAL METHODS FOR SPATIAL DATA ANALYSIS. O. Schabenberger and C.A. Gotway.
STEREOLOGY FOR STATISTICIANS. A. Baddeley and L.B. Vedel Jensen.
AN R AND S-PLUS® COMPANION TO MULTIVARIATE ANALYSIS. B.S. Everitt.London: Springer-Verlag.
AN INTRODUCTION TO MULTIVARIATE DATA ANALYSIS. T.F. Cox.
GENERALIZED LATENT VARIABLE MODELING: MULTILEVEL, LONGITUDINAL AND STRUCTURAL EQUATION MODELS. A. Skrondal and S. Rabe-Hesketh.
STATISTICAL METHODS IN BIOINFORMATICS: AN INTRODUCTION, 2nd edition. W.J. Ewens and G.R. Grant.
STATISTICAL CONCEPTS AND APPLICATIONS IN CLINICAL MEDICINE. J. Aitchison, J.W. Kay and I.J. Lauder.
THE EVALUATION OF SURROGATE ENDPOINTS. T. Burzykowski, G. Molenberghs and M. Buyse (Eds.).
REGRESSION METHODS IN BIOSTATISTICS: LINEAR, LOGISTIC, SURVIVAL AND REPEATED MEASURES MODELS. E. Vittinghoff, D.V. Glidden, S.C. Shiboski, C.E. McCulloch.
ELEMENTARY STATISTICAL QUALITY CONTROL, 2nd edition. J.T. Burr.
PERMUTATION, PARAMETRIC AND BOOTSTRAP TESTS OF HYPOTHESES, 3rd edition. P. Good.
EXTREME VALUE AND RELATED MODELS WITH APPLICATIONS IN ENGINEERING AND SCIENCE. E. Castillo, A.S. Hadi, N. Balakrishnan and J.M. Sarabia.
STATISTICS OF EXTREMES: THEORY AND APPLICATIONS. J. Beirlant, Y. Goegebeur, J. Segers and J. Teugels. With contributions from D. De Waal and C. Ferro.
LAWS OF SMALL NUMBERS: EXTREMES AND RARE EVENTS, 2nd, revised and extended edition. M. Falk, J. Hüsler and R.-D. Reiss.
LARGE DEVIATIONS AND METASTABILITY. E. Olivieri and M. Eulália Vares.
DISCRETE-TIME MARKOV JUMP LINEAR SYSTEMS. O.L.V. Costa, M.D. Fragoso and R.P. Marques.
APPLIED TIME SERIES ECONOMETRICS. H. Lütkepohl and M. Krätzig (Eds.).
AN INTRODUCTION TO CONTINUOUS TIME STOCHASTIC PROCESSES. THEORY, MODELS AND APPLICATIONS TO FINANCE, BIOLOGY, AND MEDICINE. V. Capasso and D. Bakstein.
BRANCHING PROCESSES: VARIATION, GROWTH AND EXTINCTION OF POPULATIONS. P. Haccou, P. Jagers and V.A. Vatutin.
OPTIMIZATION. K. Lange.
NIKO'S NATURE: THE LIFE OF NIKO TINBERGEN AND HIS SCIENCE OF ANIMAL BEHAVIOUR. H. Kruuk.
MUSINGS OF THE MASTERS, AN ANTHOLOGY OF MATHEMATICAL REFLECTIONS. R. Ayoub (Ed.).
GRAPHS, ALGORITHMS, AND OPTIMIZATION. W. Kocay and D.L. Krecher.
INSURANCE RISK AND RUIN. D.C.M. Dickson.
CAPITAL MARKET INSTRUMENTS: ANALYSIS AND VALUATION, 2nd edition. M. Choudhry, D. Joannas, R. Pereira, and R. Pienaar.
LUCK, LOGIC AND WHITE LIES -THE MATHEMATICS OF GAMES. J. Bewersdorff. Translated by D. Kramer.
FORECASTING PRODUCT LIABILITY CLAIMS: Epidemiology and Modeling in the Manville Asbestos Case. E. Stallard, K.G. Manton and J.E. Cohen.
SAUNDERS MAC LANE. A MATHEMATICAL AUTOBIOGRAPHY. S. Mac Lane.
THE GRAMMAR OF GRAPHICS, 2nd edition. L. Wilkinson.
INFÉRENCE ET PRÉVISION EN GRANDES DIMENSIONS. D. Bosq.
A MODERN INTRODUCTION TO PROBABILITY AND STATISTICS. UNDERSTANDING WHY AND HOW. F.M. Dekking, C. Kraaikamp, H.P. Lopuhaä and L.E. Meester.
PROBABILITY: A GRADUATE COURSE. A. Gut.
EPIDEMIOLOGY: Study Design and Data Analysis, 2nd edition. M. Woodward.
ANALYZING ENVIRONMENTAL DATA. W.W. Piegorsch and A.J. Bailer.
WEIGHT-OF-EVIDENCE FOR FORENSIC DNA PROFILES. D.J. Balding.
STATISTICS FOR EXPERIMENTERS: DESIGN, INNOVATION AND DISCOVERY, 2nd edition. G.E.P. Box, J.S. Hunter and W.G. Hunter.
STATISTICAL DESIGN OF EXPERIMENTS WITH ENGINEERING APPLICATIONS. K. Rekab and M. Shaikh.
EXPERIMENTAL DESIGN FOR FORMULATION. W.F. Smith.
TESTING STATISTICAL HYPOTHESES, 3rd edition. E.L. Lehmann and J.P. Romano.
MATHEMATICAL STATISTICS WITH APPLICATIONS. A.S. Kapadia, W. Chan and L. Moyé.
CONSTRUCTING MEASURES: AN ITEM RESPONSE MODELING APPROACH. M. Wilson.
CORRESPONDENCE ANALYSIS AND DATA CODING WITH JAVA AND R. F. Murtagh.
CLUSTERING FOR DATA MINING: A DATA RECOVERY APPROACH. B. Mirkin.
MINING IMPERFECT DATA: DEALING WITH CONTAMINATION AND INCOMPLETE RECORDS. R.K. Pearson.
APPLIED LINEAR REGRESSION, 3rd edition. S. Weisberg.
FUNCTIONAL DATA ANALYSIS, 2nd edition. J.O. Ramsey and B.W. Silverman.
IMAGE PROCESSING AND JUMP REGRESSION ANALYSIS. P. Qiu.
STATISTICS FOR FISSION TRACK ANALYSIS. R.F. Galbraith.
THE TEN MOST WANTED SOLUTIONS IN PROTEIN BIOINFORMATICS. A. Tramontano.
BEYOND BETA. Other Continuous Families of Distributions with Bounded Support and Applications. S. Kotz and J.R. van Dorp.
STATISTICAL AND INDUCTIVE INFERENCE BY MINIMUM MESSAGE LENGTH. C.S. Wallace.
STRUCTURAL ASPECTS IN THE THEORY OF PROBABILITY. A PRIMER IN PROBABILITIES ON ALGEBRAIC-TOPOLOGICAL STRUCTURES. H. Heyer.
FLOWGRAPH MODELS FOR MULTISTATE TIME-TO-EVENT DATA. A.K. Huzurbazar.
RELIABILITY AND RISK MODELS: SETTING RELIABILITY REQUIREMENTS. M.T. Todinov.
MEASUREMENT THEORY AND PRACTICE: THE WORLD THROUGH QUANTIFICATION. D.J. Hand.
GRAPHIC DISCOVERY. A Trout in the Milk and Other Visual Adventures. H. Wainer.
EXPLORATORY DATA ANALYSIS WITH MATLAB. W. Martinez and A.R. Martinez.
FINANCIAL DERIVATIVES, 2nd edition. R.W. Kolb.
INTRODUCTION TO MODERN PORTFOLIO OPTIMIZATION WITH NuOPTtm, S-PLUSâ AND S+Bayestm. B. Scherer and R.D. Martin.
ROBUST LIBOR MODELLING AND PRICING OF DERIVATIVE PRODUCTS. J. Schoenmakers.
FINANCIAL DERIVATIVES: PRICING, APPLICATIONS AND MATHEMATICS. J. Baz and G. Chacko.
STATISTICAL ANALYSIS AND DATA DISPLAY: AN INTERMEDIATE COURSE WITH EXAMPLES IN S-PLUS, R, AND SAS. R.M. Heiberger and B. Holland.
LINEAR MODELS WITH R. J.J. Faraway.
MONTE CARLO STATISTICAL METHODS, 2nd edition. C.P. Robert and G. Casella.
PLANNING, CONSTRUCTION AND STATISTICAL ANALYSIS OF COMPARATIVE EXPERIMENTS. F.G. Giesbrecht and M.L. Gumpertz.
ANALYSIS OF VARIANCE FOR RANDOM MODELS, Volume II: Unbalanced Data. Theory, Methods, Applications and Data Analysis. H. Sahai and M.M. Ojeda.
DESIGN AND ANALYSIS OF ACCELERATED TESTS FOR MISSION CRITICAL RELIABILITY. M.J. LuValle, B.G. Lefevre and S. Kannan.
ANALYSIS AND MODELLING OF SPATIAL ENVIRONMENTAL DATA. M. Kanevski and M. Maignan.
HEIRARCHICAL MODELLING AND ANALYSIS FOR SPATIAL DATA. S. Banerjee, B.P. Carlin and A.E. Gelfand.
ADVANCED DISTANCE SAMPLING. S.T. Buckland, D.R. Anderson, K.P. Burnham, J.L. Laake, D.L. Borchers and L. Thomas.
EXPLANATORY ITEM RESPONSE MODELS: A GENERALIZED LINEAR AND NONLINEAR APPROACH. P. DeBoeck and M. Wilson (Eds.).
TEST EQUATING, SCALING, AND LINKING, 2nd edition. M.J. Kolen and R.L. Brennan.
MULTIVARIATE STATISTICAL METHODS, 4th edition. D.F. Morrison.
APPLIED ECONOMETRIC TIME SERIES, 2nd edition. W. Enders.
SKEW-ELLIPTICAL DISTRIBUTIONS AND THEIR APPLICATIONS. A Journey Beyond Normality. M.G. Genton (Ed.).
back to index


Title WEIBULL MODELS.
Author D.N. Prabhakar Murthy, M. Xie and R. Jiang.
Publisher Hoboken, New Jersey: Wiley. 2004, pp. xvi + 383, £58.95.

Contents:
PART A: Overview
1. Overview
2. Taxonomy for Weibull models
PART B: Basic Weibull Model
3. Model analysis
4. Parameter estimation
5. Model selection and validation
PART C: Types I and II Models
6. Type I Weibull models
7. Type II Weibull models
PART D: Type III Models
8. Type III(a) Weibull models
9. Type III(b) Weibull models
10. Type III(c) Weibull models
11. Type III(d) Weibull models
PART E: Types IV to VII Models
12. Type IV Weibull models
13. Type V Weibull models
14. Type VI Weibull models (multivariate models)
15. Type VII Weibull models
PART F: Weibull modeling of data
16. Weibull modeling of data
PART G: Applications in reliability
17. Modeling product failures
18. Product reliability and Weibull failure models

Readership: "Everyone from reliability engineers to applied statisticians involved with reliability and survival analysis"

The main expressed aim is to develop a taxonomy for the various forms that this widely applied probability model can take through transformations and generalization, and to integrate the literature. Sixty pages of standard theory are followed by one hundred and eighty in which the model is split into seven types and over forty sub-types. Each chapter has exercises asking for analyses to be repeated on sample sets of data. There is a large reference section.
Model selection is driven by the Weibull probability plot of y = ln (-ln R(t)) on x = Int, where R(t) is the survivor function. There is no attempt at comparison outside the Weibull stable.
The index is hopeless. "Weibull probability plot" gives pages 95 and 99; "WPP" gives 53, 55, ...; "Plot, Weibull" gives 87. The plot is actually described on page 12. Under "Well" we have "well mixed" and "well separated", both on 169 and 177, referring to Models III(a)-1 and III(a)-2. "Distribution function" directs us to Type III(c)-1 in Chapter 10.
A search for "profile likelihood", used in estimation, yields nothing, "likelihood" gives only "likelihood function" and "likelihood ratio test", the latter in respect of its use with Model VII(a)-3. The references include D.R. Cox's paper "Partial likelihood", but there is no author index to enable me to trace it.
Pedants might also question "Baye's (sic) estimate" (index and 269) and "percentile (also referred to as fractile or quantile)" and "0.50 percentile is called the median".

Reviewer:
Institute Imperial College of Science, Technology and Medicine
Place London, U.K.
Name R. Coleman

Back to index


Title COMPUTATIONAL METHODS IN STATISTICS AND ECONOMETRICS.
Author H. Tanizaki.
Publisher New York: Dekker. 2004, pp. xix + 494 + CD, US$150.00/£86.00.

Contents:
1. Elements of Statistics
PART I: Monte Carlo Statistical Methods
2. Random number generation I
3. Random number generation II
PART II: Selected Applications of Monte Carlo Methods
4. Bayesian estimation
5. Bias correction of OLSE in AR models
6. State space modeIing
PART III: Nonparametric Statistical Methods
7. Difference between two-sample means
8. Independence between samples

Readership: Researchers and mature students of statistics and econometrics who seek computational tools to support the evaluation and application of statistical methodologies that could otherwise be intractable

Monte Carlo concepts have long been a respectable approach to difficult integration problems, and have helped to overcome early criticisms of Bayesian methods as impractical. In the first half of this interesting book, the author develops sub-routines in Fortran and C to generate random samples from a large variety of discrete and continuous distributions, and then in the second half of the book he applies these sub-routines to illustrate Bayesian estimation, to compare the power of various estimators, and to obtain small sample properties of a number of nonparametric methods.
Chapter 1 is a summary of a first graduate course in mathematical statistics, which should be enough to arm the reader to tackle the rest of the book.
All of the subroutines are given in the book together with their theoretical bases, and are provided as well on a CD which is included with the book. The sub-routines are a useful addition to the toolbox for researchers who need to generate random numbers from unusual distributions to solve their problems.

Reviewer:
Institute Brookfield,
Place U.S.A.
Name C.A. Fung

Back to index


Title STATISTICS FOR MICROARRAYS.
Author S. Wit and J. McClure.
Publisher Chichester, U.K.: Wiley. 2004, pp. xi + 265, £45.00.

Contents:
1. Preliminaries
PART I: Getting Good Data
2. Set-up of a microarray experiment
3. Statistical design of microarrays
4. Normalisation
5. Quality assessment
6. Microarray myths: Data
PART II: Getting Good Answers
7. Microarray discoveries
8. Differential expression
9. Predicting outcomes with gene expression profiles
10. Microarray myths: Inference

Readership: Statisticians, biostatisticians, bioinformaticians, numerate biologists
and clinicians.

This book addresses the range of questions and problems that biostatisticians and bioinformaticians face when analyzing genomic data generated by recently developed microarray(biochip)-based technologies. It focuses specifically on microarray experiments that generate measurements of expression levels of thousands of genes under various experimental conditions.
This clear and synthetic text is aimed deliberately at an applied minded audience and the numerous examples help understanding of the techniques. In contrast to recent edited books on the subject that present often a somewhat disjointed collection of topics and authors, this book adopts a structured and pedagogical approach with a focus on conveying the benefits of a systematic statistical approach in this domain. Thus, it gives an up-to-date and personal account of methods of analysis, and lively advice on best use and potential misuse. Particularly noteworthy are the chapters on differential expression and prediction. The accompanying set of R routines which makes the book self-contained is valuable. This book is an excellent reference text for any researcher interested in the analysis of transcriptomic data.

Reviewer:
Institute Imperial College of Science, Technology and Medicine
Place London, U.K.
Name S. Richardson

Back to index


Title ANALYZlNG MICROARRAY GENE EXPRESSION DATA.
Author G.J. McLachlan, K.-A. Do and C. Ambroise.
Publisher Hoboken, New Jersey: Wiley. 2004, pp. xx + 320, £52.95.

Contents:
1. Microarrays in gene expression studies
2. Cleaning and normalization
3. Some cluster analysis methods
4. Clustering of tissue samples
5. Screening and clustering of genes
6. Discriminant analysis
7. Supervised classification of tissue samples
8. Linking microarray data with survival analysis

Readership: "biologists who will undertake the statistical analyses of their own experimental microarray data and biostatisticians entering the field of microarray gene expression data analysis."(Preface)

Data analysis methods for the field of microarray gene expression are the primary focus of this book. It begins with two background chapters that will be particularly useful to statisticians new to microarray data. In addition to providing a reasonably comprehensive overview of the range of statistical and computational techniques at use in the field, the authors make a good case for the advantages of some approaches over others, particularly emphasizing mixture models, a principal component-based approach known as gene shaving, and support vector machines. They also discuss methods to address the biases introduced by multiple testing and overfitting with many genes. Inherent in the presentation are useful ways to combine techniques to address the biological questions of interest and some good ideas concerning strategies for the analysis of these complex data. Throughout, the methods are applied and compared in a number of well-known sets of data from the literature, all concerned with clinically-oriented investigations of human cancer. The references are an excellent source of current statistical work in what is a rapidly evolving field. A helpful web page associated with the book (www.maths.uq.edu.au/~rbean/book/ ) points to the authors' software and the sets of data analyzed.
Although biologists analyzing their own data would learn a great deal from this book about appropriate use of various methods, these are advanced multivariate methods usually taught at the graduate level in statistics and biostatistics programs. The book would serve as a very good resource for a graduate topics seminar in advanced applied statistics.

Reviewer:
Institute University of Toronto
Place Toronto, Canada
Name S.B. Bull

Back to index


Title DESIGN AND ANALYSIS OF DNA MICROARRAY INVESTIGATIONS.
Author R.M. Simon, E.L. Korn, L.M. McShave, M.D. Radmacker, G.W. Wright and Y. Zhao.
Publisher New York: Springer-Verlag. 2004, pp. x + 199, US$59.95.

Contents:
1. Introduction
2. DNA microarray technology
3. Design of DNA microarray experiments
4. Image analysis
5. Quality control
6. Array normalization
7. Class comparison
8. Class prediction
9. Class discovery

APPENDIX A: Basic Biology of Gene Expression
APPENDIX B: Description of Gene Expression Datasets
APPENDIX C: BRB Array Tools

Readership: Biomedical researchers using statistical tools

The design and analysis of microarray experiments presents a difficult and important endeavour for modern statistics. Over the past five to ten years, some old tools have been adapted to this area, and some newer tools have been developed. The toolbox has finally begun to gel. This book presents a concise overview of many useful statistical approaches to microarray data, running the gamut from design of experiments, image analysis of the raw data, data normalization and analysis. This broad range of topics makes the book unique and useful. The authors are all experienced researchers and practioners in this area.
At less than two hundred pages, however, the depth of the book is (understandably) lacking in places. For example, there is discussion of diagonal linear discriminant analysis and there is no mention of nearest centroids (an equivalent method) or nearest shrunken centroids, a useful generalization with automatic feature selection. And support vector machines, which are widely used (deservedly or not) are given less than half a page.
These qualms aside, this book is a valuable resource for microarray experimenters. I enjoyed it and recommend it for anyone working in this area.

Reviewer:
Institute Stanford University
Place Stanford, U.S.A.
Name R. Tibshirani

Back to index


Title BAYESIAN NONPARAMETRICS VIA NEURAL NETWORKS.
Author H.K.H. Lee.
Publisher Philadelphia: Society for Industrial and Applied Mathematics. 2004, pp. x + 96, US$42.50.

Contents:
1. Introduction
2. Nonparametric models
3. Priors for neural networks
4. Building a model
5. Conclusions

Readership: Students and practitioners of statistics and computational sciences

This book presents neural-network models in a statistical framework in order to expel their myth that they are at worst magical black boxes and at best machine learning algorithms. As the author discusses, there is a definite probability model behind neural networks and begins by showing that neural networks are nothing more than a statistical model for nonparametric regression and classification.
The text has been developed from the author's 1998 Ph.D. thesis and whilst not providing an all-inclusive introduction to the field, does present material in a self-contained manner with references provided for further details of the many issues not directly addressed.
Two examples, one on ozone pollution and the other on loan applications, are used throughout the text. The author does not assume that the reader has any previous knowledge of neural networks but introduces topics in a self-contained way. However, the reader is assumed to have a basic understanding of the Bayesian approach, mathematical statistics and linear regression that might limit its initial readership.

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

Back to index


Title INTRODUCTION TO RANDOM TIMES AND QUANTUM RANDOMNESS. New Edition.
Author K.L. Chung and J.-C. Zambrini.
Publisher Singapore: World Scientific. 2003, pp. xii + 211.

Contents:
PART 1: Introduction to Random Time
1. Prologue
2. Stopping time
3. Martingale stopped
4. Random past and future
5. Other times
6. From first to last
7. Gapless time
8. Markov chain in continuum time
9. The trouble with the infinite
PART 2: Introduction to Quantum Randomness
1. Classical prologue
2. Standard quantum mechanics
3. Probabilities in standard quantum mechanics
4. Feynman's approach to quantum probabilities
5. Schrödinger's Euclidean quantum mechanics
6. Beyond Feynman's approach
7. Time for a dialogue

Readership: Advanced students in stochastics or quantum physics; mathematical researchers in those areas

This is a quite unusual and highly stimulating book, written by two masters of their fields. While directed primarily towards talented, broadminded and ambitious younger students of stochastics or mathematical physics, many more senior researchers will find points of considerable interest in the expositions. The level of exposition is relatively elementary in both parts, but the style is far from that of a textbook, being rather colloquial, with many historical insights and some anecdotal material, the latter particularly in Part 1. That part, written by Professor Kai Lai Chung, is in the classical probabilistic (Kolmogorovian) framework and centres around stopping times for discrete time Markov processes. Part 2, by Professor Jean-Claude Zambrini, addresses a basic difficulty of the theory of quantum physics, as recognized from earliest days on by the founders of the theory, namely the fact that time (and random time) is not a quantum observable. Zambrini offers a detailed and lucid discussion of this problem, based on a succinct introduction to the mathematics of quantum physics, and outlines a possible research programme that aims at constructing a sort of middle way between classical probability, in the sense of KoImogorov, and quantum probability, in the sense of von Neumann and Schrödinger, and drawing heavily on Feynman's path integral approach. The hope is that such a middle way, while presumably not fully satisfactory for either of the two probabilistic universes, will throw new light on the foundations of quantum mechanics, and in particular on the puzzle of time, as well as opening up completely new avenues for research in classical probability.
I would recommend any reader of the book to start with the two forewords and the final section of Part 2.

