Short Book Reviews On-Line 1999

AN INTRODUCTION TO COPULAS.	R.B. Nelson.
IMAGE PROCESSING AND DATA ANALYSIS: The Multiscale Approach.	J.-L. Starch, F. Murtagh and A. Bijaoui.
STOCHASTIC DIFFERENTIAL EQUATIONS AND APPLICATIONS.	X. Mao.
ERGODICITY AND STABILITY OF STOCHASTIC PROCESSES.	A.A. Borovkov.
RECORDS.	B.C. Arnold, N. Balakrishnan and H.N. Nagaraja.
MULTIVARIATE QUALITY CONTROL. THEORY AND APPLICATIONS.	C. Fuchs and R.S. Kenett.
LINEAR SEMI-INFINITE OPTIMIZATION.	M.A. Goberna and M.A. López.
R&D AND PRODUCTIVITY. The Econometric Evidence.	Z. Griliches.
VISUAL EXPLANATIONS. Images and Quantities, Evidence and Narrative.	E.R. Tufte.
FITTING LINEAR RELATIONSHIPS: A History of the Calculus of Observations, 1750–1900.	R.W. Farebrother.
ECONOMICS: THE CULTURE OF A CONTROVERSIAL SCIENCE.	M.W. Reder.
STATISTICS FOR THE ENVIRONMENT 4: Statistical Aspects of Health and the Environment.	V. Barnett, A. Stein and K. Turkman (Eds.).
NUMERICAL LINEAR ALGEBRA FOR APPLICATIONS IN STATISTICS.	J.E. Gentle.
STATISTICAL DATA ANALYSIS.	G. Cowan.
MODELS FOR DISCRETE DATA.	D. Zelterman.
INTRODUCTORY STATISTICS AND RANDOM PHENOMENA. Uncertainty, Complexity, and Chaotic Behaviour in Engineering and Science.	M. Denker and W.A. Woyczynski.
NONPARAMETRIC STATISTICAL METHODS, 2nd edition.	M. Hollander and D.A. Wolfe.
LONGITUDINAL DATA ANALYSIS: DESIGNS, MODELS AND METHODS.	C.J. Bijleveld and L.J. van der Kamp with C.C.J. Bijleveld, W.A. van der Kloot, R. van der Leeden and F. van der Berg.
APPLIED SURVIVAL ANALYSIS. REGRESSION MODELING OF TIME TO EVENT DATA.	D.W. Hosmer Jr. and S. Lemeshow.
THE USES AND MISUSES OF DATA AND MODELS. THE MATHEMATIZATION OF THE HUMAN SCIENCES.	W.J. Bradley and K.C. Schaefer.
LA REGRESSION PLS.	M. Tenenhaus.
REGRESSION GRAPHICS. Ideas for Studying Regressions through Graphics.	R.D. Cook.
PROCESS CAPABILITY INDICES IN THEORY AND PRACTICE.	S. Kotz and C.R. Lovelace.
MODEL SELECTION AND INFERENCE: A PRACTICAL INFORMATION-THEORETIC APPROACH.	K.P. Burnham and D.R. Anderson.
STATISTICAL DESIGN AND ANALYSIS FOR INTERCROPPING EXPERIMENTS. Volume II: Three or More Crops.	W.T. Federer.
THEORY AND METHODS OF SURVEY SAMPLING.	P. Mukhopadhyay.
ELEMENTS OF LARGE-SAMPLE THEORY.	E.L. Lehmann.
ASYMPTOTIC STATISTICS.	A.W. van der Vaart.
DECOUPLING: FROM DEPENDENCE TO INDEPENDENCE. Randomly Stopped Processes, U-Statistics and Processes, Martingales and Beyond.	V.H. de la Peña and E. Giné.
EPIDEMIC MODELLING: AN INTRODUCTION.	D.J. Daley and J. Gani.
SAMPLE-PATH ANALYSIS OF QUEUEING SYSTEMS.	M. El-Taha and S. Stidham Jr.
LARGE DEVIATIONS TECHNIQUES AND APPLICATIONS, 2nd edition.	A. Dembo and O. Zeitouni.
STATISTIQUE THEORIQUE ET APPLIQUEE: Tome 2, Inférence statistique à une et à deux dimensions.	P. Dagnelie.
MISUSED STATISTICS, 2nd edition.	H.F. Spirer, L. Spirer and A.J. Jaffe.
STATISTICAL CASE STUDIES. A Collaboration between Academe and Industry.	R. Peck, L.D. Haugh and A. Goodman.
FORECASTING ECONOMETRIC TIME SERIES.	M.P. Clements and D.F. Hendry.
TIME SERIES MODELS FOR BUSINESS AND ECONOMIC FORECASTING.	P.H. Franses.
STOCHASTIC PROCESSES FOR INSURANCE AND FINANCE.	T. Rolski, H. Schmidt, V. Schmidt and J.F. Teugels.
MATHEMATICS OF FINANCIAL MARKETS.	R.J. Elliott and P.K. Kopp.
APPLIED STOCHASTIC MODELS AND CONTROL FOR FINANCE AND INSURANCE.	C.S. Tapiero.
ARBITRAGE THEORY IN CONTINUOUS TIME.	T. Björk.
STOCHASTIC DYNAMIC PROGRAMMING AND THE CONTROL OF QUEUEING SYSTEMS.	L.I. Sennott.
CASE STUDIES IN ENVIRONMENTAL STATISTICS.	D. Nychka, W.W. Piegorsch and L.H. Cox (Eds.).
KENDALL's ADVANCED THEORY OF STATISTICS. Volume 2A. Classical Inference and the Linear Model, 6th edition	A. Stuart, J