Reviewer:
Institute University of Aarhus
Place Aarhus, Denmark
Name O.E. Barndorff-Nielsen

Back to index


Title DISTRIBUTION THEORY OF RUNS AND PATTERNS AND ITS APPLICATIONS. A FINITE MARKOV CHAIN IMBEDDING APPROACH.
Author J.C. Fu and W.Y.W. Lou.
Publisher River Edge, New Jersey: World Scientific. 2003, pp. x + 162.

Contents:
1. Introduction
2. Finite Markov chain imbedding
3. Runs and patterns in a sequence of two-state trials
4. Runs and patterns in multi-state trials
5. Waiting-time distributions
6. Random permutations
7. Applications

Readership: Researchers in applied statistics, quality control and engineering systems

Traditionally, the distributions of runs and patterns were studied using combinatorial analysis but finding the appropriate combinatorial identities to derive the appropriate probability distributions can be difficult, if not impossible. The author's approach to the problem is to make use of finite Markov chain imbedding techniques for studying the distributions of runs and patterns. A great advantage of this approach is that it can be applied not only to independent and identically distributed cases but also to Markov-dependent multi-state trials.
Applications date back to the early 1980s for the purpose of evaluating the reliabilities of various engineering systems, to hypothesis testing and DNA sequence matching in the late 1990s, and to health care in 2000.
The text is neither a review book for the theory of runs and patterns, nor is it intended as a course textbook. It is mainly aimed at researchers and the contents are largely based upon recent developments in the area.

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

Back to index


Title ELEMENTS OF THE RANDOM WALK: AN INTRODUCTION FOR ADVANCED STUDENTS AND RESEARCHERS.
Author J. Rudnick and G. Gaspari.
Publisher Cambridge University Press, 2004, pp. xv + 329, £40.00/US$65.00. Contents:

1. Introduction to techniques
2. Generating functions I
3. Generating functions II: Recurrence, sites visited, and the role of dimensionality
4. Boundary conditions, steady state, and the electrostatic analogy
5. Variations on the random walk
6. The shape of a random walk
7. Path integrals and self-avoidance
8. Properties of the random walk: Introduction to scaling
9. Scaling of walks and critical phenomena
10. Walks and the O(n) model: Mean field theory and spin waves
11. Scaling, fractals, and renormalization
12. More on the renormalization group

Readership: Graduate students and researchers of statistics, mathematics and engineering

The random walk is a useful model across a range of scientific disciplines and has played a role in the analysis of stock prices and quantum field theory, to name but two novel applications. This book is self-contained in the basic topics of the random walk and provides an excellent introduction to emerging topics such as fractals, scaling and path integrals.
Central to the mathematical treatment is the generating function but mathematical background is provided in supplements at the end of each chapter, whenever appropriate. Exercises are included in a number of the chapters. The text is well-written and would provide an excellent course companion for both an intermediate course and an advanced course in probability.

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

Back to index


Title WHY STOCK MARKETS CRASH. Critical Events in Complex Financial Systems.
Author D. Sornette.
Publisher Princeton University Press. 2003, pp. xx + 421, US$19.95.

Contents:
1. Financial crashes: What, how, why, and when?
2. Fundamentals of financial markets
3. Financial crashes are "outliers"
4. Positive feedbacks
5. Modeling financial bubbles and market crashes
6. Hierarchies, complex fractal dimensions, and log-periodicity
7. Autopsy of major crashes: Universal exponents and log-periodicity
8. Bubbles, crises, and crashes in emergent markets
9. Prediction of bubbles, crashes, and antibubbles
10. 2050: The end of the growth era?

Readership: Anyone interested in the functioning of financial markets

No doubt, the author has thought carefully about how to present a wealth of potential material to an audience not particularly expert in the behaviour of complex systems. Starting from such systems, the author explains how certain periodicities may come naturally in financial markets. The book is more on data-driven observations than on mathematical theory. For most of the models presented, theory however does exist. I find the book lively, well written with lots of common sense insight on the functioning of our financial system. Is this the new "unifying" theory for understanding financial markets? Surely not; I did find however interesting information scattered throughout the text which students and researchers working on other approaches to the same questions might find interesting for their own research.

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

Back to index


Title THE CONCEPTS AND PRACTICE OF MATHEMATICAL FINANCE.
Author M.S. Joshi.
Publisher Cambridge University Press. 2003, pp. xvii + 473, £30.00.

Contents:
1. Risk
2. Pricing methodologies and arbitrage
3. Trees and option pricing
4. Practicalities
5. The Itô calculus
6. Risk neutrality and martingale measures
7. The practical pricing of a European option
8. Continuous barrier options
9. Multi-look exotic options
11. Multiple sources of risk
12. Options with early exercise features
13. Interest rate derivatives
14. The pricing of exotic interest rate derivatives
15. Incomplete markets and jump-diffusion processes
16. Stochastic volatility
17. Variance gamma models
18. Smile dynamics and the pricing of exotic options

APPENDIX A: Financial and mathematical jargon
APPENDIX B: Computer projects
APPENDIX C: Elements of probability theory
APPENDIX D: Hints and answers to exercises

Readership: Students or practitioners new to mathematical finance

In view of the fact that there are now many books on mathematical finance, several of which have been reviewed in these pages, the obvious question is what is different about this one? The author's aims are to impart a conceptual understanding of the basic ideas in mathematical finance, and to show how these are translated into practicalities. I think he succeeds in both of these objectives. The book starts at a satisfactorily introductory level, introducing basic motivations of risk and the concept of arbitrage. It examines things from several different perspectives, thus presenting an attractive overview of the subject, and certainly equipping readers with a broader understanding than would some of the technically deeper but narrower and more partisan books. The book has been very nicely produced by Cambridge University Press. I would certainly recommend that anyone teaching an introductory or intermediate course on this topic seriously consider this book as a potential course text.

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

Back to index


Title DYNAMICS OF MARKETS. ECONOPHYSICS AND FINANCE.
Author J.L. McCauley.
Publisher Cambridge University Press. 2004, pp. xvi + 209, £45.00/US$65.00.

Contents:
1. The moving target
2. Neo-classical economic theory
3. Probability and stochastic processes
4. Scaling the ivory tower of finance
5. Standard betting procedures in portfolio selection theory
6. Dynamics of financial markets, volatility and option pricing
7. Thermodynamic analogies vs instability of markets
8. Scaling, correlations and cascades in finance and turbulence
9. What is complexity?

Readership: Science graduate students, finance specialists

This is a refreshing text about the main ideas of econophysics, basically from a mathematical point of view. The 'dismal science' normally takes as its basis the neo-classical equilibrium model, with the underlying notion of Adam Smith's 'Stabilising invisible hand'. In contrast to the usual resulting comparisons of real data with how the markets 'should' behave, the approach here is to develop a new empirically-based model of markets which clarifies their basic volatility and instability.
The author's main background is in physics and particularly in nonlinear dynamics. Full advantage of this text would come only for those readers with at least good basic mathematics. However, the main ideas and arguments should be apparent to all.

Reviewer:
Institute Imperial College of Science, Technology and Medicine
Place London, U.K.
Name F.H. Berkshire

Back to index


Title HANDBOOK OF COMPUTATlONAL AND NUMERICAL METHODS IN FINANCE.
Author S.T. Rachev (Ed.).
Publisher Boston: Birkhäuser. 2004, pp. vi + 435, US$79.95.

Contents:
1. Skewness and kurtosis trades
2. Valuation of a credit spread put option: The stable Paretian model with copulas
3. GARCH-type processes in modeling energy prices
4. Malliavin calculus in finance
5. Bootstrap unit root tests for heavy-tailed time series
6. Optimal portfolio selection and risk management: A comparison between the stable Paretian approach and the Gaussian one
7. Optimal quantization methods and applications to numerical problems in finance
8. Numerical methods for stable modeling in financial risk management
9. Modern heuristics for finance problems: A survey of selected methods and appIications
10. On relation between expected regret and conditional value-at-risk
11. Estimation, adjustment and applications of transition matrices in credit risk models
12. Numerical analysis of stochastic differential systems and its applications in finance

Readership: Postgraduate students and researchers in quantitative finance

The title of this book may be somewhat misleading-as an edited volume, it contains several papers on some issues in quantitative finance. Most contributions have a computational/numerical slant. It is no surprise that several papers concentrate on heavy-tailed models, in particular Pareto-type models figure prominently. For me, the highlight is the paper by Kohatsu-Higa and Montero on "Malliavan Calculus in Finance". This seventy plus page paper gives a very readable introduction to this important field of current research. A further enjoyable paper is "Modern Heuristics for Finance Problems: A Survey of Selected Methods and Applications" by Schlottmann and Seese.

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

Back to index


Title C++ DESIGN PATTERNS AND DERIVATIVE PRICING.
Author M.S. Joshi.
Publisher Cambridge University Press. 2004, pp. xiii + 199, £35.00/US$55.00.

Contents:
1. A simple Monte Carlo model
2. Encapsulation
3. Inheritance and virtual functions
4. Bridging with a virtual constructor
5. Strategies, decoration and statistics
6. A random number class
7. An exotics engine and the template pattern
8. Trees
9. Solvers, templates and implied volatilities
10. The factory
11. Design patterns revisited

APPENDIX A: Black-Scholes formulas
APPENDIX B: Distribution functions
APPENDIX C: A simple array class
APPENDIX D: The code
APPENDIX E: Glossary

Readership: Readers who know the basics of mathematical finance and of C++ and who want to put the two together

This is a short book, but an elegant one. Assuming a basic knowledge of mathematical finance and of C++, it iIlustrates the ideas of object oriented programming, and how they can be used to implement financial models. It proceeds largely by example, with the examples being formulated to illustrate design principles, rather than aspects such as numerical efficiency. It includes a CD containing source code. It would serve as an excellent course text for a course on the practical aspects of mathematical finance.

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

Back to index


Title CONVEX OPTIMIZATION.
Author S. Boyd and L. Vandenberghe.
Publisher Cambridge University Press. 2004, pp. xiii + 716, £45.00/US$65.00.

Contents:
1. Introduction
PART I: Theory
2. Convex sets
3. Convex functions
4. Convex optimization problems
5. Duality
PART
II: Applications
6. Approximation and fitting
7. Statistical estimation
8. Geometric problems
PART III: Algorithms
9. Unconstrained minimization
10. Equality constrained minimization
11. Interior point methods

APPENDIX A: Mathematical background
APPENDIX B: Problems involving two quadratic functions
APPENDIX C: Numerical linear algebra background

Readership: Mathematicians, mathematical programmers

The development of interior point methods for convex optimization means that realistically sized problems can now be efficiently and reliably solved. Linear programming and least squares are special cases of convex programs. The reader of this text has to be sufficiently mathematically sophisticated so as to be comfortable with advanced calculus and linear algebra. There are almost no numerical examples even in the applications chapters; the applications are described abstractly. In the section on algorithms, simplified variants of a few robust and fast algorithms are presented. If the reader is at ease with a mathematical description of the problem areas, the solution algorithms in the text will read smoothly. This is a refererence text; I highly recommend it either if you teach nonlinear optimization at the graduate Ievel for a supplementary reading list and for your library, or if you solve optimization problems and wish to know more about solution methods and applications.

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

Back to index


Title INTRODUCTION TO LOGISTIC SYSTEMS PLANNING AND CONTROL.
Author G. Ghiani, G. Laporte and R. Musmanno. With a Foreword by M. Goetschalck.
Publisher Hoboken, New Jersey: Wiley. 2004, pp. xxiv + 352.

Contents:
1. lntroducing logistics systems
2. Forecasting logistics requirements
3. Designing the Iogistics network
4. Solving inventory management problems
5. Designing and operating a warehouse
6. Planning and managing long-haul freight transportation
7. Planning and managing short-haul freight transportation
8. Linking theory to practice

Readership: Operations researchers

This text is an introduction to logistics through mathematical modelling with all the consequential simplifications and idealizations. lt assumes that the reader is familiar with statistics and probability theory, and comfortable with mathematical programming models and algorithms. The models presented are deterministic; they assume that the data concerned with costs, demand, and distances are all known with certainty. The book is intended for an advanced graduate course, in logistics or operations management, to demonstrate the use of modelling in a logistics environment. It is unlikely to be of value to practitioners because of its mathematical demands and because of the inadequacy of the models to solve real life complex problems.

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

Back to index


Title THE CROSS-ENTROPY METHOD. A Unified Approach to Combinatorial Optimization, Monte-Carlo Simulation, and Machine Learning.
Author R.Y. Rubinstein and D.P. Kroese.
Publisher New York: Springer-Verlag. 2004, pp. xx + 300, US$84.95.

Contents:
1. Preliminaries
2. A tutorial introduction to the cross-entropy method
3. Efficient simulation via cross-entropy
4. Combinatorial optimization via cross-entropy
5. Continuous optimization and modifications
6. Noisy optimization with CE
7. Application of CE to COPs
8. Application of CE to machine learning

Readership: Engineers, computer scientists, mathematicians and statisticians interested in numerical methods

This book describes the cross-entropy method for a range of optimization problems. The cross-entropy (distance) itself is a very familiar object in statistics and information theory and is usually known as the Kullback-Leibler divergence (distance). The authors start with a rather short introduction to the probabilistic background, then describe a few simple examples to illustrate the main ideas; the rest of the book is a careful elaboration and generalization of these ideas.
It is a substantial contribution to stochastic optimization and more generally to the stochastic numerical methods theory.

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

Back to index


Title THE NATURE OF SCIENTIFIC EVIDENCE: STATISTICAL, PHILOSOPHICAL, AND EMPIRICAL CONSIDERATIONS.
Author M.L. Taper and S.R. Lele (Eds.).
Publisher University of Chicago Press. 2004, pp. xviii + 567, US$30.00.

Contents:
Foreword by C.R. Rao
PART I: Scientific Process. Overview, by N. Lewin-koh, M.L. Taper and S.R. Lele
1. A brief tour of statistical concepts, by N. Lewin-koh, M.L. Taper and S.R. Lele
2. Models of scientific inquiry and statistical practice: Implications for the structure of scientific knowledge, by B.A. Baurer
3. Experiments, observations, and other kinds of evidence, by S.M. Scheiner
PART II: Logics of Evidence. Overview, by V.P. Godambe
4. An error-statistical philosophy of evidence, by D.G. Mayo
5. The likelihood paradigm for statistical evidence by R. Royall
6. Why likelihood?, by M. Forster and E. Sober
7. Evidence functions and the optimality of the law of likelihood, by S.R. Lele
PART III: Realities of Nature. Overview, by M.S. Boyce
8. Whole-ecosystem experiments: Replication and arguing from error, by J.A. Miller and T.M. Frost
9. Dynamical models as paths to evidence, by M.L. Taper and S.R. Lele
10. Constraints on negative relationships: Mathematical causes and ecological consequences, by J.H. Brown, E.J. Bedrick, S.K.M. Ernest, J.-L.E. Cartron and J.F. Kelly
PART IV: Science, Opinion, and Evidence. Overview, by M.L. Taper and S.R. Lele
11. Statistics and the scientific method in ecology by B. Dennis
12. Taking the prior seriously: Bayesian analysis without subjective probability, by D. Goodman
13. Elicit data, not prior: on using expert opinion in ecological studies, by S.R. Lele
PART V: Models, Reality, and Evidence. Overview, by M.L. Taper and S.R. Lele
14. Statistical distances as loss functions in assessing model adequacy, by B.G. Lindsay
15. Model identification from many candidates, by M.L. Taper
PART VI: Conclusion
16: The nature of scientific evidence: A forward-looking synthesis, by M.L. Taper and S.R. Lele

Readership: 'Any reader engaged in the quantification and evaluation of data.' Students as well as established scientists

This book is a collection of papers, with the titles given in the contents, each followed by (typically) two commentaries and a rejoinder. The various authors are ecologists, statisticians and philosophers, but the editors hope that the volume will be of substantial interest to a much wider audience in the scientific community.
The book was preceded by two workshops and two symposia, and most of the contributors took part in these, so they will have had an opportunity to hear and digest others' views before writing their contribution. This gives the book a useful coherence: it is less susceptible to the criticism of many collections, that they are disconnected. The book also includes introductory and concluding synthesis chapters, which further help to integrate the contents.
Although the book is centred around ecology, perhaps this is a good thing. Discussions of the philosophy of science which are not grounded in a practical application can often drift off into the realms of prescriptive fantasy, and certainly ecology is confronted by most if not all of the deep problems of inference.
This is a book which would amply repay detailed study. I have already learnt much from it, and I look forward to learning much more.

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

Back to index


Title R.L. MOORE, MATHEMATICIAN AND TEACHER.
Author J. Parker.
Publisher Washington, D.C.: The Mathematical Association of America. 2005, pp. xiv + 587, US$ 45.95.

Contents:
1. Roots and influences (1882-1897)
2. Of richest promise (1897-1902)
3. On to Chicago (1903)
4. A veritabie hothouse (1903-1905)
5. Uneasy progress (1905-1908)
6. A settling experience (1908-1916)
7. Back to Texas (1916-1920)
8. A rewarding decade (1920-1930)
9. A change of direction (1930-1932)
10. Politics and persuasion (1933-1938)
11. Moore the teacher: A new era (1939-1944)
12. Blacklisted! (1943)
13. Class of '45 (1945)
14. Clash of the titans (1944-1950)
15. His female students
16. Moore's calculus (1945-1969)
17. Changing times (1953-1960)
18. Axiomatics continued: (1953-1965)
19. The final years (1965-1969)

APPENDIX 1: The Moore Genealogy Project
APPENDIX 2: The PhD Students of R.L. Moore
APPENDIX 3: Publications of Robert Lee Moore
APPENDIX 4: Descriptions of Courses

Readership: General reading; educators, scientists

R.L. Moore (1882-1974) was certainly a pioneering figure in the mathematical landscape of 20th century America. In many ways, he was also controversial. Though his research area was in point-set topology, he is more famous for his Moore method of teaching where students are taught the least and are invited to figure things out for themselves without the aid of any textbooks or teachers. This method, of course, is not without its price and can only be said to work some of the time. John Parker's biography of this fascinating personality makes for absorbing reading. In his lifetime, Moore produced fifty doctoral students including Mary Ellen Rudin, Burton Jones and Raymond Wilder. Moore's influence can be seen in the works of these mathematical descendants. This well-written biography is also packed with action and drama of the seldom seen academic world. For instance, Moore's "feud" with the number theorist H.S. Vandiver, well-known for his work on Fermat's Last Theorem, is described in detail in this biography in the chapter entitled 'Clash of the Titans.' The book is indeed colourful reading, not only for mathematicians and educators, but the general scientific public as well.

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

Back to index


Title ?: A BIOGRAPHY OF THE WORLD'S MOST MYSTERIOUS NUMBER.
Author A. Posamentier and I. Lehmann. With an Afterword by H.A. Hauptman.
Publisher Amherst, New York: Prometheus. 2004, pp. 324, US$26.00.

1. What is ??
2. The history of ?
3. Calculating the value of ?
4. ? Enthusiasts
5. ? Curiosities
6. Applications of ?
7. Paradox in ?

Epilogue

APPENDIX A: A Three-DimensionaI Example of a Rectilinear Equivalent to a Circular Measurement
APPENDIX B: Ramanujan's Work
APPENDIX C: Proof that e? > ?e
APPENDIX D: A Rope around the Regular Polygons

Readership: Scientists, non-specialists, high-school teachers

The ubiquity of ? is well known in science. This fundamental constant pops up everywhere. Yuri Manin once said that ? and the prime numbers are somehow interwoven into the fabric of the universe through the formula

where the product is over prime numbers p. This book is an entertaining voyage through history on the number ?. Describing the work from Archimedes to Ramanujan, the book surveys how great minds applied sophisticated mathematical tools to find better and better approximations to the number. The book also includes an afterword by Herbert Hauptman whose work in crystallography earned him the 1985 Nobel Prize in Chemistry.

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

Back to index


Title A TOUR THROUGH MATHEMATICAL LOGIC.
Author R.S. Wolf.
Publisher Washington, D.C.: The Mathematical Association of America. 2005, pp. xv + 397, US$52.95.

Contents:
1. Predicate logic
2. Axiomatic set theory
3. Recursion theory and computability
4. Gödel's incompleteness theorem
5. Model theory
6. Contemporary set theory
7. Nonstandard analysis
8. Constructive Mathematics

APPENDIX A: Deductive System for First-order Logic
APPENDIX B: Relations and Orderings
APPENDIX C: Cardinal Arithmetic
APPENDIX D: Groups, Rings and Fields

Readership: Mathematicians

This book is a leisurely excursion through some of the significant chapters in the history of mathematical logic. The writing style is friendly and the book is sprinkled with humorous anecdotes and historical vignettes. Thus, it is ideaI for anyone desiring a sympathetic introduction to the subject.
Apart from its amicable style, the book is not immune from criticisms. Here is one glaring omission. In Chapter 3 where the author discusses complexity theory, he gives a passing mention that primality testing can now be done in polynomial time, a long-standing open question. This 2002 result of Agrawal, Kayal and Saxena has a notable human interest side. Agrawal is an engineer by training and Kayal and Saxena are his undergraduate students! Moreover, their solution is elegantly short and simple that undergraduates can read their work and understand it, something that should be noted in a monograph aimed at undergraduates. What is even more striking is that the paper is not even listed in the references.
In spite of such omissions, readers will enjoy the book's informal style and its scenic tour through mathematical logic.

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

Back to index


Title DATA ANALYSIS AND GRAPHICS USING R: AN EXAMPLE-BASED APPROACH.
Author J. Maindonald and J. Braun.
Publisher Cambridge University Press. 2003, pp. xxiii + 362, £45.00/US$70.00.[Reprinted 2004].

Contents:
1. A brief introduction to R
2. Styles of data analysis
3. Statistical models
4. An introduction to formal inference
5. Regression with a single parameter
6. Multiple linear regression
7. Exploiting the linear model framework
8. Logistic regression and other generalized linear models
9. Multi-level models: Time series and repeated measures
10. Tree-based classification and regression
11. Multivariate data exploration and discrimination
12. The R system-Additional topics

Epilogue Models

APPENDIX: S-PLUS Differences

Readership: Researchers requiring practical skills in data analysis or students looking for examples of applications to complement a more theoretically based course

Today's modern statistical software, such as R, is freely available and provides sophisticated tools for researchers to use to manipulate and display their data. This book is an example-based introduction to data analysis using R. The mathematical content here has been kept to a minimum whilst the statistical and scientific issues are explored in more depth. Only basic statistical knowledge, such as that covered in a first undergraduate course, is required to follow the book. The user will have to have R installed on their machine to be able to do the exercises, which are provided at the end of the chapters.
The text includes a wealth of practical examples, drawn from a variety of practical applications which should be easily understood by the reader. The methods demonstrated are suitable for use in areas such as biology, social science, medicine and engineering. The core of the book is taken up with detailed discussion of regression methods which leads onto more advanced statistical concepts. Each chapter starts with a brief explanation and ends with a recap, references and suggestions for further reading and exercises for the reader to complete. In the appendix the authors draw attention to those differences between S-PLUS and R that may be important in the adaptation of code that appears in the book.

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

Back to index


Title DATA ANALYSIS OF ASYMMETRIC STRUCTURES. Advanced Approaches to Computational Statistics.
Author T. Saito and H. Yadohisa.
Publisher New York: Dekker. 2005, pp. vii + 258.

Contents:
1. Introduction
2. Paired comparisons with asymmetry
3. Graphical representation of asymmetric data
4. Multidimensional scaling of asymmetric data
5. Cluster analysis of asymmetric data
6. Network analysis of asymmetric data
7. Multivariate analysis of asymmetry between data sets

Readership: Students, graduate students and research workers interested in methodology and/or applications

The authors claim, correctly I believe, that this is the first reference book concerned with the analysis of asymmetry. Square asymmetric tables arise in many applications: input/output, import/export, father's occupations/son's occupations, immigration/emigration, pecked/pecking, confusion matrices, citing/cited journals, dom-inance data … . The main characteristic of these structures is that their rows and columns are classified by different modes of the same things. This book brings together the many models and visualizations that have been developed to analyze and understand such types of data. The models may be divided into those that use directly model asymmetric relationships and those that use separate parameters to analyze symmetry and departures from symmetry. The methodology is scattered throughout the journal literature, much of it not easily accessible. The authors, who themselves have made major contributions, have done a useful service in unifying this material.

Reviewer:
Institute The Open University
Place Milton Keynes, U.K.
Name J.C. Gower

Back to index


Title MATHEMATICAL LABORATORIES FOR MATHEMATICAL STATISTICS: EMPHASIZING SIMULATION AND COMPUTER INTENSIVE METHODS.
Author J.A. Baglivo.
Publisher Philadelphia: Society for Industrial and Applied Mathematics/Alexandria, Virginia: American Statistical Association. 2005, pp. xx + 260 + CD,US$70.00.

Contents:
1. Introductory probability concepts
2. Discrete probability distributions
3. Continuous probability distributions
4. Mathematical expectation
5. Limit theorems
6. Translation to statistics
7. Estimation theory
8. Hypothesis testing theory
9. Order statistics and quantiles
10. Two sample analysis
11. Permutation analysis
12. Bootstrap analysis
13. Multiple sample analysis
14. Linear least squares analysis
15. Contingency table analysis

Readership: Mathematical statistics students

This book takes students and instructors through a fairly traditional undergraduate mathematical statistics course. However, each chapter is supplemented with a computer based laboratory and problems written in the mathematics package Mathematica that are found on the accompanying CD. Note that one can only run the accompanying laboratories if one has already installed Mathematica; therefore the book is not completely self-contained. The Mathematica material is written to enable students to explore analytically, graphically and via simulation techniques many statistical concepts and these can be read in a relatively self-contained way. I did find the choice of Mathematica as the computational engine rather surprising given the large number of good statistical packages available. Its choice reflects the mathematical, rather than statistical, roots of this work. Thus Mathematica allows symbolic manipulation of, for example, likelihood functions, while putting less emphasis on exploratory data analysis.

Reviewer:
Institute University of Waterloo
Place Waterloo, Canada
Name P. Marriott

Back to index


Title PROBABILITY AND COMPUTING: RANDOMIZED ALGORITHMS AND PROBABILISTIC ANALYSIS.
Author M. Mitzenmacher and E. Upfal.
Publisher Cambridge University Press. 2005, pp. xvi + 352, £30.00/US$55.00.

Contents:
1. Events and probability
2. Discrete random variables and expectation
3. Moments and deviations
4. Chernoff bounds
5. Balls, bins and random graphs
6. The probabilistic method
7. Markov chain and random walks
8. Continuous distributions and the Poisson process
9. Entropy, randomness, and information
10. The Monte Carlo method
11. Coupling of Markov chains
12. Martingales
13. Pairwise independence and universal hash functions
14. Balanced allocations

Readership: Advanced undergraduate or beginning postgraduate students of computer science and applied mathematics having an elementary background in discrete mathematics

The authors are both professors of computer science, and the book was written to support an undergraduate course in computer science. Consequently, in the first chapter we find applications to verifying polynomial identities, verifying matrix multiplication and a randomized min-cut algorithm. There is an emphasis on graph theory. In the second half, the authors delve into more advanced topics which include (according to the blurb) continuous probability, but does not give any mention of the normal distribution. There are weaknesses in the presentation of elementary probability ("[1.k)" is used for the discrete set {1, 2, …, k}, "E_1 - E_2" means "the occurrence of an event that is in E_1 but not in E_2", and we have dice which "are 1", "land on 1", "are tossed'' and "are summed"). Despite this there is much of interest that is new to me in the handling of the advanced topics.

Reviewer:
Institute Imperial College of Science,Technology and Medicine
Place London, U.K.
Name R. Coleman

Back to index


Title STATISTICAL METHODS FOR SPATIAL DATA ANALYSIS.
Author O. Schabenberger and C.A. Gotway.
Publisher Boca Raton, Florida: Chapman and Hall/CRC Press. 2005, pp. xvii + 488, US$89.95/£39.99.

Contents:
1. Introduction
2. Some theory on random fields
3. Mapped point patterns
4. Semivariograms and covariance function analysis and estimation
5. Spatial prediction and kriging
6. Spatial regression models
7. Simulation of random fields
8. Non-stationary covariance
9. Spatio-temporal processes

Readership: Statisticians and graduate students, geographers, scientists and practitioners who deal with spatial data of any kind and who are familiar with linear model theory and matrix algebra

The appearance of this text is particularly timely since there have been considerable developments recently in the statistical methods for the analysis of spatial data. The special methods required to deal with the unique challenges of analyzing spatial data are brought together using a unified approach with numerous practical examples from a variety of application areas. In the introductory chapter the authors begin by distinguishing between the different types of spatial data and introduce the concept of autocorrelation and its various measures. This is followed by a description of the theoretical framework of random fields in both the spatial and frequency domains, and a discussion of spectral density functions and linear filters. Point patterns, binomial and Poisson processes are illustrated using data on woodpeckers, birth weights and lightning strikes in Chapter 3. Next, a substantial chapter deals with estimation and modelling of the covariance function and the semivariogram. These are treated in some detail using maximum likelihood, restricted maximum likelihood and generalized estimating equations, and also includes non-parametric methods using step functions and kernel functions. This leads naturally into linear prediction using various kinds of kriging from simple, ordinary and universal kriging to disjunctive kriging and isofactorial models. Perhaps the most important chapter is Chapter 6 on spatial regression models; this begins with the standard linear model with uncorrelated errors using ordinary least squares with residual diagnostics and neighbourhood adjustments, then develops autoregressive models with correlated errors, and goes on to consider generalized linear models and Bayesian hierarchical models. Finally, three short chapters on simulating spatial data, different types of non-stationarity and spatio-temporal processes, respectively, are used to discuss recent advances in these areas.
This is a well-presented research-level text with interesting examples and an extensive reference list, much of which relates to work which has appeared during the last five years or so.

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

Back to index


Title STEREOLOGY FOR STATISTICIANS.
Author A. Baddeley and L.B. Vedel Jensen.
Publisher Boca Raton, Florida: Chapman and Hall/CRC Press. 2005, pp. xvi + 395.

Contents:
1. Introduction
2. Classical stereology
3. Overview of modern stereology
4. Geometrical identities
5. Geometrical probability
6. Statistical formulations of stereology
7. Uniform and isotropic uniform designs
8. Vertical and local designs
9. Ratio estimation
10. Discrete sampling and counting
11. Inference for particle populations
12. Design of stereological experiments
13. Variance of stereological estimators
14. Frontiers and open problems

APPENDIX: Sampling Theory

Readership: Statisticians who might be called on for advice in the biological, materials, and geological sciences and image processing

Stereology can be summarized as the sampling theory for quantitative microscopy, yet is largely unknown to the statistical community. Its objective is to give statistical descriptions of 3-d processes from measurements of features on a micrograph taken on a section through a specimen. The emphasis here is on design-based inference, when the images are from random sections, as opposed to model-based stereology which assumes a spatial homogeneity within the process and on the image.
The authors are mathematical statisticians who have been active participants in much of the theoretical development of the subject. The book is up-to-date and well illustrated. There are exercises and over 600 references. All that seems to be missing are some case studies following the transfer of micrograph measurement to statistical report. This book will assist statisticians in responding to queries from microscopists, and, difficult though the mathematics will seem to the latter, will help microscopists communicate with statisticians.

Reviewer:
Institute Imperial College of Science, Technology and Medicine
Place London, U.K.
Name R. Coleman

Back to index


Title AN R AND S-PLUS® COMPANION TO MULTIVARIATE ANALYSIS.
Author B.S. Everitt.London: Springer-Verlag.
Publisher 2005, pp. xiii + 221, US$69.95. Contents:

1. Multivariate data analysis
2. Looking at multivariate data
3. Principal component analysis
4. Exploratory factor analysis
5. Multidimensional scaling and correspondence analysis
6. Cluster analysis
7. Grouped multivariate data: Multivariate analysis of variance and discriminant function analysis
8. Multiple regression and canonical correlation
9. Analysis of repeated measures data

APPENDIX: An Aide Memoir for R and S-Plus®

Readership: Graduate students and advanced undergraduates on applied statistics courses

This book, as the title implies, is a companion to multivariate analysis, i.e. it is a companion to a multivariate textbook, which covers all the theory necessary. The emphasis is on the practice of multivariate analysis not the theory. Whilst there are many books on multivariate analysis on the market this one uses a wealth of practical examples to explain and demonstrate how multivariate sets of data can be analyzed using the appropriate software. The author uses R and S-Plus®, throughout to demonstrate the use of multivariate data analysis. The user will have to have R and/or S-Plus® installed on their machine to be able to do the exercises which are provided at the end of the chapters. It is assumed that the readers have some prior experience of using these packages but do not have to be experts. The appendix briefly describes the main features of the packages but is not a manual in itself. For further information the author recommends several references for the user to consult.
This book is not an introductory text. It presumes some knowledge of basic statistical theory and practice and in particular multivariate analysis theory. All the sets of data and code used in the book are available from the website http://biostatistics.iop.kcl.ac.uk/publications/everitt/. The text also contains a wealth of references for the reader to pursue.

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

Back to index


Title AN INTRODUCTION TO MULTIVARIATE DATA ANALYSIS.
Author T.F. Cox.
Publisher London: Hodder Arnold. 2005, pp. vii + 232, £19.99.

Contents:
1. Introduction
2. Matrix algebra
3. Basic multivariate statistics
4. Graphical representation of multivariate statistics
5. Principal components analysis
6. Biplots
7. Correspondence analysis
8. Cluster analysis
9. Multidimensional scaling
10. Linear regression analysis
11. Multivariate analysis of variance
12. Canonical correlation analysis
13. Discriminant analysis and canonical variate analysis
14. Loglinear modelling
15. Factor analysis
16. Other latent variable models
17. Graphical modelling
18. Data mining

Readership: Students using statistics for research

It is common for students on their first introduction to multivariate statistics to be overwhelmed by the large number of different techniques and indeed different names for essentially the same technique. For such students, and for those who need multivariate methods in their research work, this book would be a very useful purchase. It takes a concise tour of a very large number of methods giving the basic motivation and theory, the details of computation and a nicely judged numerical example for each. Connections between different methods are made and strengths and weaknesses are pointed out. The volume is remarkable in the amount of material that is covered for its size. The book does assume a familiarity with the basics of matrix algebra and perhaps, if used as a basis for an undergraduate course, an instructor would need to ensure that students had the required fluency by using supplementary material.

Reviewer:
Institute University of Waterloo
Place Waterloo, Canada
Name P. Marriott

Back to index


Title GENERALIZED LATENT VARIABLE MODELING: MULTILEVEL, LONGITUDINAL AND STRUCTURAL EQUATION MODELS.
Author A. Skrondal and S. Rabe-Hesketh.
Publisher Boca Raton, Florida: Chapman and Hall/CRC Press. 2004, pp. xiii + 508.

Contents:
PART 1: Methodology
1. The omni-presence of latent variables
2. Modeling different response processes
3. Classical latent variable models
4. General model framework
5. Identification and equivalence
6. Estimation
7. Assigning values to latent variables
8. Model specification and inference
PART 2: Applications
9. Dichotomous responses
10. Ordinal responses
11. Counts
12. Durations and survival
13. Comparative responses
14. Multiple processes and mixed responses

Readership: Statisticians interested in latent variable models

The "Generalized Linear Latent and Mixed Model" (GLLAMM) framework can be described as follows, using the terminology of multi-level models. A predictor is a form which is linear in observable and latent explanatory variables defined for units at various levels at or above the level of the predictor. The observed response has an exponential family form, linear in the predictor. The authors show that this framework includes as special cases random coefficient models, measurement models, and structural equation models. (They have developed an analysis package gllamm which is an add-on to Stata, and which carries out estimation using maximum likelihood via adaptive Gaussian quadrature.) The treatment in the book provides an elegant and illuminating unification of concepts and models from divers disciplines. The final application chapters deal with a broad collection of interesting applications, to areas such as meta-analyses, disease mapping, confirmatory factor analysis, and case-control studies. The book is well worth acquiring, and would be a suitable text for advanced graduate courses.

Reviewer:
Institute University of Waterloo
Place Waterloo, Canada
Name M.E. Thompson

Back to index


Title STATISTICAL METHODS IN BIOINFORMATICS: AN INTRODUCTION, 2nd edition.
Author W.J. Ewens and G.R. Grant.
Publisher New York: Springer-Verlag. 2005, pp. xx + 597, US$89.95.

Contents:
1. Probability theory (i): One random variable
2. Probability theory (ii): Many random variables
3. Statistics (i): An introduction to statistical inference
4. Stochastic processes (i): Poisson process and Markov chains
5. The analysis of one DNA sequence
6. The analysis of multiple DNA or protein sequences
7. Stochastic processes (ii): Random walks
8. Statistics (ii): Classical estimation theory
9. Statistics (iii): Classical hypothesis testing theory
10. BLAST
11. Stochastic processes (iii): Markov chains
12. Hidden Markov chains
13. Gene expression, microarrays, and multiple testing
14. Evolutionary models
15. Phylogenetic tree estimation

APPENDIX A: Basic Notions In Biology
APPENDIX B: Mathematical Formulae and Results
APPENDIX C: Computational Aspects of the Binomial and Generalized Geometric Distribution Function
APPENDIX D: BLAST: Sum of Normalized Scores

Readership: Biostatisticians and statisticians who want to learn about bioinformatics

This is the second edition of "a very substantial and highly professional contribution to bioinformatics and applied statistics"; see my review for the first edition [Short Book Reviews, Vol. 22, p. 8]. The authors make the text more consistent shuffling and improving a few chapters. They paid a tribute to the novel and rapidly developing statistical techniques in genomics. Once more I have found a confirmation of the new edition law: the latest edition is never shorter than the previous one!

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

Back to index


Title STATISTICAL CONCEPTS AND APPLICATIONS IN CLINICAL MEDICINE.
Author J. Aitchison, J.W. Kay and I.J. Lauder.
Publisher Boca Raton, Florida: Chapman and Hall/CRC Press. 2005, pp. iv + 339, US$79.95/£39.99.

Contents:
1. The field of application
2. Relating the present patient to past experience
3. A review of statistical methodology
4. Further statistical methodology
5. Experience
6. Observation and measurement
7. Indirect measurement: Assay and calibration
8. Diagnosis
9. Special aspects of diagnosis
10. Prognosis and treatment
11. Assessment

APPENDIX A: Data and Software

Readership: Persons wishing to gain an appreciation for how applied statisticians confront clinical data and problems

This text is written from the perspective of an applied statistician, with one goal being to educate the reader on how to approach problems arising in a clinical setting. Rather than taking the traditional approach of developing a theory and then illustrating its application with an example, this text uses real examples to motivate the methods. As such, it represents a refreshing text for persons interested in gaining an appreciation of how applied statisticians usually function. Two chapters are devoted to developing statistical methods, but readers interested in learning these methods would do better to refer to more specialized texts for a more thorough development.

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

Back to index


Title THE EVALUATION OF SURROGATE ENDPOINTS.
Author T. Burzykowski, G. Molenberghs and M. Buyse (Eds.).
Publisher New York: Springer-Verlag. 2005, pp. xxiii + 408, US$74.95.

Contents:
1. Introduction
2. Setting the scene
3. Regulatory aspects of using surrogate markers in clinical trials
4. Notation and motivating studies
5. The history of surrogate endpoint validation
6. Validation using single-trial data: Mixed binary and continuous outcomes
7. A meta-analytic validation framework for continuous outcomes
8. The choice of units
9. Extensions of the meta-analytic approach to surrogate endpoints
10. Meta-analytic validation of binary outcomes
11. Validation in the case of two failure-time endpoints
12. An ordinal surrogate for a survival true endpoint
13. A combination of longitudinal and survival endpoints
14. Repeated measures and surrogate endpoint validation
15. Bayesian evaluation of surrogate endpoints
16. Surrogate marker validation in mental health
17. The evaluation of surrogate endpoints in practice: Experience in HIV
18. An alternative measure for meta-analytic surrogate endpoint validation
19. Discussion: Surrogate endpoint definition and evaluation
20. The promise and peril of surrogate endpoints in cancer research

Readership: Medical statisticians, clinical trial methodologists

This edited volume deals with a topic that has been the subject of much debate since publication of Prentice's 1989 attempt to formalize the definition of a surrogate marker or endpoint. This debate, in the words of the editors, "has since been laden with a certain amount of skepticism". This work focuses on evaluation of a surrogate endpoint. This is, typically, the determination of whether or not a shorter-term endpoint can replace a longer-term endpoint of primary interest. To quote the editors again, "different people approach surrogate marker evaluation with various degrees of comfort" and the editors' hope is that "the current text does proper justice to all views, not just the editors' views."
This is a somewhat unusual edited volume in that nine of the twenty chapters have one or more of the editors as an author. Also the editing is said to have been strong. In spite of this, it is true that most of the views surrounding surrogate endpoints are discussed. Non-statisticians may need to be selective in their reading as the technical level is reasonably high. However, an attempt has been made to maintain common notations whenever possible. The editors have succeeded in producing a very useful book.

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

Back to index


Title REGRESSION METHODS IN BIOSTATISTICS: LINEAR, LOGISTIC, SURVIVAL AND REPEATED MEASURES MODELS.
Author E. Vittinghoff, D.V. Glidden, S.C. Shiboski, C.E. McCulloch.
Publisher New York: Springer-Verlag. 2004, pp. xv + 340, US$79.95.

Contents:
1. Introduction
2. Exploratory and descriptive methods
3. Basic statistical methods
4. Linear regression
5. Predictor selection
6. Logistic regression
7. Survival analysis
8. Repeated measures analysis
9. Generalized linear models
10. Complex surveys
11. Summary

Readership: Biostatistics readers, post-graduate research physicians

The subtitle of this book is "Linear, logistic, survival, and repeated measures models". All four authors are in the Department of Epidemiology and Biostatistics in the University of California, San Francisco. This text is nicely written and well arranged and provides excellent, reasonably brief, information on the selected topics. The exercises (or "problems" as they are called here) are somewhat disappointing; many are small questions deriving from data earlier on (e.g. "Verify the calculation of the predicted values and residuals in …"). Others ask the reader to describe something "from your own area of interest." In the nine pages of references, I noted two books which now have later editions. The authors' data are analyzed using STATA; as they point out (p. 320), there are several "excellent ... packages ... that implement the ... techniques used here."

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

Back to index


Title ELEMENTARY STATISTICAL QUALITY CONTROL, 2nd edition.
Author J.T. Burr.
Publisher New York: Dekker. 2005, pp. xviii + 449.

Contents:
1. Why statistics?
2. Characteristics of data and how to describe them
3. Simple probability and probability distributions
4. Control charts in general
5. Control charts for attributes
6. Control charts for measurements: Process control
7. Process capability
8. Further topics in control charts and appIications
9. Acceptance sampling for attributes
10. Some standard sampling plans for attributes
11. Sampling by variables
12. Tolerances for mating parts and assemblies
13. Studying relationships between variables by linear correlation and regression
14. A few reliability concepts

Readership: Students and practitioners of basic statistics, especially quality control, in manufacturing environments

This is the second edition of an introductory text on quality controI originally written by the author's father, Irving W. Burr, first published in 1979. This edition updates some terminology and examples, and adds a discussion of process capability, but emphasizes that the fundamental, traditional, methods of quality control covered in the original edition are still central to quality improvement in industry no matter what management model may be overlaid on them (TQC, TQM, MBNQA, Six Sigma, etc.).
The second edition preserves most of the structure and friendly narrative style of the original. The book is intended to be accessible to a wide range of industrial users. Therefore coverage is elementary and no prior background in statistics is assumed, but the reader does need to be competent with basic algebra. The book contains no derivations or proofs of concepts, but illustrations are given to convince the reader of the plausibility of the claims that are made about the various methods.
As with most statistics texts, not all the chapters in this book need be read in sequence. After the first four chapters of introductory concepts, it becomes a reference book that one can dip into as needed for particular topics of interest. Process control and acceptance sampling are areas that resonate most with people who have already experienced such issues at work. Having said that, however, I found the chapters quite pleasant to read as first exposures to each topic.
This is a nice book to have, both for beginners and for practitioners who need to communicate with beginners.

Reviewer:
Institute ###
Place Brookfield, U.S.A.
Name C.A. Fung

Back to index


Title PERMUTATION, PARAMETRIC AND BOOTSTRAP TESTS OF HYPOTHESES, 3rd edition.
Author P. Good.
Publisher New York: Springer-Verlag. 2005, pp. xix + 315 + CD, US$84.95. [Original 1994, Short Book Reviews, Vol. 14, p. 46; 2nd edition 2000].

Contents:
1. A wide range of applications
2. Optimal procedures
3. Testing hypotheses
4. Distributions
5. Multiple tests
6. Experimental designs
7. Multifactor designs
8. Categorical data
9. Multivariate analysis
10. Clustering in time and space
11. Coping with disaster
12. Solving the unsolved and the impossible
13. Publishing your results
14. Increasing computational efficiency

APPENDIX: Theory of Testing Hypotheses

Readership: Practising statisticians and their teachers

The multifaceted and unorthodox table of contents (with its substructure covering more than eight pages), the author's frequent references to himself and folksy expressions and images set the tone of the book. This may not appeal to many readers.
The few pages read in detail indicate that the book needs heavy copy-editing and serious review by unflinching colleagues. For example, on page 27 alone, the integral signs in the Bayes risks and marginal density are missing. The proof of the Neyman-Pearson Lemma in Appendix A is completely botched, with reversed inequalities, skipped steps, and missing essential details. It is difficult to fathom the utility of such an Appendix for practicing statisticians.
Table 2.1a seeks to explain the concept of Type I and Type II errors but is poorly laid out, with column headings and cell contents blended. Table 2.1b is actually wrong, in classifying the rejection of the hypothesis as a Type II error. Perhaps this is simply a typographical error, but how will the student recognize this?
It is difficult to reconcile the fact that this book is a third edition (although the title seems to have changed) given its many flaws. Regardless of how much valid material may appear elsewhere in the book, its numerous flaws make reading it very painful. It is of limited value for the practicing applied statistician and cannot be recommended as a text for students.

Reviewer:
Institute Boeing
Place Seattle, U.S.A.
Name F.W. Scholz

Back to index


Title EXTREME VALUE AND RELATED MODELS WITH APPLICATIONS IN ENGINEERING AND SCIENCE.
Author E. Castillo, A.S. Hadi, N. Balakrishnan and J.M. Sarabia.
Publisher Chichester, U.K.: Wiley. 2005, pp. xiv + 362, £48.95.

Contents:
PART I: Data, Introduction and Motivation
1. Introduction and motivation
PART II: Probabilistic Models Useful in Extremes
2. Discrete probabilistic models
3. Continuous probabilistic models
4. Multivariate probabilistic models
PART III: Model Estimation, Selection and Validation
5. Model estimation
6. Model selection and validation
PART IV: Exact Models for Order Statistics and Extremes
7. Order statistics
8. Point processes and exact models
PART V: Asymptotic Models for Extremes
9. Limit distributions of order statistics
10. Limit distributions of exceedances and shortfalls
11. Multivariate extremes

Readership: Theoretical and applied statisticians, hydrologists, engineers and scientists who meet extreme values in their research

This text provides an extensive coverage of the distribution theory of extreme values and illustrates the methods using several small sets of data related to applications in structural engineering, hydrology, transportation, climatology and environmental science. The sets of data are described in the introductory chapter which forms Part I of the book. The three chapters in Part II cover univariate discrete, univariate continuous and multivariate distributions, respectively, in a clear and concise way, using moment generating functions and characteristic functions to derive many of the results. Point and interval estimation using maximum likelihood, the method of moments, probability weighted moments, quantile least squares and the truncation method are covered in the first chapter of Part III, which is followed by a chapter on probability plotting, describing in particular the use of P-P and Q-Q plots.
Part IV covers models for order-statistics and extreme values, joint distributions of the maximum and minimum, return periods and probabilities of exceedances. In Chapter 8 various exact models are introduced, these include point processes using the Poisson distribution, the Poisson flaws model, mixture models, the competing risk models and the Poisson storm model. The limit distributions of order statistics (Weibull, Gumbel and the general extreme value distributions (GEVDs)) are discussed in the next chapter, together with the development of confidence intervals for their parameters. This section concludes with some more complicated ideas involving dependent observations and moving average models. The generalized Pareto distribution is examined in Chapter 10 on limit distributions of exceedances and shortfalls. The final chapter extends some of the earlier results to multivariate extremes. Here some parametric bivariate models (logistic, direchlet and biIogistic) are examined as is the peaks-over-threshoId multivariate model.
In each chapter, the methods described are illustrated using one or more of the given sets of data, extensively supported by probability plots. Each chapter ends with a set of exercises. The text is well presented with clearly displayed equations and diagrams. This text provides a particularly useful reference book for the multitude of distributions encountered in the statistics of extreme values.

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

Back to index


Title STATISTICS OF EXTREMES: THEORY AND APPLICATIONS.
Author J. Beirlant, Y. Goegebeur, J. Segers and J. Teugels. With contributions from D. De Waal and C. Ferro.
Publisher Chichester, U.K.: Wiley. 2005, pp. xiii + 490, £60.00.

Contents:
1. Why extreme value theory?
2. The probabilistic side of extreme value theory
3. Away from the maximum
4. Tail estimation under Pareto-type models
5. Tail estimation for all domains of attraction
6. Case studies
7. Regression analysis
8. Multivariate extreme value theory
9. Statistics of multivariate extremes
10. Extremes of stationary time series
11. Bayesian methodology in extreme value statistics

Readership: Graduate students and researchers in applied probability and statistics; statisticians and scientists involved in application areas including hydrology, finance and insurance, environmental research and meteorology, metallurgy, geology, engineering

E.J. Gumbel's seminal text Statistics of Extremes was published by Columbia University Press in 1958. After that, relatively little further progress in the area was made until a period of rapidly accelerating development began around 1970. The present authors believe that extreme value theory has matured in the past two decades (more than half the references in the book were published during the last ten years); hence they deliberately chose the same title for their book as Gumbel. Like him, they provide an exposition of the theory and an overview of the current state of knowledge.
As the subtitle indicates, applications are an important aspect of the book. In the first chapter a wide variety of examples are used to illustrate basic ideas. Many of these exemples appear again later on, and some form the case studies of Chapter 6. Essentially, Chapters 1-5 are all expositions of well-established methodology, whilst Chapters 7-11 are concerned with areas that are still being very actively developed.
The authors say that they conceived the book as a graduate or advanced undergraduate course text. Knowledge of basic probability theory and statistics is assumed (preferably with some knowledge of Poisson processes), together with a reasonable level of mathematics. Only well-known results from analysis, probability, and statistics are used. The book is well written and the authors make good use of graphical procedures to illustrate and illuminate their exposition.

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

Back to index


Title LAWS OF SMALL NUMBERS: EXTREMES AND RARE EVENTS, 2nd, revised and extended edition.
Author M. Falk, J. Hüsler and R.-D. Reiss.
Publisher Basel: Birkhäuser. 2004, pp. xiii + 376.[Original 1995, Short Book Reviews, Vol. 15, p. 56].

Contents:
PART I: The IID Case: Functional Laws of Small Numbers
1. Functional laws of small numbers
2. Extreme value theory
3. Estimation of conditional curves
PART II: The IID Case: Multivariate Extremes
4. Basic theory of multivariate maxima
5. Multivariate extremes: The Pickands approach
6. The Pickands approach in the bivariate case
7. Multivariate extremes: Supplementary concepts and results
PART III: Non IID Observations
8. Introduction to the Non IID Case
9. Extremes of random sequences
10. Extremes of Gaussian processes
11. Extensions for rare events
12. Statistics of extremes

Readership: Researchers in probability theory, statistics and number theory; statisticians; graduate and postgraduate students

This second, revised and extended edition of the book mainly aims at giving a mathematically oriented development of the theory of extremes and rare events underlying a large number of applications. Because of the increasing interest for this active research area and its numerous practical applications, various new results from the statistical literature are incorporated in this updated edition which includes now three parts.
The first two parts are devoted to "The IID case". Part I deals with functional laws of small numbers, provides basic elements from extreme value theory and discusses conditional problems, the whole in an univariate framework. Part II, which is added to this edition, includes recent developments in multivariate extremes based on the Pickands approach and deals with further multivariate concepts such as maxima, exceedances and upper order statistics. The last part "Non IID observations" of the book pertains to extremes and rare events of non IID random sequences, among which Gaussian processes, along with some applications. Recent approaches are also included.
As a remark, the section and appendix on the statistical software Xtremes and the software itself are no longer included in the present edition. Moreover, the chapter treating the statistics of extremes is a bit restricted and refer the interested reader to the new book by J. Beirlant, Y. Goegebear and J. Teugels (2004), Statistics of Extremes: Theory and Applications [Short Book Reviews, Vol. 25, p. 27] and also to the references therein to supplement the bibliography.

Reviewer:
Institute Catholic University Leuven
Place Leuven, Belgium
Name S.A. Ladoucette

Back to index


Title LARGE DEVIATIONS AND METASTABILITY.
Author E. Olivieri and M. Eulália Vares.
Publisher Cambridge University Press. 2004, pp. xv + 512, £80.00/US$140.00.

Contents:
1. Large deviations: Basic results
2. Small random perturbations of dynamical systems: Basic estimates of Freidlin and Wentzell
3. Large deviations and statistical mechanics
4. Metastability. General description. Curie-Weiss model. Contact process
5. Metastability. Models of Freidlin and Wentzell
6. Reversible Markov chains in the Freidlin-Wentzell regime
7. Metastable behaviour for lattice spin models at low temperature

Readership: Graduate students and researchers in mathematics, probability theory and theoretical physics

This monograph is volume 100 in the prestigious series 'Encylopedia of Mathematics and its Applications'. Chapters 1 to 3 give general introductions to large deviation theory (e.g. theorems of Cramèr, Chernoff and Sanov), metastability (e.g. Freidlin- Wentzell theory) and statistical mechanics (e.g. Gärtner-Ellis theorem).
Chapters 4 to 7 of the book give a thorough overview of metastable behaviour of stochastic systems, from the physical description in thermodynamics to the most rigorous mathematical treatment. This impressive monograph provides a quite complete survey of the topic of metastability and will certainly be a very useful reference for all research activity in this field.

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

Back to index


Title DISCRETE-TIME MARKOV JUMP LINEAR SYSTEMS.
Author O.L.V. Costa, M.D. Fragoso and R.P. Marques.
Publisher London: Springer-Verlag. 2005, pp. x + 280, US$89.95.

Contents:
1. Markov jump linear systems
2. Background material
3. On stability
4. Optimal control
5. Filtering
6. Quadratic optimal control with partial information
7. H?-control
8. Design techniques and examples

APPENDIX A: Coupled Algebraic Riccati Equations
APPENDIX B: Auxiliary Results for the Linear Filtering Problem with ?(k) Unknown
APPENDIX C: Auxiliary Results for the H2-Control Problem

Readership: Graduate students in stochastic control theory or more generally in stochastic processes

The book is concerned with the control of discrete-time stochastic linear systems whose dynamics can abruptly change. The changes are assumed to be governed by a Markov chain. There is a substantial and growing literature in this topic. Three core approaches are multiple models, hidden Markov models and operator theory. This book focuses on the latter of these approaches. It gives a comprehensive theory for the control of these systems within the operator theoretic framework including stability, H2 and H?-control with full and partial state information. The book is highly recommended to anybody with an interest in control problems associated with Markov Jump Linear Systems.

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

Back to index


Title APPLIED TIME SERIES ECONOMETRICS.
Author H. Lütkepohl and M. Krätzig (Eds.).
Publisher Cambridge University Press. 2004, pp. xxv + 323, £48.00/US$85.00 Cloth; £21.99/US$37.99 Paper.

Contents:
1. Initial tasks and overview
2. Univariate time series analysis
3. Vector autoregressive and vector error correction models
4. Structural vector autoregressive modeling and impulse responses
5. Conditional heteroskedasticity
6. Smooth transition regression modeling
7. Nonparametric time series modeling
8. The software JMulTi

Readership: Econometricians, time series analysts

This book provides a well-written and lucid discussion of many of the of the important and ubiquitous time series methods used in modern econometrics. Topics covered include cointegration, conditional volatility, ARCH models and their generalizations, and non-parametric approaches. The style is uniform and appealing and liberally sprinkled with applications to real sets of data. The final chapter discusses the JAVA program JMulTi (Java-based Multiple Time Series software) which is freely availeble via the internet. The book is nicely produced with a list of notations, a full set of references, and, importantly for an edited volume, an adequate index.

Reviewer:
Institute Imperial College of Science,Technology and Medicine
Place London, U.K.
Name A.T. Walden

Back to index


Title AN INTRODUCTION TO CONTINUOUS TIME STOCHASTIC PROCESSES. THEORY, MODELS AND APPLICATIONS TO FINANCE, BIOLOGY, AND MEDICINE.
Author V. Capasso and D. Bakstein.
Publisher Boston: Birkhäuser, pp. xi + 343, £87.74. Contents:

1. Fundamentals of probability
2. Stochastic processes
3. The Itô integral
4. Stochastic differential equations
5. Applications to finance and insurance
6. Applications to biology and medicine

APPENDIX A: Measure and Integration
APPENDIX B: Convergence of Probability Measures on Metric Spaces
APPENDIX C: Maximum Principles of Elliptic and Parabolic Operators
APPENDIX D: Stability of Ordinary Differential Equations

Readership: Anyone with a background in pre-measure theory probability and calculus and an interest in a rigorous treatment of stochastic processes

This book is an introduction to stochastic processes, stochastic integration and differential equations. It intends to be largely self-contained and accessible to scientists and professionals from a business or academic background, but its rigorous mathematical style will hold more appeal to those with a taste for the mathematics of stochastic processes. There are several books which cover the basic results (Chapters 1-4) in probability and stochastic integration. The last two chapters provide somewhat less common and interesting applications to Black-Scholes theory, interest rate models, insurance mathematics and to population dynamics, epidemics, self-organizing systems and the neurosciences.

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

Back to index


Title BRANCHING PROCESSES: VARIATION, GROWTH AND EXTINCTION OF POPULATIONS.
Author P. Haccou, P. Jagers and V.A. Vatutin.
Publisher Cambridge University Press. 2005, pp. xii + 316, £60.00/US$95.00.

Contents:
1. Generalities
2. Discrete-time branching processes
3. Branching in continuous time
4. Large populations
5. Extinction
6. Development of populations
7. Specific models
Readership: Academic (applied probability, statistics, biology: researchers and practitioners)

Peter Jagers' book, Branching Processes with Biological Applications, appeared thirty years ago. The present book has the same underlying theme, that observation (collection of biological data) and theory (applied probability modelling) should be close relatives. (The thought crosses my mind that close relatives sometimes argue vigorously, which is not inappropriate here.) The main authors comprise 'one biologist and two mathematicians' and many sections in the later chapters are written by other contributors, of whom there are fifteen.
The authors say that their book is aimed primarily at biologists and that they believe that it 'can be read, in full, by an interested biologist with a basic command of calculus, linear algebra, and probability theory'. In my experience, such a belief would be rather optimistic, certainly so for England. However, the book should serve well, among other things, as an excellent introduction to applied probabilists looking for interesting and worthwhile applications.
The authors opine that much of mathematical biology should be treated probabilistically rather than deterministically. They also advocate stochastic modelling at the individual level rather than deterministic modelling at the population level. As they say, the former can be scaled up and approximated transparently whereas the latter may contain hidden unrealistic features. Such thoughts, and much more wide-ranging wisdom, are presented in Chapter 1, a good read for any statistician.
Finally, I cannot resist quoting: 'Too many math-ematicians, in our view, work on intellectual riddles, while important scientific problems escape their attention.'

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

Back to index


Title OPTIMIZATION.
Author K. Lange.
Publisher New York: Springer-Verlag. 2004, pp. x + 252, US$79.95.

Contents:
1. Elementary optimization
2. The seven C's of analysis
3. Differentiation
4. Karush-Kuhn-Tucker theory
5. Convexity
6. The MM algorithm
7. The EM algorithm
8. Newton's method
9. Conjugate gradient and quasi-Newton
10. Analysis of convergence
11. Convex programming

APPENDIX: The Normal Distribution

Readership: Statisticians

This text is a member of the series Springer Texts in Statistics. It is concerned with solving highly non-linear, most commonly unconstrained, optimization problems. Such problems arise when determining the parameters of statistical models; many of the examples come from bio-statistics. The emphasis is the motivation and theoretical properties of the algorithms presented. There is no discussion of the details needed for a computational implementation of the algorithms. This is a graduate text. It assumes the reader is fluent in calculus and linear algebra; it would also help if the reader has been introduced to elementary unconstrained optimization.

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

Back to index


Title NIKO'S NATURE: THE LIFE OF NIKO TINBERGEN AND HIS SCIENCE OF ANIMAL BEHAVIOUR.
Author H. Kruuk.
Publisher Oxford University Press. 2003, pp. xiv + 391, £20.00.

Contents:
1. Wild birds and science
2. A Dutch upbringing
3. Student years and Greenland
4. Ethologist in the 1930s
5. The Second World War and after
6. Starting again: Oxford in the 1950s
7. Niko's two worlds: Oxford in the 1960s
8. The Nobel Prize and human behaviour
9. Winding down
10. Niko's legacy

Niko Tinbergen's publications

Readership: Biologists, ethologists, and anyone interested in the development of observational sciences or in the evolution of one of the twentieth century's great biological scientists

This book is the first full biography of Niko Tinbergen, 1907-1988. Tinbergen and his friend Konrad Lorenz created a new branch of science-ethology: the study of animal behaviour. Tinbergen's early academic career was not outstanding; he scraped through at school and struggled to pass his PhD in 1932. But once he had found his metier, he made tremendous advances, winning numerous honours, including the Nobel Prize for Physiology or Medicine in 1973. Interestingly, his eldest brother Jan Tinbergen was awarded the Nobel Prize for Economics in 1969. (These two awards must have made it tough on their other siblings!)
Hans Kruuk, the author of this book was one of Tinbergen's students (along with such luminaries as Desmond Morris and Richard Dawkins) and became a good friend. In his preface, Kruuk thanks his friends and colleagues for supporting him while he wrote this book, since he had never attempted a biography before. Their confidence was well founded. This is a fascinating account of a fascinating man.

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

Back to index


Title MUSINGS OF THE MASTERS, AN ANTHOLOGY OF MATHEMATICAL REFLECTIONS.
Author R. Ayoub (Ed.).
Publisher Washington, D.C.: The Mathematical Association of America. 2004, pp. xvi + 277, US$46.50.

PART I: Mathematics and the Intellect
Mathematics and thinking mathematically, by Mary Cartwright.
Mathematical invention, by Henri Poincaré
Thoughts on the heuristic method, by Jacques Hadamard
Mathematical proof, by G.H. Hardy
The unity of knowledge, by Hermann Weyl
PART II: Mathematics and Human Understanding
Mathematics and the arts, by Marston Morse
Intuition, reason and faith in science, by George David Birkhoff
Logic and the understanding, by David Hilbert
The cultural basis of mathematics, by Raymond Wilder
PART III: Mathematics and Society
Presidential Address to the British Association, by J.J. Sylvester
The mathematician, by John von Neumann
The community of scholars, by André Lichnerowicz
History of mathematics: Why and how, by André Weil
PART IV: Miscellaneous
Does God exist?
by Paul Levy
Goethe and mathematics, by Wilhelm Maak
Leonardo and mathematics, by Francesco Severi
The highest good, by Norbert Wiener
Sources of the Articles

Readership: General reading; educators, scientists

Professor Raymond Ayoub has done a splendid job in collecting these articles by celebrated mathematicians on the nature of mathematics. In this introduction, Ayoub quotes Gauss, who wrote, "The enchanting charms of this sublime subject do not reveal themselves except to those who have the courage to plumb its depths" and this, perhaps, is the general feeling that pervades this collection. Mathematics is often viewed as a science done by a cold, unfeeling inteIlect. As these great mathematicians reveal, it is aIso an art guided by a sense of beauty. This book makes fascinating reading for a wide audience and it definitely gets a strong endorsement.

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

Back to index


Title GRAPHS, ALGORITHMS, AND OPTIMIZATION.
Author W. Kocay and D.L. Krecher.
Publisher Boca Raton, Florida: Chapman and Hall/CRC Press. 2005, pp. x + 483.

Contents:
1. Graphs and their complements
2. Paths and walks
3. Some special classes of graphs
4. Trees and cycles
5. The structure of trees
6. Connectivity
7. Alternating paths and matchings
8. Network flows
9. Hamilton cycles
10. Digraphs
11. Graph colorings
12. Planar graphs
13. Graphs and surfaces
14. Linear programming
15. The primal-dual algorithm
16. Discrete linear programming

Readership: Computer scientists, non-specialist mathematicians

This textbook presents graph theory from an algorithmic viewpoint and is suitable for a second or third year undergraduate course with a title such as "Graph and Network Algorithms". The algorithms are presented in a generic style so that they could be implemented in any major programming language. At the end of every chapter there are exercises (without solutions) and notes for further reading. The material is presented in a clear, accessible way; and the authors frequently exploit the visual nature of the graphs to illustrate their points.

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

Back to index


Title INSURANCE RISK AND RUIN.
Author D.C.M. Dickson.
Publisher Cambridge University Press. 2005, pp. xii + 229, £35.00/US$60.00.

Contents:
1. Probability distributions and insurance applications
2. Utility theory
3. Principles of premium calculation
4. The collective risk model
5. The individual risk model
6. Introduction to ruin theory
7. Classical ruin theory
8. Advanced ruin theory
9. Reinsurance

Readership: Students in insurance risk theory

This is a nice introduction to some basic aspects of risk theory, with special emphasis on utility theory, recursive methods and ruin theory.
The book is written in a style which is close to what a presentation in classroom might be: It is narrative and moves ahead from definitions to results in a flow of mathematical arguments and accompanying explanations. No formal statements of theorems are given, and only the core of a theorem is retained as a numbered formula which, of course, can be referred to but does not contain the conditions under which it is true. The author's intention to give a simple presentation and a tendency to avoid certain necessary but subtle probabilistic arguments is most evident (and also somewhat dangerous) in the chapters involving stochastic processes.
The book greatly benefits from numerical illustrations, numerous examples and many exercises and their solutions. What I appreciate most is the way in which the author explains the different topics dealt with in this book.

Reviewer:
Institute Technische Universitaet Dresden
Place Dresden, Germany
Name K.D. Schmidt

Back to index


Title CAPITAL MARKET INSTRUMENTS: ANALYSIS AND VALUATION, 2nd edition.
Author M. Choudhry, D. Joannas, R. Pereira, and R. Pienaar.
Publisher Basingstoke, U.K.: Palgrave MacMillan. 2005, pp. xxvii + 548, £125.00.

Contents:
PART I: Introduction
1. Introduction to financial market instruments
2. Market-determined interest rates, and the time value of money
PART II: Debt Capital Market Cash Instruments
3. Money market instruments and foreign exchange
4. Fixed income securities I: The bond markets
5. Fixed income securities II: Interest-rate risk
6. Fixed income securities III: Option-adjusted spread analysis
7. Interest rate modelling
8. Fitting the yield curve
9. B-spline modelling and fitting the term structure
10. Inflation-indexed bonds and derivatives
PART III: Structured Financial Products
11. An introduction to asset-backed bonds and securitisation
12. Mortgage-backed securities
13. Collateralised debt obligations
PART IV: Derivative Instruments
14. Short-term interest-rate derivatives
15. Swaps
16. Options I
17. Options II
18. Options III
19. Credit derivatives
PART V: Equity Capital Markets
20. Introduction to equity instrument analysis
21. Introduction to financial ratio analysis
PART VI: Risk Measurement and Value-at-Risk
22. Value-at-risk and credit VaR
PART VII: Rate Applications Software
23. RATE computer software

Readership: Students and professionals in the world of global derivatives. Front, middle, and back office staff in banks who are involved to some extent in the capital markets. Undergraduate and postgraduate students of finance and economics. Corporate and local authority treasurers, risk managers, capital market lawyers, auditors, and financial journalists. Graduate trainees in financial services and investment banking

As an application domain of mathematics, the financial sector is rather unusual because it is one of few areas where frontier mathematical research finds direct applications. As a consequence of this, although the first edition of this book appeared only three years ago, this new edition includes discussions of new financial instruments which have appeared in the interim. The focus is on techniques for analysis and valuation with practical application in mind. As the contents list indicates, this book is fairly comprehensive. It is also comprehensible, and I recommend it as a broad introduction to capital market instruments. It would form an excellent course text, though it would need to be supplemented by exercises.

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

Back to index


Title LUCK, LOGIC AND WHITE LIES -THE MATHEMATICS OF GAMES.
Author J. Bewersdorff. Translated by D. Kramer.
Publisher Wellesley, Massachusetts: A.K. Peters. 2005, pp. xvii + 486, US$44.00.

Contents:
PART I: Games of Chance
1. Dice and probability
2. Waiting for a double 6
3. Tips on playing the lottery: More equal than equal?
4. A fair division: But how?
5. The red and the black: The law of large numbers
6. Asymmetric dice: Are they worth anything?
7. Probability and geometry
8. Chance and mathematical certainty: Are they reconcilable?
9. In quest of the equiprobable
10. Winning the game: Probability and value
11. Which die is best?
12. A die is tested
13. The normal distribution: A race to the finish!
14. And not only at roulette: The Poisson distribution
15. When formulas become too complex: The Monte Carlo method
16. Markov chains and the game Monopoly
17. Blackjack: A Las Vegas fairy tale
PART II: Combinatorial Games
18. Which move is best?
19. Chances of winning and symmetry
20. A game for three
21. Nim: The easy winner!
22. Lasker Nim: Winning along a secret path
23. Black-and-white Nim: To each his (or her) own
24. A game with dominoes: Have we run out of space yet?
25. Go: A cIassical game with a modern theory
26. Misère games: Loser wins!
27. The computer as game partner
28. Can winning prospects always be determined?
29. Games and complexity: When calculations take too long
30. A good memory and luck: And nothing eIse?
31. Backgammon: To double or not to double?
32. Mastermind: Playing it safe
PART III: Strategic Games
33. Rock-paper-scissors: The enemy's unknown plan
34. Minimax versus psychology: Even in poker?
35. Bluffing in poker: Can it be done without psychology?
36. Symmetric games: Disadvantages are avoidable, but how?
37. Minimax and linear optimization: As simple as can be
38. Play it again, Sam: Does experience make us wiser?
39. He/Her: Should I exchange?
40. Deciding at random: But how?
41. Optimal play: Planning efficiently
42. Baccarat: Draw from a five?
43. Three person poker: Is it a matter of trust?
44. QUAAK! Child's play?
45. Mastermind: Color codes and minimax

Readership: General readership, particularly game-players

This book is a translation of a book which was first published in German by Vieweg & Sohn Verlag in 2001. The author is (among other things) general manager of a game design company based in Limburg, Germany and the book is written to reach as broad a readership as possible.
The range is very wide and in broadly three categories. As the contents indicate, these are Games of Chance ('Luck'), Combinatorial Games ('Logic'), Strategic Games ('White Lies').
In the first of these categories, of course, there are features of the games (e.g. cards, dice, lotteries) which are not within the control of the player(s), even if probabilities of these features are known. In the second category are games (e.g. chess, checkers, nim, go) in which complete knowledge of the current state of play-and how it was arrived at-is known to the players, but where complexity comes from the vast numbers of moves and consequences that then need to be evaluated. Finally, in the third category are games (e.g. Rock-Paper-Scissors, Mastermind, poker) where knowledge is definitely incomplete and the goal is to ascertain and exploit the strategy of an opponent, while protecting one's own strategy. This book is certainly not a textbook. As the author admits, the mathematical detail given is too superficial and incomplete for experts, but there is a wealth of information and results to satisfy the most avid taste. The general readership should find it very fascinating and motivating to get to know more; these such readers who wish to follow up specifics, but also the experts, should find the multitude of references given throughout the text and in extensive chapter notes a mine of information. These references are also useful in putting aspects of the games in a historical perspective. The style of the book is engaging and attractive.

Reviewer:
Institute Imperial College of Science, Technology and Medicine
Place London, U.K.
Name F.H. Berkshire

Back to index


Title FORECASTING PRODUCT LIABILITY CLAIMS: Epidemiology and Modeling in the Manville Asbestos Case.
Author E. Stallard, K.G. Manton and J.E. Cohen.
Publisher New York: Springer-Verlag. 2005, pp. xxix + 394, US$84.95.

Contents:
1. Overview
2. Epidemiology of asbestos-related diseases
3. Forecasts based on direct estimates of exposure
4. Forecasts based on indirect estimates of exposure
5. Uncertainty in forecasts based on indirect estimates
6. Updated forecasts based on indirect estimates of exposure
7. Uncertainty in updated forecasts
8. Forecasts based on a hybrid model
9. Uncertainty in forecasts based on a hybrid model
10. Conclusions and implications

Readership: General

Over 750,000 claimants have filed suit against some 8000 U.S. companies for illnesses and deaths related to exposure to asbestos, and at least 65 companies had been driven to bankruptcy; these numbers continue to grow. The largest producer was the Johns-Manville Corporation, whose name has been most prominently attached to the problems of asbestos (and this book). Claims may eventually reach $210 billion dollars (US). These authors were members of a six-person panel that advised the presiding judge about the probable future course of the numbers and sizes of new claims, as an aid to allocating the much smaller available funds fairly. This work led to substantial advances in the art of forecasting the number, timing, and nature of new claims. The authors present a lucid explanation of these advances and a description of how the matters in litigation have been settled. The presiding judge, Jack Weinstein, has contributed an informative preface and a plea for the broad changes needed to deal with the medical-legal-entrepeneurial system that created such a mess.

Reviewer:
Institute University of Chicago
Place Chicago, U.S.A.
Name J.C. Bailar

Back to index


Title SAUNDERS MAC LANE. A MATHEMATICAL AUTOBIOGRAPHY.
Author S. Mac Lane.
Publisher Wellesley, Massachusetts: A.K. Peters, 2005, pp. xvi + 358, US$39.00. Contents:

PART One: Early Years
PART Two: First Teaching
PART Three: Collaborative Research
PART Four: The War Years
PART Five: Eilenberg and Mac Lane
PART Six: Harvard Years
PART Seven: Chicago in the Fifties
PART Eight: Mathematical Developments
PART Nine: National Academy of Science
PART Ten: The Sixties and Beyond
PART Eleven: National Science Policy
PART Twelve: Travels
PART Thirteen: Advising
PART Fourteen: Later Developments
PART Fifteen: Contemplating

Readership: Mathematicians

Saunders Mac Lane is one of the pioneers of 20th century mathematics. His seminal contributions to the theory of cohomology of groups and homological algebra led to the creation of "category theory", a fundamental formalism that encapsulates theorems from diverse areas of mathematics into visual images consisting of objects and morphisms. In this captivating autobiography, Mac Lane takes us back to the early years of his mathematical youth. For example, in Chapter 12 he describes how the quaternions inspired the construction of crossed product algebras and the discovery of group extensions. From then on, the subject develops naturally connecting itself to algebraic topology, number theory and ultimately, algebraic geometry. The book is interspersed with amusing anecdotes along with a description of his famous collaborative work with Eilenberg. I would recommend this biography to any serious mathematics student. Modern algebra often appears dry, but the reading of this book will do much to show it is a fertile creation of marvellous minds.

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

Back to index


Title THE GRAMMAR OF GRAPHICS, 2nd edition.
Author L. Wilkinson.
Publisher New York: Springer-Verlag, 2005, pp. xviii + 690, US$79.95. [Original, 1999; Short Book Reviews, Vol. 20, p. 3] Contents:

1. Introduction
2. How to make a pie
3. Data
4. Variables
5. Algebra
6. ScaIes
7. Statistics
8. Geometry
9. Coordinates
10. Aesthetics
11. Facets
12. Guides
13. Space
14. Time
15. Uncertainty
16. Analysis
17. Control
18. Automation
19. Reader
20. Coda

Readership: Statisticians, computer scientists and others interested in visualizing data

To most people who work with data the production of graphical displays, through a favourite computer package, has become something that is completely taken for granted. If one thinks of the process at all it is only when one is frustrated when one cannot get the package to do exactly what one wants. This fascinating book deconstructs the process of producing graphics and in doing so raises many fascinating questions on the nature and representation of information. The book is split into two parts; syntax, where the author discusses an abstract framework for manipulating data into graphics, and semantics in which he discusses the meaning, or lack of it, underlying displays. The approach is multi-disciplinary and illustrates the power of how good, well thought-out abstraction can illuminate and give insight to general problems. This second edition is almost twice the size of the originaI, with six new chapters and substantial revisions.

Reviewer:
Institute University of Waterloo
Place Waterloo, Canada
Name P. Marriott

Back to index


Title INFÉRENCE ET PRÉVISION EN GRANDES DIMENSIONS.
Author D. Bosq.
Publisher Paris: Economica, 2005, pp. 194. Première partie: Prévision

1. Prévision statistique
2. Prévision asymptotique
Deuxième partie: Inférence par Projection
3. Estimation par projection adaptive
4. Estimation par projection adaptive dans les processus
5. Tests fonctionnels
6. Prédiction non paramétrique
Troisième partie: Processus Linéaires en Grandes Dimensions
7. Processus linéaires fonctionnels
8. Estimation et prévision des processus linéaires fonctionnels

Lecture: Probabilistes et statisticiens

Dans cet ouvrage I'auteur nous offre une étude bien détaillée sur I'inférence et la prévision dans le cas où les données et/ou le paramètre sont en grande dimension (éventuellement infinie). L'accent est sur des méthodes non paramétriques. Plus précisément la projection adaptive est utilisée pour estimer par exemple une densité, une fonction de régression, une densité spectrale, etc. On considère aussi des tests d'ajustement. Des résultats nouveaux apparaissent sur les processus linéaires en dimension infinie. Le traitement est basé sur la théorie de la mesure et I'inférence statistique mathématique.

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

Back to index


Title A MODERN INTRODUCTION TO PROBABILITY AND STATISTICS. UNDERSTANDING WHY AND HOW.
Author F.M. Dekking, C. Kraaikamp, H.P. Lopuhaä and L.E. Meester.
Publisher London: Springer-Verlag, 2005, pp. xv + 486, US$49.95. Contents:

1. Why probability and statistics?
2. Outcomes, events, and probability
3. Conditional probability and independence
4. Discrete random variables
5. Continuous random variables
6. Simulation
7. Expectation and variance
8. Computations with random variables
9. Joint distributions and independence
10. Covariance and correlation
11. More computations with more random variables
12. The Poisson process
13. The law of large numbers
14. The central limit theorem
15. Exploratory data analysis: Graphical summaries
16. Exploratory data analysis: Numerical summaries
17. Basic statistical models
18. The bootstrap
19. Unbiased estimators
20. Efficiency and mean squared error
21. Maximum likelihood
22. The method of least squares
23. Confidence intervals for the mean
24. More on confidence intervals
25. Testing hypotheses: Essentials
26. Testing hypotheses: Elaboration
27. The t-test
28. Comparing two samples

APPENDIX A: Summary of Distributions
APPENDIX B: Table of the Normal and t-distributions
APPENDIX C: Answers to Selected Exercises
APPENDIX D: Full Solutions to Selected Exercises

Readership: Undergraduate students in mathematics, engineering, econometrics

This textbook provides a well-written first course in probability and statistics. The text is self-contained with fourteen chapters on probability (from probability rules to the central limit theorem) and fourteen chapters on statistics (from exploratory data analysis to the two sample t-test). There are of course many introductory books on the market, but the present one can show a number of pluses. The focus is on understanding the methods intuitively (by using interesting examples) and to deal with them at an acceptable mathematical level. Another plus is the wealth of more than three hundred and fifty exercises, going from easy to more challenging. Many of them have shorter or longer solutions in the Appendix. It is a book that has been written based on the long teaching experience of the authors and I would certainly recommend it for university coursework.

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

Back to index


Title PROBABILITY: A GRADUATE COURSE.
Author A. Gut.
Publisher New York: Springer-Verlag, 2005, pp. xxiii + 603, US$79.95. Contents:

1. Introductory measure theory
2. Random variables
3. Inequalities
4. Characteristic functions
5. Convergence
6. The law of large numbers
7. The central limit theorem
8. The law of the iterated logarithm
9. Limit theorems: Extensions and generalizations
10. Martingales

APPENDIX A: Some Useful Mathematics

Readership: Those who need a sound knowledge of undergraduate real analysis and mathematical statistics

This is more substantial than the usual graduate course in probability; it contains many useful and interesting details that previously were scattered around the literature and gives clear evidence that the writer has a great deal of experience in the area. It is a thorough treatment and the author has been careful to ensure transparency in proofs. Indeed he remarks that the literature of the subject is replete with highly abbreviated proofs containing large gaps in logic which are dismissed with comments such as "it is clear that ...". He has done an excellent job in ensuring such gaps are very few in number; it is easy to dip into a chapter, readily pick up the argument and follow it through. The author writes in a clear-headed straight-forward style and provides plenty of exercises at the end of each chapter. The material covered goes beyond most graduate texts on probability; for example the extensive discussion of rates of convergence in the Central Limit Theorem, the chapter on the Law of the Iterated Logarithm, the detailed treatment of the Strong Law of Large Numbers, etc. These give the book some of the flavour of Feller Volume 2. It would be an excellent text for those who wish to develop a research interest in probability theory or a related area. The author recommends virtually all of it (except Chapter 9) as forming a graduate course. However, unhappily there would be very few universities in Australia now that would be able to run a viable graduate course covering the contents of this book. The decline in the level of mathematical knowledge of undergraduates coupled with the need to retain student numbers has led to a purging, or at least a dilution, of mathematics from many statistics programs. This of course is not unique to Australia; it seems to be happening everywhere.

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

Back to index


Title EPIDEMIOLOGY: Study Design and Data Analysis, 2nd edition.
Author M. Woodward.
Publisher Boca Raton, Florida: Chapman and Hall/CRC Press, 2005, pp. xxii + 849, US$79.95/£39.99. Contents:

1. Fundamental issues
2. Basic analyticaI procedures
3. Assessing risk factors
4. Confounding and interaction
5. Cohort studies
6. Case-control studies
7. Intervention studies
8. Sample size determination
9. Modelling quantitative outcome variables
10. Modelling binary outcome data
11. Modelling follow-up data
12. Meta-analysis

Readership: Researchers wanting to understand statistical methods applied in epidemiology and statisticians wanting to examine applications in epidemiology

This second edition [first edition, 2000; Short Book Reviews, Vol. 20, p. 2] has added about 150 pages with a new chapter on meta analysis accounting for about fifty pages and a companion web site.
Discussions of various epidemiological research designs, and of concepts such as confounding and interaction are very well presented. The chapter on sample size describes fairIy basic methods, but common extensions such as to correlations or studies involving more than two conditions are not included. In the modeling chapters, for the most part, explanations are clear (for example, of the need for special models for binary data) for those with knowledge of logarithms, the exponential function and probability distributions. However, in an attempt to cover many different situations, cursory details are given; for example, the Poisson regression is presented with almost no discussion of the Poisson distribution, its properties and uses. Models for (matched) case-control studies receive very little attention. As well, estimates and tests for generalized linear models, survival models, etc. appear in the text through SAS computer outputs without discussion of their origin. This book provides very good coverage of major issues in the design of epidemiological studies, and a decent, but very quick, tour of commonly used statistical models for such studies.

Reviewer:
Institute University of Waterloo
Place Waterloo, Canada
Name K.S. Brown

Back to index


Title ANALYZING ENVIRONMENTAL DATA.
Author W.W. Piegorsch and A.J. Bailer.
Publisher Chichester, U.K.: Wiley, 2005, pp. xv + 496, £45.00. Contents:

1. Linear regression
2. Nonlinear regression
3. Generalized linear models
4. Quantitative risk assessment with stimulus-response data
5. Temporal data and autoregressive modelling
6. Spatially correlated data
7. Combining environmental information
8. Fundamentals of environmental sampling

APPENDIX A: Review of Probability and Statistical Inference
APPENDIX B: Tables

Readership: Students and researchers in environmental statistics

This is a substantial and thorough book which covers a large variety of the statistical techniques which are of use in environmental statistics. Most of the techniques are demonstrated by application to real data, either directly in the text (with frequent reference to sample SAS programs) or embedded in the many exercises-the larger sets of data having been made available at the publishers' website. Anyone who mastered the contents of this book would be well equipped as an applied statistician in a more general sense, not just restricted to the analysis of environmental data. Indeed, the neat summaries of topics, as varied as overdispersion, Weibull growth curves, bootstrap confidence intervals and semivariograms, etc., might make this a handy reference book for any statistician's bookshelf.

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

Back to index


Title WEIGHT-OF-EVIDENCE FOR FORENSIC DNA PROFILES.
Author D.J. Balding.
Publisher Chichester, U.K.: Wiley, 2005, pp. x + 184, £45.00. Contents:

1. Introduction
2. Crime on an Island
3. Assessing evidence via likelihood ratios
4. Typing technologies
5. Some popular genetics for DNA evidence
6. Identification
7. Relatedness
8. Other approaches to weight of evidence
9. Issues for the courtroom
10. Solutions to exercises

Readership: Forensic scientists, expert witnesses

As the author states, "the analysis of forensic DNA profiles is a complex topic"; but this book should provide a good starting point for any reader with a reasonably developed background in both statistics and genetics. There are set exercises to assist such a reader in developing his or her understanding. Possibly the most important formula in genetics is stated to be F=1/(1+4N?), which is shown to yield the curious implication that the effective size of the human population (i.e. N) for displaying all observed genetic variation is about 10,000 people! The author proposes that "evidential weight can be measured by likelihood ratios and combined to assess the totality of the evidence using the appropriate version of Bayes Theorem." That this can be highly controversial in a legal context is acknowledged and carefully analyzed by commentary on several of the important cases, including the well-known case of Regina versus D.J. Adams which went to the Appeal Court twice.

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

Back to index


Title STATISTICS FOR EXPERIMENTERS: DESIGN, INNOVATION AND DISCOVERY, 2nd edition.
Author G.E.P. Box, J.S. Hunter and W.G. Hunter.
Publisher Hoboken, New Jersey: Wiley, 2005, pp. xvii + 633, £54.50. Contents:

1. Catalyzing the generation of knowledge
2. Basics (probability, parameters, and statistics)
3. Comparing two entities: Reference distributions, tests, and confidence intervals
4. Comparing a number of entities, randomized blocks and Latin squares
5. Factorial designs at two levels
6. Fractional factorial designs
7. Additional fractionals and analysis
8. Factorial designs and data transformation
9. Multiple sources of variation
10. Least squares and why we need designed experiments
11. Modeling, geometry, and experimental design
12. Some applications of response surface methods
13. Designing robust products and processes: An introduction
14. Process control, forecasting, and time series: An introduction
15. Evolutionary process operation

Readership: Experimenters in science and engineering, statisticians

This is a welcome second edition of a much-
Ioved book. The new subsidiary title reflects how the statistical design and analysis of experiments have become key elements of scientific advance and training since the book first appeared in 1978.
Much of the material from the first edition has been rewritten and updated, as have the valuable references, further reading, exercises and problems at the end of the chapters. Many new topics have been included, such as Bayesian methods for identifying active factors from small experiments, and for selecting follow-on runs for an experiment to guide the choice between competing models. The treatment of other topics has been expanded, for example, split-plot designs and components of variance. A particularly useful addition is an extensive discussion of the "hands on" design of a paper helicopter. This is an excellent means of conveying understanding of basic principles and techniques of experimentation to people as diverse as industrial engineers and mathematical undergraduates.

Reviewer:
Institute University of Southampton
Place Southampton, U.K.
Name S.M. Lewis

Back to index


Title STATISTICAL DESIGN OF EXPERIMENTS WITH ENGINEERING APPLICATIONS.
Author K. Rekab and M. Shaikh.
Publisher Boca Raton, Florida: Chapman and Hall/CRC Press, 2005, pp. x + 252, US$89.95/£49.99. Contents:

1. Introduction
2. Designing and conducting the experiment
3. Optimization of the location parameter
4. Minimization of the dispersion
5. Taguchi's approach to the design of experiments
6. Statistical optimization of the location parameter
7. Statistical minimization of the dispersion parameter
8. Validity of the prediction equation
9. Three-Ievel factorial designs
10. Second order analysis

APPENDIX 1: Two-Level Fractional Factorial Designs
APPENDIX 2: Plackett-Burman Designs
APPENDIX 3: Taguchi Designs
APPENDIX 4: Standardized Normal Distribution
APPENDIX 5: Percentiles of the t Distribution
APPENDIX 6: Percentiles of the F Distribution
APPENDIX 7: Some Useful Box-Behnken Designs
APPENDIX 8: Matrix Algebra

Readership: Experimenters who want a beginning text in response surface methodology

This volume offers a pared-down tour of response surfaces that "avoids frustrating and unnecessary time spent on theory". Although it has the feel of a series of lists, it provides a useful introduction to readers who are happy with such an approach. Some of the references to books are dated and refer to earlier editions, and Figure 9.6 is incomplete. The "problems" (exercises) at the ends of the chapters are mostly uninspiring, 29 out of 39 asking for a repetition of book material. Of the ten problems with data, one refers to an example in the text, six form parts of only two distinct problems. Of the remaining three, the single Chapter 3 problem is a version of the Box and J.S. Hunter example in Technometrics, 1961, pp. 334-337. The example also appears in the Box, Hunter and Hunter book Statistics for Experimenters, first edition, pp. 424-429, and in the 2005 second edition, pp. 252-257.

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

Back to index


Title EXPERIMENTAL DESIGN FOR FORMULATION.
Author W.F. Smith.
Publisher Philadelphia: Society for Industrial and Applied Mathematics/Alexandria, Virginia: American Statistical Association, 2005, pp. xix + 367, US$105.00.

1. Introduction
2. Mixture space
3. Models for a mixture setting
4. Designs for simplex-shaped regions
5. Designs for non-simplex shaped regions
6. Design evaluation
7. Blocking mixture experiments
8. Building models in a mixture setting
9. Model evaluation
10. Model revision
11. Effects
12. Optimization
13. Including process variables
14. Collinearity

Readership: Scientists interested in experiments with mixtures

The author gained his experience in this area working for Eastman Kodak and with the help of John Cornell of mixture fame, whom he characterizes as his mentor. Both can be proud of this volume which offers a "step-by-step guide to the design and analysis of experiments involving formulations" (namely mixtures). Recommended.

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

Back to index


Title TESTING STATISTICAL HYPOTHESES, 3rd edition.
Author E.L. Lehmann and J.P. Romano.
Publisher New York: Springer-Verlag, 2005, pp. xvi + 789, US$89.95. Contents:

PART I: Small Sample Theory
1. The general decision problem
2. The probability background
3. Uniformly most powerful tests
4. Unbiasedness: Theory and first applications
5. Unbiasedness: Applications to normal distributions
6. Invariance
7. Linear hypotheses
8. The minimax principle
9. Multiple testing and simultaneous inference
10. Conditional inference
PART II: Large-Sample Theory
11. Basic large-sample theory
12. Quadratic mean differentiable families
13. Large-sample optimality
14. Testing goodness of fit
15. General large-sample methods

APPENDIX A: Auxiliary Results

Readership: Graduate students and senior undergraduate students in statistics and related fields

The third edition of this book differs principally from the second in that the treatment of asymptotic optimality is greatly enhanced in Part II. There is also a more extensive discussion of goodness-of-fit as well as multiple testing and an introduction to the bootstrap and related techniques. This new edition of the classic and fundamental text on the theory of testing hypotheses is an essential addition to the bookshelf of mathematical statisticians.

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

Back to index


Title MATHEMATICAL STATISTICS WITH APPLICATIONS.
Author A.S. Kapadia, W. Chan and L. Moyé.
Publisher Boca Raton, Florida: Chapman and Hall/CRC Press, 2005, pp. xxiii + 617, US$89.95/£49.99. Contents:

Introduction
1. Review of mathematics
2. Probability theory
3. Random variables
4. Discrete distributions
5. Continuous random variables
6. Distributions of order statistics
7. Asymptotic distribution theory
8. Point estimation
9. Hypothesis testing
10. Interval estimation
11. Introduction to computational methods

Readership: Scientists who need mathematical statistics for their applications

The book starts with an introductory chapter, Chapter 1, of forty-five pages on mathematical concepts and tools. Personally I would prefer to see this review in the appendix of the book since it is not very inviting for the rest of the book. The remaining ten chapters are a rather classical introduction to probability and statistics using a lot of calculus. Each chapter has a number of exercises, in total about two hundred and fifty. No solutions or hints are provided. Although the topics are the ones you can find in numerous other books, it is worth mentioning that the authors also deal with some Bayesian inference, some nonparametric tests and some modern computational techniques such as EM, Gibbs sampler and MCMC.

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

Back to index


Title CONSTRUCTING MEASURES: AN ITEM RESPONSE MODELING APPROACH.
Author M. Wilson.
Publisher Mahwah, New Jersey: Lawrence Erlbaum, 2005, pp. xix + 228, US$59.95. Contents:

PART I: A Constructive Approach to Measurement
1. Construct modeling: The "four building blocks" approach
PART II: The Four Building Blocks
2. Construct maps
3. The item designs
4. The outcome space
5. The measurement model
PART III: Quality Control Methods
6. Choosing and evaluating a measurement model
7. Reliability
8. Validity
PART IV: A Beginning Rather Than A Conclusion
9. New steps in measuring

APPENDIX 1: The Cases Archive (on CD)
APPENDIX 2: Grade Map (on CD)

Readership: Advanced students of item test, or instrument development, measurement, item response theory, or Rasch analysis in a variety of departments including education and psychology, and those developing instruments in industrial, organizational, educational and other contexts

This book describes the process of constructing scales for measuring phenomena in the behavioural and social sciences. It proceeds through four stages. The construct map presents 'a coherent and substantive definition for the content' in the form of a uni-dimensional continuum. The items design describes how to capture observations which bear on the underlying concept being measured. The outcome space describes how responses to items are coded. And the measurement model relates the items and their outcome codes back to the underlying concept, so enabling observations on the former to shed light on the value of the latter. This last section essentially gives a very informal introduction to item response models. Part III of the book introduces issues of reliability and validity in the context of evaluating such measurement procedures. From a statistical perspective, the book is elementary. From the perspective of someone new to measurement issues, and coming from an educational or psychological background, I think it would provide an excellent initial introduction to the ideas and methods.

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

Back to index


Title CORRESPONDENCE ANALYSIS AND DATA CODING WITH JAVA AND R.
Author F. Murtagh.
Publisher Boca Raton, Florida: Chapman and Hall/CRC Press, 2005, pp. 230, US$79.95/£44.99. Contents:

1. Introduction
2. Theory of correspondence analysis
3. Input data coding
4. Examples and case studies
5. Content analysis of text
6. Concluding remarks

Readership: Statisticians, computer scientists, researchers in many disciplines

The structure and relationships in multivariate data is often not revealed by simple summary statistics and correlations, or chi-squared tests. This book shows how correspondence analysis can display these hidden structures. A complete guide to the implementation of this technique is offered. Detailed examples of its application to data are drawn from an astonishingly wide variety of fields; astronomy, financial modeIing and forecasting, comparisons of prehistoric and modern groups of dogs, ancient goblets and measurements on ancient Egyptian skulls.
An entire chapter is devoted to the analysis of textual data by correspondence analysis; an approach which is quite different from the current conventional methods. Based on the frequencies of "tool" words such as "to", "and" and "the", this method is relatively easy to automate. It can distinguish between different forms of writing, such as novels, fairy tales and technical reports; between authors and between works by the same author.
However, this book goes beyond applications. A terse presentation of the mathematical basis of correspondence analysis is included and full listings of programs written in R to implement the examples are given. These can be downloaded from the web and are suitable for use with Windows, Mackintosh or Unix systems. Similar programs are also available in Java.
Often data has to be categorized before doing a correspondence analysis. The recommendations given on how this should be done to avoid distortion and minimize the inevitable loss of information, could be usefully applied in many other situations.
All in all this book can be recommended as a succinct reference on all aspects of correspondence analysis, theoretical, computational and practical.

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

Back to index


Title CLUSTERING FOR DATA MINING: A DATA RECOVERY APPROACH.
Author B. Mirkin.
Publisher Boca Raton, Florida: Chapman and Hall/CRC Press, 2005, pp. xxiii + 266, US$79.95/£44.99. Contents:

1. What is clustering?
2. What is data?
3. K-means clustering
4. Ward hierarchical clustering
5. Data recovery models
6. Different clustering approaches
7. General issues

Conclusions: Data recovery approach in clustering

Readership: Readers new to cluster analysis, and researchers in cluster analysis methods

Cluster analysis is characterized by its wide range of techniques and its ad hoc nature. Recent years have seen various attempts to formulate a sounder theoretical base, often in the form of model-based approaches. This book also attempts to establish a stronger foundation, though from a rather different perspective. This perspective is indicated by the word 'recovery' in the book's title, indicating that the clusters should be regarded as summaries of the data in the sense that good approximations to the data could be generated ('recovered') if one knew the cluster structure. This yields a decomposition of the data into that part which is explained by the cluster structure, and that part which is not-in a way similar to the decomposition of many other standard statistical tools into explained and unexplained variation. The particular decomposition studied in this book is the decomposition of the total sum of squares matrix into between and within cluster components, and the book develops this decomposition, and its associated diagnostics, further than I have seen them developed for cluster analysis before. Overall, the book presents an unusual, perhaps even rather idiosyncratic approach to cluster analysis, from the perspective of someone who is clearly an enthusiast for the insights these tools can bring to understanding data.

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

Back to index


Title MINING IMPERFECT DATA: DEALING WITH CONTAMINATION AND INCOMPLETE RECORDS.
Author R.K. Pearson.
Publisher Philadelphia: Society for Industrial and Applied Mathematics, 2005, pp. x + 305, US$70.00. Contents:

1. Introduction
2. Imperfect datasets: Character, consequences, and causes
3. Univariate outlier detection
4. Data pretreatment
5. What is a "good" data characterization?
6. GSA
7. Sampling schemes for a fixed dataset
8. Concluding remarks and open questions

Readership: Anyone concerned with data mining who has to confront imperfect data

Modern statistics is such a large subject that it is difficult to know which parts to cover when teaching, and which to omit, but the training of every statistician should include an introduction to issues of imperfect data. After all, outside the confines of the text book all data are likely to be imperfect: poor quality data is a ubiquitous problem, and all real data should be approached with suspicion. For this reason, I am pleased to see this book. Although the title refers to data mining, it seems to me to cover material which one might find in a statistics text on the problem: it is concerned with model building in the face of imperfect data, rather than anomaly detection for its own sake.
The author tackles the detection and resolution of data quality problems in data mining in two phases. The first is data pretreatment, concerned with detecting and handling outliers, missing data, noninformative variables, misalignments, and other anomalies. The second is analytical validation, concerned with assessing structures once they have been found to decide if they could be merely a consequence of poor quality data. This second aspect is tackled via the notion of generalized sensitivity analysis, the GSA of the contents list, an informal graphical approach to exploring the sensitivity of an analysis to changes in a set of data, and the particulars of which appear to be the author's own development.
One of the difficulties with attempts at general treatments of quality of data is that many of the issues are context dependent. I think that this inevitably means that such treatments run the risk of some idiosyncrasy, and this book is no exception. Nevertheless, the tools described here will find frequent use when mining perfect data.

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

Back to index


Title APPLIED LINEAR REGRESSION, 3rd edition.
Author S. Weisberg.
Publisher Hoboken, New Jersey: Wiley, 2005, pp. xvi + 310, £48.95. Contents:

1. Scatterplots and regression
2. Simple linear regression
3. Multiple regression
4. Drawing conclusions
5. Weights, lack of fit, and more
6. Polynomials and factors
7. Transformations
8. Regression diagnostics: Residuals
9. Outliers and influence
10. Variable seIection
11. Nonlinear regression
12. Logistic regression

Readership: Statisticians, scientists and engineers using regression models

This is an up-dated third edition [1980, 1st edition; Short Book Reviews, Vol. 1, p. 2] of a well-presented text on the theory and practice of regression analysis, illustrated by many examples and supported by a wealth of sets of data available for downloading from a dedicated website. The book begins with a new Chapter 1 introducing the ideas of graphing the data in various ways and using smoothers to assist with the identification of any relationships which might be present. This is followed by two chapters on the theory of least squares in simple and multiple regression, each with a set of practical examples and exercises. Chapter 4 on Drawing Conclusions focuses on understanding the interpretation of the model parameter estimates, the effects of dropping terms, the meaning of R2 and dealing with missing data. A short section on computationally intensive methods such as the bootstrap concludes this chapter. There follow chapters on weighted regression, assessing lack-of-fit, the use of polynomials and factors, and on the use of transformations of the response variable and/or the predictor variables. Later chapters emphasize the importance of diagnostic tools to examine the models and to investigate for outliers and influential points. Much changed from earlier editions is the chapter on variable selection, collinearity and the use of predicted residuals. The last two chapters provide an introduction to non-linear regression and logistic regression respectively. The material is well explained and clearly illustrated, with numerous examples and end-of-chapter exercises using the sets of data available. The theory uses matrix notation but little calculus, with the emphasis on understanding the methods employed. This is an excellent book which could easily be used as a course text.

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

Back to index


Title FUNCTIONAL DATA ANALYSIS, 2nd edition.
Author J.O. Ramsey and B.W. Silverman.
Publisher New York: Springer-Verlag, 2005, pp. xxix + 426, [Original 1997, Short Book Reviews, Vol. 18, p. 11] Contents:

1. Introduction
2. Tools for exploring functional data
3. From functional data to smooth functions
4. Smoothing functional data by least squares
5. Smoothing functional data with a smoothness penalty
6. Constrained functions
7. The registration and display of functional data
8. Principal components analysis for functional data
9. Regularized principal components analysis
10. Principal components analysis for mixed data
11. Canonical correlation and discriminant analysis
12. Functional linear models
13. Modelling functional responses with multivariate covariates
14. Functional responses, functional covariates and the current model
15. Functional linear models for scalar responses
16. Functional linear models for functional responses
17. Derivatives and functional linear models
18. Differential equations and operators
19. Principal differential analysis
20. Green's functions and reproducing kernels
21. More general roughness penalties
22. Some perspectives on FDA
Readership: Researchers and statisticians working with repeated measures

The first edition presented a large, coherent collection of techniques for the analysis of data of repeated measures of smooth functions. Growth curves are an example. This second edition, more than a third longer, presents a significant expansion. New analytic and graphical tools have been added. Approximate confidence intervals are included. The topics are introduced with more discussion and the examples are described in greater detail. This edition is useful to a broader audience.
This is a book for data analysts. It begins "Figure 1.1 provides ... data". The book is a valuable source of techniques. The author's software is available. Exploratory graphical methods are uniquely useful in learning from data. While not all will agree with the interpretations of the results (see page 121), the discussion illustrates the great potential of what might be discovered.

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

Back to index


Title IMAGE PROCESSING AND JUMP REGRESSION ANALYSIS.
Author P. Qiu.
Publisher Hoboken, New Jersey: Wiley, 2005, pp. ix + 301, US$94.95. Contents:

1. Introduction
2. Basic statistical concepts and conventional smoothing techniques
3. Estimation of jump regression curves
4. Estimation of jump location curves of regression surfaces
5. Jump-preserving surface estimation by local smoothing
6. Edge detection in image processing
7. Edge-preserving image restoration

Readership: Image analysts and statisticians

The human eye is extremely good at image analysis in particular at detecting edges and non-smooth features in complex images. Such a task is non-trivial for many conventional statistical regression-based methods that are based on local smoothness. Jump regression analysis attempts to bridge this gap. This book would be suitable for a graduate level course on statistically based image analysis with the emphasis on smoothing and edge detection. Such a course would only require a basic knowledge of statistical methodology. It takes a fairly practical approach to topics with application emphasized over proof.

Reviewer:
Institute University of Waterloo
Place Waterloo, Canada
Name P. Marriott

Back to index


Title STATISTICS FOR FISSION TRACK ANALYSIS.
Author R.F. Galbraith.
Publisher Boca Raton, Florida: Chapman and Hall/CRC Press, pp. xiv + 219, US$79.95/£44.99. Contents:

1. Introduction
2. The Poisson line segment model
3. Track counts and densities: Fission track dating
4. The population method
5. Discrete mixture of ages
6. Continuous mixture of ages
7. Probability distributions of lengths and angles
8. Observational features of track measurements
9. Further developments

Readership: Experimental scientists and statisticians
The book is a good survey for scientists who have worked in the field. The book deals mainly with length measurements as deduced from fission tracks. The book does not mention all the recent advances in stereology, image analysis and stochastic geometry which deal with most of the problems addressed in this text.
This would be an excellent introductory text for those wishing to enter this field, especially if they were also aware of the new statistical tools available. It would be an asset to have data sets at the end of chapters to give readers the chance to practice the theory cited and to perhaps try some new theories on the same data.

Reviewer:
Institute University of Calgary
Place Calgary, Canada
Name E. Enns

Back to index


Title THE TEN MOST WANTED SOLUTIONS IN PROTEIN BIOINFORMATICS.
Author A. Tramontano.
Publisher Boca Raton, Florida: Chapman and Hall/CRC Press, 2005, pp. xx + 186, US$69.95/£39.99. Contents:

Introduction
Problem 1. Protein sequence alignment
Problem 2. Predicting protein features from the sequence
Problem 3. Function prediction
Problem 4. Protein structure prediction
Problem 5. Membrane proteins
Problem 6. Functional site identification
Problem 7. Protein-protein interaction
Problem 8. Protein-small molecule interaction
Problem 9. Protein design
Problem 10. Protein engineering
Conclusions

Readership: Students and newcomers to biostatistics either from biology or from computational mathematics and statistics

Bioinformatics is a fuzzy mixture of biology, mathematics and engineering. To introduce newcomers to this area is not an easy task and this book is a serious attempt to help many of them. The author is balancing somewhere between a popular science book and a medium level textbook. The book would be more attractive if it were not overloaded with high tech terminology (biological and in computational mathematics/statistics).

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

Back to index


Title BEYOND BETA. Other Continuous Families of Distributions with Bounded Support and Applications.
Author S. Kotz and J.R. van Dorp.
Publisher Hackensack, New Jersey: World Scientific, 2004, pp. xvi + 289. Contents:

1. The triangular distribution
2. Some early extensions of the triangular distribution
3. The standard two-sided power distribution
4. The two-sided power distribution
5. The generalized trapezoidal distribution
6. Uneven two-sided power distributions
7. The reflected generalized Topp and Leone distribution
8. A generalized framework for two-sided distributions

Epilogue

APPENDIX A: Graphical Overview of Continuous Univariate Families of Distributions Possessing a Bounded Domain
APPENDIX B: The Johnson SB Distribution

Readership: Probabilists and statisticians

The most used family of density functions on a bounded interval is that of the beta distribution (including the uniform as a special case). This book explores alternative families of densities with bounded support. The simplest is the triangular distribution, which has various extensions: the Topp and Leone distribution, the trapezoidal distribution, the distribution of convex linear combinations of uniform random variables. Further families are the two-sided power distributions and many other generalizations, providing for example 'peaked' alternatives for the beta densities. For each of the families, properties and examples are given as well as methods for estimating the parameters. Together with the historical notes, this monograph is an interesting piece of work.

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

Back to index


Title STATISTICAL AND INDUCTIVE INFERENCE BY MINIMUM MESSAGE LENGTH.
Author C.S. Wallace.
Publisher New York: Springer-Verlag, 2005, pp. xv + 429, US$79.95. Contents:

1. Inductive inference
2. Information
3. Strict minimum message length (SMML)
4. Approximations to SMML
5. MML: Quadratic approximations to SMML
6. MML details in some interesting cases
7. Structural models
8. The feathers on the arrow of time
9. MML as a descriptive theory
10. Related work

Readership: Graduate students and researchers in machine learning and data mining, scientists and analysts in various disciplines wishing to make use of computer techniques for hypothesis discovery, statisticians and econometricians interested in the philosophy of science

I am really pleased to see this book in print at last; it has been in preparation for a long time. I recall many years ago discussing with Chris Wallace the need to have his seminal thoughts on inference, and in particular on his minimum message length approach, recorded in an accessible form. His insights and deep understanding of inferential issues deserve a wider audience, and especially deserve the attention of statisticians. The minimum message length approach to inference is based on formalizing the Occam's razor principle, that the 'best' explanation for a body of data is the shortest. The approach is not without its technical complexities, and it is possible that this has held back its wider acceptance. With luck, the appearance of this book will relax that constraint. Any statistician interested in the foundations of the discipline, or of the deeper philosophical issues of inference, will find this volume a rewarding read.
Chris Wallace died in August 2004, with the book partly finished. D.W. Albrecht and I. Zukerman, two of his colleagues, made minor corrections and amendments to bring the volume to publication.

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

Back to index


Title STRUCTURAL ASPECTS IN THE THEORY OF PROBABILITY. A PRIMER IN PROBABILITIES ON ALGEBRAIC-TOPOLOGICAL STRUCTURES.
Author H. Heyer.
Publisher River Edge, New Jersey: World Scientific, 2004, pp. x + 388, US$78.00. Contents:

1. Probability measures on metric spaces
2. The Fourier transform in a Banach space
3. The structure of infinitely divisible probability measures
4. Harmonic analysis of convolution semigroups
5. Negative definite functions and convolution semigroups
6. Probabilistic properties of convolution semigroups

APPENDIX A: Topological Groups
APPENDIX B: Topological Vector Spaces
APPENDIX C: Commutative Banach Algebras

Readership: Graduate students and researchers

This is an attractive book. The author leads the reader through a second course of probability theory and leaves her at the point of current research. The text assumes a first course in probability using measures as in the standard books, e.g. Billingsley, Breiman, Chung or Durrett. The theme of this book is to use functional analysis as the backbone on which to develop a structural theory of probability. In this respect the author assumes basic knowledge of Banach spaces. In the appendices, background material is provided without proof. In the second half of the book, the author develops the basic theory of harmonic analysis on locally compact Abelian groups. This is used in turn as a base for Markov processes on Abelian groups. The table of contents gives a good description of the material covered. There are no exercises.

Reviewer:
Institute Queen's University
Place Kingston, Canada
Name J.A. Mingo

Back to index


Title FLOWGRAPH MODELS FOR MULTISTATE TIME-TO-EVENT DATA.
Author A.K. Huzurbazar.
Publisher Hoboken, New Jersey: Wiley, 2005, pp. xii + 270, £54.50. Contents:

1. Multistate models and flowgraph models
2. Flowgraph models
3. Inversion of flowgraph moment generating functions
4. Censored data histograms
5. Bayesian prediction for flowgraph models
6. Computational implementation of flowgraph models
7. Semi-Markov processes
8. Incomplete data
9. Flowgraph models for queueing systems

Readership: Stochastic modelers especially in biological and engineering sciences, analysts of multi-state survival data

A valuable and original synthesis is given of the following ideas: discrete multistate systems in continuous time; their representation by flowgraphs; a calculus of moment generating functions (Laplace transforms) attached to such systems; the inversion of moment generating functions into density or distribution functions by saddle-point methods; provision of computer algorithms for such inversion where analytical inversion is not available; some corresponding statistical issues. The emphasis is mostly on the calculation of the distribution taken in these models, typically semi-Markov processes, to reach a specified terminal state. The account is lucid, largely self-contained and is illustrated with interesting examples from a number of fields. The book is particularly welcome in the light of an increasing tendency to tackle such problems solely by computer simulation.

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

Back to index


Title RELIABILITY AND RISK MODELS: SETTING RELIABILITY REQUIREMENTS.
Author M.T. Todinov.
Publisher Chichester, U.K.: Wiley, 2005, pp. xviii + 322, £65.00. Contents:

1. Some basic reliability concepts
2. Common reliability and risk models and their applications
3. Reliability and risk models based on mixture distributions
4. Building reliability and risk models
5. Load-strength (demand-capacity) models
6. Solving reliability and risk models using a Monte Carlo simulation
7. Analysis of the properties of inhomogeneous media using Monte Carlo simulations
8. Mechanisms of failure
9. Overstress reliability integral and damage factorisation law
10. Determining the probability of failure for components containing flaws
11. Uncertainty associated with the location of the ductile-to-brittle transition region of multi-run welds
12. Modelling the kinetics of deterioration of protective coatings due to corrosion
13. Minimising the probability of failure of automotive suspension springs by delaying the fatigue failure model
14. Reliability governed by the relative locations of random variables in a finite domain
15. Reliability dependent on the existence of minimum critical distances between the locations of random variables in a finite interval
16. Reliability analysis and setting reliability requirements based on the cost of failure

Readership: Industry (manufacturing, maintenance, engineering); academic (researchers and postgraduate students in engineering)

The treatment is firmly embedded in practical engineering. The situations are tackled intelligently, sometimes departing from the conventional wisdom and even challenging the time-honoured approaches. My overall impression is of a book written by an engineer deeply immersed in certain practical problems but able to apply sound statistical methods innovatively. This strikes me as unusual. Some technical engineering knowledge seems to be assumed, e.g. 'MPa' is used without explanation on page 53. On the other hand, no statistical knowledge is assumed, Chapters 1 to 6 covering elementary aspects of probability and statistics. The later chapters tend to focus on particular aspects, presumably reflecting the author's own interests and publications; see the list of references. These applications are dominated by the specialist field of crack propagation and metal fatigue. However, the statistical methods presented can be useful in other fields of engineering. Good features include the large number of diagrams used to illustrate the text, and the provision of pseudo-code for many of the methods.

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

Back to index


Title MEASUREMENT THEORY AND PRACTICE: THE WORLD THROUGH QUANTIFICATION.
Author D.J. Hand.
Publisher London: Arnold. 2004, pp. xi + 320, £45.00.

Contents:
1. Introduction
2. The nature of measurement
3. The process of measurement
4. Accuracy of measurement
5. Measurement in psychology
6. Measurement in medicine
7. Measurement in the physical sciences
8. Measurement in economics and the social sciences
9. Measurement in other areas

Readership: Users and teachers of quantitative research methods, statisticians

From the preface: "... in translating our observations about the world into numerical form, we are mapping from the real world to an artificial one and, in particular, one in which we can apply mathematical tools. This book is about such mappings, describing how they are carried out, the properties they must have ...". Measurement is at the root of all statistical analysis, and Hand provides an elegant, authoritative overview of both fundamental ideas (Chs. 1-4) and the design and use of specific models and instruments in major applied fields (Chs. 5-8). This is a serious work in which the author's humour shines through, and the many examples and quotations make it easily accessible to statisticians and non-statisticians alike. The book is clearly the culmination of many years' work, and it deserves to be very widely read.

Reviewer:
Institute University of Warwick
Place Coventry, U.K.
Name D. Firth

Back to index


Title GRAPHIC DISCOVERY. A Trout in the Milk and Other Visual Adventures.
Author H. Wainer.
Publisher Princeton University Press. 2005, pp. xvi +192, US$29.95/£18.95.

Contents:
PART I:
1. Why Playfair?
2. Who was Playfair?
3. William Playfair: A daring and worthless fellow
4. Scaling the heights (and widths)
5. A Priestly view of international currency exchanges
6. Tom's veggies and the American way
7. The graphical inventions of Duboug and Ferguson: Two precursors to William Playfair
PART II:
8. Winds across Europe: Francis Galton and the graphical discovery of weather patterns
9. A graphical investigation of the scourges of Vietnam
10. Two mind-bending statistical paradoxes
11. Order in the court
12. No order in the court
13. Like a trout in the milk
14. Scaling the market
15. Sex, smoking and life insurance: A graphical view
16. There they go again!
17. Sex and sport: How quickly are women gaining?
18. Clear thinking made visible: Redesigning score reports for students
PART III:
19. John Wilder Tukey: The father of twenty-first century graphical displays
20. Graphical tools for the twenty-first century: Spinning and slicing
21. Graphical tools for the twenty-first century: Nearness and smoothing engines
22. Epilogue: A selection of selections anomalies

Readership: Non-academic general interest

The author takes a personalized, and mildly quirky, tour through a series of topics all related to graphical displays in statistics. Starting in Part I, we learn about the pioneers of the use of graphical methods, in particular on William Playfair's contributions. Part II is a series of short essays on topics where graphical methods are shown to be powerful in discovering patterns in data. Part III concentrates on the influence of Tukey on interactive computer graphics such as spanning, brushing and smoothing.
The tour that the reader is taken on is rather idiosyncratic, arriving at some unusual places and bypassing other areas where a more thorough treatment would certainly stop. For example, a student, interested in learning about graphical methods, who flips through the index would be surprised to find 'Henry VllI' but not 'histogram', 'bourgeois' but not 'box plot'.

Reviewer:
Institute University of Waterloo
Place Waterloo, Canada
Name P. Marriott

Back to index


Title EXPLORATORY DATA ANALYSIS WITH MATLAB.
Author W. Martinez and A.R. Martinez.
Publisher Boca Raton, Florida: Chapman and Hall/CRC Press. 2005, pp. xv + 405, US$79.95/£44.99.

Contents:
1. Introduction to exploratory data analysis
2. Dimensionality reduction - linear methods
3. Dimensionality reduction - nonlinear methods
4. Data tours
5. Finding clusters
6. Model-based clustering
7. Smoothing scatter plots
8. Visualizing clusters
9. Distribution shapes
10. Multivariate visualization

APPENDIX A : Proximity Measures
APPENDIX B: Software Resources for EDA
APPENDIX C: Description of Data Sets
APPENDIX D: Introduction to MATLAB
APPENDIX E: MATLAB Functions

Readership: Scientists, statisticians, data miners, engineers, computer scientists, biostatisticians, social scientists and any discipline that deals with the analysis of raw data

This book, as the title implies, is centred on the use of MATLAB® to illustrate the computational aspects of Exploratory Data Analysis (EDA). MATLAB® and the MATLAB® Statistics Toolbox are used extensively to demonstrate the practical applications of EDA. The theory of EDA is alluded to but this is not the main thrust of the text. Pseudo code is also present so that users of other software packages can implement the aIgorithms. The main goal of the book is to show the key concepts and methods of computational statistics and how they can be implemented in MATLAB®.
This book presumes that the reader has an understanding of basic linear algebra and has completed an introductory course in probability and statistics. In particular the reader should know about random variables, probability distributions, density functions, regression and basic descriptive measures on the statistics front, and also about array multiplication, matrix inverse, determinants and array transfers on the algebra side.
MATLAB code in the form of EDA Toolbox is provided with the text. This includes the functions, GUIs and sets of data that are described in the book. This is available for download at and
.
The author recommends you review the 'readme' file for installation instructions and information on any changes. M-files that contain the MATLAB commands for the exercises are also available for download. I would recommend that these be downloaded for use when required.
Each chapter has exercises for the reader to complete, however, there no answers are supplied. A brief summary and further reading is included with each chapter. The text also contains a wealth of references for the reader to pursue on related issues.
This is a book for those who have a good grounding in linear algebra and statistics who wish to use MATLAB® for statistical investigations.

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

Back to index


Title FINANCIAL DERIVATIVES, 2nd edition.
Author R.W. Kolb.
Publisher Malden, Massachusetts: Blackwell, 2000, pp. ix + 261, £25.99. [Original, 1996] Contents:

1. Introduction
2. Futures
3. Options
4. The swaps market
5. Synthesizing and transforming securities
6. Risk management and financial engineering

Readership: Students interested in an overview of
financial derivates that avoids
the use of mathematics

The book succeeds in providing supplemental material for an introductory course that treats futures, options and swaps. The material is well-presented and the index is good. New concepts are highlighted in bold. A sixth chapter is added to this second edition promoting derivates as instruments for managing risk. The deficiencies of this new edition show up in precisely this context-the tables are not updated from 1996, the case studies (e.g., Barings, Orange County, P&G) are not completed and, perhaps most importantly, there is no mention of the rapidly growing interest in managing credit risk through CDO's, CDS's, etc.

Reviewer:
Institute Imperial College of Science, Technology and Medicine
Place London, U.K.
Name C. Albanese

Back to index


Title INTRODUCTION TO MODERN PORTFOLIO OPTIMIZATION WITH NuOPTtm, S-PLUSâ AND S+Bayestm.
Author B. Scherer and R.D. Martin.
Publisher New York: Springer-Verlag, 2005, pp. xxi + 405, US$79.95. Contents:

1. Linear and quadratic programming
2. General optimization with SIMPLE
3. Advanced issues in mean-variance optimization
4. Resampling and portfolio choice
5. Scenario optimization: Addressing non-normality
6. Robust statistical methods for portfolio construction
7. Bayes methods

Readership: Practitioners in quantitative finance, students and scientists in financial engineering, operations research or statistics

The construction of optimal investment strategies is a basic and hot topic of financial engineering in both practice and academia. In this book variants and generalizations of the classical one-period Markowitz mean-variance portfolio optimization approach are considered. Topics addressed range from introductions to linear and quadratic programming, robust statistics, and Bayesian estimation to risk measures, such as conditional value-at-risk or mean absolute deviation that constitute alternatives to the portfolio return variance. These introductions are accompanied by practical advice on how to use statistical packages such as NuOPTtm, S-PLUSâ and S+Bayestm to solve the corresponding static portfolio problems. With regard to static portfolio optimization, the book gives a good survey on the development from the basic Markowitz approach to state of the art models and is in particular valuable for direct use in practice or for lectures combined with practical exercises. However, it does not contain any significant part on modern multi-period or continuous time portfolio optimization at all.

Reviewer:
Institute University of Kaiserslautern
Place Kaiserslautern, Germany
Name R. Korn

Back to index


Title ROBUST LIBOR MODELLING AND PRICING OF DERIVATIVE PRODUCTS.
Author J. Schoenmakers.
Publisher Boca Raton, Florida: Chapman and Hall/CRC Press, 2005, pp. xvi + 202, US$89.95/£99.95. Contents:

1. Arbitrage-free modelling of effective interest rates
2. Parametrisation of the Libor market model
3. Implied calibration of a Libor market model to caps and swaptions
4. Pricing of exotic European style products
5. Pricing of Bermudan style Libor derivatives
6. Pricing long dated products via Libor approximations

Readership: Practitioners, financial engineers, students with an interest in (Libor Market Model) interest-rate modelling

This book provides an introduction to the Libor market model, one of the current tools for modelling interest rates and interest rate derivatives. The author begins with the necessary financial mathematics, discusses parametrizations of the Libor model that are economically feasible, and calibration of the Libor Market Model to market data. The latter exercise is made more challenging by the fact that typically the model has a parameter of higher dimension than the cardinality of the market quotes that are being matched. The impact of various calibration methods on the pricing of various interest-rate derivatives, Libor trigger swaps, Ratchet caps, sticky Caps, and Auto-flex caps, as well as Bermudan style Libor derivatives is discussed. Finally, lognormal approximations are used to price long-dated interest-rate derivatives and these approximations assessed by simulation.

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

Back to index


Title FINANCIAL DERIVATIVES: PRICING, APPLICATIONS AND MATHEMATICS.
Author J. Baz and G. Chacko.
Publisher Cambridge University Press, 2004, pp. xi + 338, £35.00. Contents:

1. Preliminary mathematics (17pp)
2. Principles of financial valuation (56pp)
3. Interest rate models (106pp)
4. Mathematics of asset pricing (85pp)

Bibliography (57pp)

Readership: MBA and MSc students in finance, Quant analysts in the finance industry

There is now a huge number of books on this subject. This one is at an intermediate level: no measure theory but much more mathematical and practical detail than elementary texts. The emphasis can be gleaned from the page count of the chapters. Chapter 2 covers the Black-Scholes theory, but with more emphasis than usual on principles of risk, utility and equilibrium. The lengthy interest rate chapter includes valuable detail on convexity corrections etc. and covers some standard models, but does not cover important modern developments such as the Libor Market Model. The final chapter covers standard material in stochastic calculus at a generally informal level. Occasionally the authors slip up-for example, their proof that Brownian motion is continuous also shows that the Poisson process is continuous (which it is, in probability). There is a colossal bibliography.

Reviewer:
Institute Imperial College of Science, Technology and Medicine
Place London, U.K.
Name M.H.A. Davis

Back to index


Title STATISTICAL ANALYSIS AND DATA DISPLAY: AN INTERMEDIATE COURSE WITH EXAMPLES IN S-PLUS, R, AND SAS.
Author R.M. Heiberger and B. Holland.
Publisher New York: Springer-Verlag. 2004, pp. xxiv + 729 , US$79.95.

Contents:
1. Introduction and motivation
2. Data and statistics
3. Statistics concepts
4. Graphs
5. Introductory inference
6. One-way analysis of variance, ANOVA
7. Multiple comparisons
8. Linear regression by least squares
9. Multiple regression - more than one predictor
10. Multiple regression - variables and contrasts
11. Multiple regression - regression diagnostics
12. Two-way analysis of variance
13. Design of experiments - factorial designs
14. Design of experiments - complex designs
15. Bivariate statistics - discrete data
16. Nonparametrics
17. Logistic regression
18. Time series analysis

Readership: Statistics majors at the master's level, other quantitatively oriented disciplines at the Ph.D. level

As indicated by the subtitle, this is an intermediate-level text covering the use of three popular examples of statistical software applied to the display and statistical analysis of data. The range of statistical techniques is broad but not as comprehensive as, say, Modern Applied Statistics with S (4th edition) by Venables and Ripley [2nd edition, Short Book Reviews, Vol. 15, p. 7]. This book does cover the use of SAS which "MASS" does not. I think the authors are accurate in targeting the master's level student with this book as opposed to the Ph.D. student for whom MASS may be more appropriate. Considering that the book emphasizes data display and interpretation, the quality of the graphics is sometimes disappointing.

Reviewer:
Institute University of Wisconsin
Place Madison, U.S.A.
Name D.V. Bates

Back to index


Title LINEAR MODELS WITH R.
Author J.J. Faraway.
Publisher Boca Raton, Florida: Chapman and Hall/CRC Press. 2004, pp. vii + 229, US$69.95/£39.99.

Contents:
1. Introduction
2. Estimation
3. Inference
4. Diagnostics
5. Problems with predictions
6. Problems with error
7. Transformations
8. Variable selection
9. Shrinkage methods
10. StatisticaI strategy and model uncertainty
11. Insurance redlining - A complete example
12. Missing data
13. Analysis of variance
14. One-way analysis of variance
15. Factorial designs
16. Block designs

APPENDIX A: R Installation, Functions and Data
APPENDIX B: Quick Introduction to R

Readership: Researchers, teachers and students who have a background in statistics

There are many books on regression and analysis of variance on the market, but this one is unique and has a novel approach to these statistical methods. The author uses R throughout the text to teach data analysis. The user will have to have R installed on their machine to be able to do the exercises which are provided at the end of the chapters. The emphasis is on the practice of regression and analysis of variance with considerably less emphasis on the mathematical techniques.
This book is not an introductory text. It presumes some knowledge of basic statistical theory and practice. Readers are expected to know the essentiaIs of statistical inference such as estimation, hypothesis testing and confidence intervals. Also a basic knowledge of data analysis is presumed and some linear algebra and calculus is required. A brief introduction to R is presented in the appendix, and all the sets of data used are available from the website www.stat.lsa.umich.edu/~faraway/LMR .
One will need to follow the explanation given to be able to download and use the data. Free introductory guides to R can be obtained from www.r-project.org which is extremely useful to new users of R.
The text also contains a wealth of references for the reader to pursue on related issues. This book is recommended, for all who wish to use R for statistical investigations.

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

Back to index


Title MONTE CARLO STATISTICAL METHODS, 2nd edition.
Author C.P. Robert and G. Casella.
Publisher New York: Springer-Verlag. 2004, pp. xxx + 645, US$89.95.

Contents:
1. Introduction
2. Random variable generation
3. Monte Carlo integration
4. Controlling Monte Carlo variance
5. Monte Carlo optimization
6. Markov chains
7. The Metropolis-Hastings algorithm
8. The slice sampler
9. The two-stage Gibbs sampler
10. The multi-stage Gibbs sampler
11. Variable dimension models and reversible jump algorithms
12. Diagnosing convergence
13. Perfect sampling
14. Iterated and sequential importance sampling

Readership: Graduate students and scientists seeking to understand or apply Monte Carlo methods

This revision of the influential 1999 text [Short Book Reviews, Vol. 20, p. 12] includes changes to the presentation in the early chapters and much new material related to MCMC and Gibbs sampling. The result is a useful introduction to Monte Carlo methods and a convenient reference for much of current methodology. The theoretical bases of the methods are presented in useful detail. For example the chapter introducing the Metropolis-Hastings algorithm runs for fifty-four pages. The numerous problems include many with analytical components. The result is a very useful resource for anyone wanting to understand Monte Carlo procedures.
This excellent text is highly recommended, with one caution. The many examples that illustrate applications are conveniently simple. However, some readers may miss the fact that many examples are appropriately handled with more basic methods.

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

Back to index


Title PLANNING, CONSTRUCTION AND STATISTICAL ANALYSIS OF COMPARATIVE EXPERIMENTS.
Author F.G. Giesbrecht and M.L. Gumpertz.
Publisher Hoboken, New Jersey: Wiley. 2004, pp. xi + 693, £67.95.

Contents:
1. Introduction
2. Completely randomised designs
3. Linear models for designed experiments
4. Testing hypotheses and determining sample sizes
5. Methods of reducing unexplained variation
6. Latin squares
7. Split-plot and related designs
8. Incomplete block designs
9. Repeated treatments designs
10. Factorial experiments: The 2N system
11. Factorial experiments: The 3N system
12. Analysis of experiments without designed error terms
13. Confounding effects with blocks
14. Fractional factorial experiments
15. Response surface designs
16. Plackett- Burman Hadamard plans
17. General pN and nonstandard factorials
18. Plans for which run order is important
19. Sequences of fractions of factorials
20. Factorial experiments with quantitative factors: Blocking and fractions
21. Supersaturated plans
22. Multistage experiments
23. Orthogonal arrays and related structures
24. Factorial plans derived via orthogonal arrays
25. Experiments on the computer

Readership: Experimental scientists, statistics undergraduate/MSc students

This book covers a wide range of topics. The approach is broadly a traditional one based on the normal theory linear model and analyses of all the standard experimental designs are presented, including for example analysis of covariance in split-plot experiments. The level of detail is higher than in most other books on similar topics and therefore makes this one a useful reference tool. There are lots of numerical examples and exercises with calculations based on SAS®. This is definitely a book aimed at experimentalists; the mathematical level is not high. However, there are good accounts of confounding and fractional replication, topics which many find conceptually difficult.
We find general advice about planning experiments including a section on ethical issues and a section on Taguchi methods. Little mention is made of optimality criteria but this does not detract from the value of the book.

Reviewer:
Institute Imperial College of Science, Technology and Medicine
Place London, U.K.
Name L.V. White

Back to index


Title ANALYSIS OF VARIANCE FOR RANDOM MODELS, Volume II: Unbalanced Data. Theory, Methods, Applications and Data Analysis.
Author H. Sahai and M.M. Ojeda.
Publisher Boston: Birkhäuser. 2005, pp. xxv + 480, US$89.95.

Contents:
9. Matrix preliminaries and general linear model
10. Some general methods for making inferences about variance components
11. One-way classification
12. Two-way classification without interaction
13. Two-way classification with interaction
14. Three-way and higher-order crossed classification
15. Two-way nested classification
16. Three-way nested classification
17. General r-way nested classification

Readership: Theoretical research statisticians, applied statisticians using random effects models, graduate students

This is the second of a two-volume comprehensive review of the analysis of variance for random effects models. Volume I [Short Book Reviews, Vol. 24, p. 46] was devoted to various models using balanced data, whereas this volume is concerned with unbalanced data. The book provides an extensive coverage of the methods and techniques of point estimation, interval estimation and tests of hypotheses for random effects models. A variety of experimental designs are considered involving one, two, three and multi-factor experiments with crossed and nested factors. Throughout the text, the procedures are illustrated using numerical examples analyzed with several software systems (although in my copy and at least one other copy Figures 15.1, 16.1, 16.2 and 16.3 which should have contained computer output were blank).
This volume begins with Chapter 9, a short chapter on matrix algebra and the linear model. This is followed by a substantial chapter in which several methods of estimating the variance components in the random effects model are discussed. These include Henderson's adaptations of ANOVA, restricted maximum likelihood (REML), minimum-norm quadratic unbiased estimation (MINQUE) and Pukelshiem's convex programming approach using the existence of non-negative quadratic unbiased estimators. The authors include an examination of the merits of the different methods described, remarking on the relative simplicity of some and the computational complexity of others. The chapter closes with a set of twenty-four exercises and fifteen pages of references.
Chapter 11 on the one-way classification model sets the structure for the remaining chapters of the book. The various estimation methods described earlier are applied to this simple model in considerable detail. Sampling variances and confidence intervals for the estimators of the variance components are developed and the methods are illustrated using a numerical example analyzed with SAS, SPSS and BMDP. Subsequent chapters, which cover increasingly more complicated models with more factors, with and without interaction terms and with the factors crossed or nested, have a similar structure, each covering the detailed theory, a numerical example, a set of exercises and a list of references.
The book will be particularly useful as a reference source to the literature of random effects models with each chapter having its own bibliography and with the main extensive reference section containing more than six hundred papers.

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

Back to index


Title DESIGN AND ANALYSIS OF ACCELERATED TESTS FOR MISSION CRITICAL RELIABILITY.
Author M.J. LuValle, B.G. Lefevre and S. Kannan.
Publisher Boca Raton, Florida: Chapman and Hall/CRC Press. 2004, pp. ix + 236, US$99.95/£60.99.

Contents:
1. Background
2. Demarcation mapping: Initial design of accelerated tests
3. Interface for building kinetic models
4. Evanescent process mapping
5. Data analysis for failure time data
6. Data analysis for degradation data

APPENDIX: Installing the Software

Readership: Reliability engineers, physical scientists, statisticians

It is widely recognized that advances in the analysis and improvement of reliability ultimately rely on an understanding of the physical processes that lead to failures in equipment, materials or systems. This is especially true in the area of accelerated testing, where reliability testing under severe conditions aims to provide insight into reliability under normal conditions. This book attempts to link physical models for degradation and failure to accelerated testing and reliability analysis. The scope is fairly narrow: the book focuses on first-order chemical kinetic models and introduces a limited range of statistical methods. Nevertheless, it is a useful and welcome start in an important area. The inclusion of a software system that aids specification and visualization of kinetic models is also welcome. One hopes that this book will spur further research in this area.

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

Back to index


Title ANALYSIS AND MODELLING OF SPATIAL ENVIRONMENTAL DATA.
Author M. Kanevski and M. Maignan.
Publisher New York: EPFL Press. 2004, pp. xi + 288 + CD, US$175.00/£100.00.

Contents:
1. Introduction to environmental data analysis and modelling
2. Exploratory spatial data analysis. Analysis of monitoring networks, declustering
3. Spatial data analysis: Deterministic interpolations
4. Introduction to geostatistics. Variography
5. Geostatistical spatial predictions
6. Estimation of local probability density functions
7. Conditional stochastic simulations
8. Artificial neural networks and spatial data analysis
9. Support vector machines for environmental spatial data
10. Geographical information systems and spatial data analysis
11. Conclusions

Readership: Undergraduate statistics and/or environmental science majors

This is an odd book, and it is not clear to me what its purpose is. It comes with a "student edition" CD of "Geostat Office" (which only runs on Windows) so presumably is meant either as an elementary level text or possibly as a program manual. The preface states that "The objective of this book was to write a tutorial text with accompanying software tools for the analysis, processing and presentation of spatially distributed data." The first thing I looked for, the index, does not exist. I found the book very uneven. The only integral sign I found was in the definition of the cumulative distribution function [sic]
, unnecessarily careless for a text. Statistical definitions range from sloppy to incorrect, expectation is used without being defined, and, as in (5.6), appears to be considered equivalent to an estimate. Similarly, in discussing kriging, they refer to "stationarity" when the context implies "isotropic". Explaining "BLUE(P)" on page 90, the authors helpfully say "Linear means linear model." The editing is careless with numerous errors in English ("jackknife" is written "jack-knife"), mistakes in formulae, incomplete sentences etc. Specialized jargon is fairly dense and often not well defined: page 149 discusses "Spectral methods" with the kind of spectrum - optical, time-series, matrix eigenvalue, or energy per nucleon - left to the imagination. The graphics are generally colourful, but often lack informative axis labels. Many of the labels and numbers are in ~ 2.5 point type. In summary, I think I would be quite uncomfortable living close to a superfund site which had been certified safe by someone who had been taught how to make rigorous statistical arguments about environmental data from this book.

Reviewer:
Institute Queen's University
Place Kingston, Canada
Name D.J. Thomson

Back to index


Title HEIRARCHICAL MODELLING AND ANALYSIS FOR SPATIAL DATA.
Author S. Banerjee, B.P. Carlin and A.E. Gelfand.
Publisher Boca Raton, Florida: Chapman and Hall/CRC Press. 2004, pp. xvii + 452, US$69.95/£46.99.

Contents:
1. Overview of spatial data problems
2. Basics of point-referenced data models
3. Basics of areal data models
4. Basics of Bayesian inference
5. Heirarchical modelling for univariate spatial data
6. Spatial misalignment
7. Multivariate spatial modelling
8. Spatiotemporal modelling
9. Spatial survival models
10. Special topics in spatial process modelling

APPENDIX: Matrix Theory and Spatial Computing Methods

Readership: Graduate students and those working in the field

This book was a pleasure to review. Most of the emphasis is on insight and intuition with relatively little on traditional muItivariate techniques. I also found some of the explanations delightful. When was the last time you encountered the Cauchy-Riemann equations in a section on map projections in a statistics text? The authors state that "Our purpose in writing this book is to describe heirarchical Bayesian methods for one class of applications for which they can pay substantial dividends: spatial (and spatiotemporal) statistics." And, while they did not quite convert me to Bayesianism, made me reconsider some of my assumptions. They later state "Our book is intended as a research monograph, presenting the state of the art" and my impression is that they have succeeded.
The text is intentionally Iess mathematically demanding than N.A.C. Cressie [1993, Short Book Reviews, Vol. 12, p. 45] but at the same time it is not trivial. In many sections the formulae are augmented by showing R or S code, making it easy to actually apply the mathematics.
In summary, this is a nice book.

Reviewer:
Institute Queen's University
Place Kingston, Canada
Name D.J. Thomson

Back to index


Title ADVANCED DISTANCE SAMPLING.
Author S.T. Buckland, D.R. Anderson, K.P. Burnham, J.L. Laake, D.L. Borchers and L. Thomas.
Publisher Oxford University Press. 2004, pp. xvii + 416.

Contents:
1. Introduction to advanced distance sampling
2. General formulation for distance sampling
3. Covariate models for the detection function
4. Spatial distance sampling models
5. Temporal inferences from distance sampling surveys
6. Methods for incomplete detection at distance zero
7. Design of distance sampling surveys and Geographic Information Systems
8. Adaptive distance sampling surveys
9. Passive approaches to detection in distance sampling
10. Assessment of distance sampling estimators
11. Further topics in distance sampling

Readership: Professionals in government and environmental agencies, statisticians, biologists, wildlife managers, conservation biologists and ecologists, as well as graduate students of the density and abundance of biological populations

This is a follow up to the same team's Introduction to Distance Sampling [2002, Short Book Reviews, Vol. 22, p. 23] this time concentrating on more advanced and recent methods. Those methods used prior to 2000 are deemed historical by the editorial team. The chapter headings reveal a wide range of application of modern statistical techniques. The contributors were assigned specific topics rather than their own research, with their chapters editorially treated so that it reads like a unified applied statistics textbook. It may also be read as a thorough review of current best practice in distance sampling. A computer package that implemented the methods of the previous volume is being extended to cover those of this book.

Reviewer:
Institute Imperial College of Science, Technology and Medicine
Place London, U.K.
Name R. Coleman

Back to index


Title EXPLANATORY ITEM RESPONSE MODELS: A GENERALIZED LINEAR AND NONLINEAR APPROACH.
Author P. DeBoeck and M. Wilson (Eds.).
Publisher New York: Springer-Verlag. 2004, pp. xxii + 382, US$69.95.

Contents:
PART I: Introduction to the Framework
1. A framework for item response models
2. Descriptive and explanatory item response models
3. Models for polytomous data
4. An introduction to (generalised) (non)linear mixed models
PART II: Models with External Factors - Predictors and their Effects
5. Person regression models
6. Models with item and item group predictors
7. Person-by-item predictors
PART III: Models with Internal Factors
8. Multiple person dimensions and latent item predictors
9. Latent item predictors with fixed effects
10. Models for residual dependencies
11. Mixture models
PART IV:
12. Estimation and software
Afterword

Readership: Those wishing an advanced introduction to item response models, psychometricians, practitioners who (hope to) apply such tools. Advanced graduate students in measurement and psychometrics

This book seeks to generalize the typical perspective on item response models, putting the ideas into a broader statistical context. It begins with the assertion that statistical analysis (for models of this type) has typically been aimed at explanation, with the objective being to elucidate how 'independent' variables influence 'dependent' variables, while measurement has been aimed at constructing or deriving latent variables which measure some aspect of individuals (and items). With this as the background, the book argues that a common core of statistical models can be used for either objective, these being generalized linear mixed models and nonlinear mixed models. There are various advantages to this broader perspective, not least the fact that general purpose software can be used.
The authors have made a good job of integrating the various contributions, and I believe the book will help in increasing awareness of the potential of these types of models.

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

Back to index


Title TEST EQUATING, SCALING, AND LINKING, 2nd edition.
Author M.J. Kolen and R.L. Brennan.
Publisher New York: Springer-Verlag. 2004, pp. xxvi + 548, US$79.95.

[Original 1995, Short Book Reviews, Vol. 16, p. 6]

Contents:
1. Introduction and concepts
2. Observed score equating using the random groups design
3. Random groups - smoothing in equipercentile equating
4. Nonequivalent groups - linear methods
5. Nonequivalent groups - equipercentile methods
6. Item response theory methods
7. Standard methods of equating
8. Practical issues in equating
9. Score scales
10. Linking
11. Current and future challenges

Readership: Advanced graduate psychometric students, entry-Ievel professionals, or others approaching equating, scaling, or linking for the first time, or professionals using the book as a reference work

Equating is a statistical tool which adjusts for differences between tests which are intended to be similar in difficulty and content. This book describes such tools. The authors say 'the principal goals of this book are for the reader to understand the principles of equating, scaling, and linking; to be able to conduct equating, scaling, and linking, and to interpret the results in reasonable ways.' This second edition, published nine years after the (award-winning) first edition, has new chapters on test scaling and on test linking.
Given the perennial debates about academic standards and grade inflation, it is my view that tools such as those described in this book should be adopted much more widely by the academic community than they are at present. This book provides an excellent overview, and I strongly recommend it.

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

Back to index


Title MULTIVARIATE STATISTICAL METHODS, 4th edition.
Author D.F. Morrison.
Publisher Belmont, California, Brooks/Cole Thomson Learning. 2005, pp. viii + 469.

Contents:
1. Samples from the multivariate normal population
2. Tests of hypotheses on means
3. The multivariate analysis of variance
4. Classification by discriminant functions
5. lnferences from covariance matrices
6. The structure of multivariate observations: I. Principal components
7. The structure of multivariate observations: II. Factor analysis

Readership: Students and practitioners of statistics, and researchers in other fields who analyze multidimensional response data

This is the fourth edition of a classic introductory text on multivariate statistics that was first published in 1967. This edition differs from prior editions in that two introductory chapters on elementary statistics and matrix algebra, respectively, have been removed, since such basic information is readily available elsewhere. Citations have also been updated, and a number of real sets of data for exercises have been added.
Coverage is quite traditional, and, though elementary, readers do need some prior statistical maturity and facility with matrix algebra before tackling the book. The book is a good standard reference for any statistician who works with multivariate data.

Reviewer:
Institute Brookfield,
Place U.S.A.
Name C.A. Fung

Back to index


Title APPLIED ECONOMETRIC TIME SERIES, 2nd edition.
Author W. Enders.
Publisher Hoboken, New Jersey: Wiley. 2004, pp. xiv + 460.

Contents:
1. Difference equations
2. Stationary time-series models
3. Modeling volatility
4. ModeIs with trend
5. Multi-equation time-series models
6. Co-integration and error-correction models
7. Nonlinear time-series models

Readership: Undergraduate students of statistics, macroeconomics, agricultural economics and international finance

This book is the second edition of a previously well-read text used by both students of statistics and economics. In revising the text, the author has addressed a number of the issues raised by readers of the first edition and incorporated guidance on how to compare the forecasts of alternative time-series models. The largest change has been in the addition of an entire chapter on nonlinear time-series models.
The text emphasizes the difference equation as the foundation of time-series modelling and the author takes seriously the term applied in the title of the book. Simple exampIes are used and built upon to develop ever more general and complicated models. The author recognizes the need for the techniques to be explicitly programmed and reference is given to a number of software packages.
The statistical prerequisites for using this book are modest: a knowledge of multiple regression analysis and ordinary least squares. The reader is assumed to be familiar with the concepts correlation and covariance, together with the meaning of terms such as mean square error, significance level and unbiased estimate. A knowledge of matrix algebra is assumed as a prerequisite to solving systems of equations.

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

Back to index


Title SKEW-ELLIPTICAL DISTRIBUTIONS AND THEIR APPLICATIONS. A Journey Beyond Normality.
Author M.G. Genton (Ed.).
Publisher Boca Raton, Florida: Chapman and Hall/CRC Press. 2004, pp. xvii + 396, US$89.95/£49.99.

Contents:
PART I: Theory and Inference
1. The skew-normal distribution, by A.D. Valle
2. The closed skew-normal distribution, by G. González-Farías, J.A. Domínguez-Molina, and A. Gupta
3. Skew-elliptical distributions, by J. Liu and D.K. Dey
4. Generalized skew-normal distributions, by N.M.R. Loperfido
5. Skew-symmetric and generalized skew-elliptical distributions, by M.G. Genton
6. Elliptical models subject to hidden truncation or selective sampling, by B.C. Arnold and R.J. Beaver
7. From symmetric to asymmetric distributions: A unified approach, by R.B. Arellano-Valle and G.E. del Pino
8. Skewed link models for categorical response data, by M.-H. Chen
9. Skew-elliptical distributions in Bayesian inference, by B. Liseo
PART II: Applications and Case Studies
10. Bayesian multivariate skewed regression modeling with an application to firm size, by J.T.A.S. Ferreira, and M.F.J. Steel
11. Capital asset pricing for UK stocks under the multivariate skew-normal distribution, by C. Adcock
12. A skew-in-mean GARCH model, by G. De Luca and N.M.R. Loperfido
13. Skew-normality in stochastic frontier analysis, by J.A. Domínguez-Molina, G. González-Farías and R. Ramos-Quiroga
14. Coastal flooding and the multivariate skew-t distribution, by K.R. Thompson and Y. Shen
15. Time series analysis with a skewed Kalman filter, by P. Naveau, M.G. Genton and C. Ammann
16. Spatial prediction of rainfall using skew-normal processes, by H.M. Kim, E. Ha and B.K. Mallick
17. Shape representation with flexible skew-symmetric distributions, by S.H. Baloch, H. Krim and M.G. Genton
18. An astronomical distance determination method using regression with skew-normal errors, by L. Eyer and M.G. Genton
19. On a Bayesian multivariate survival model with a skewed frailty, by S.K. Sahu and D.K. Dey
20. Linear mixed effects models with flexible generalized skew-elliptical random effects, by Y.Y. Ma, M.G. Genton and M. Davidian

Readership: Researchers and graduate students in statistics, scientists

Suppose f(x) is a density function with f(x) = f(-x) for all x. Let 0 ? ?(x) ? 1 be a skewing function with the defining property that ?(-x) = 1 - ?(x). Then for ? > 0, the product f(x;?) = 2 f(x) ?(? x) is easily seen to be a density function, with its skewness determined by the value of ?.
For example, the symmetric case is ? = 0. When f(x) is a standard normal density function and ?(x) is the standard normal distribution function, the resulting density is said to be skewed normal. Extensions to multivariate skewed normal distributions are not hard to construct, and have interesting theoretical properties and applications.
This book brings together theoretical and applied work on skewed multivariate distributions. There is an impressive list of topics and authors. The book does a good job of advertising these skewed distributions as flexible families of error distributions for data where symmetry is not appropriate. As Nicola Loperfido says in Chapter 4, these skewed distributions are "a reasonable compromise between mathematical tractability and shape flexibility". I agree, provided that the emphasis is put on mathematical tractability. Flexible families are not hard to construct. So the mathematically tractable choices are to be recommended the most.

Reviewer:
Institute University of Waterloo
Place Waterloo, Canada
Name C.G. Small

Back to index


Back to ISI Home Page

Back to SBR Index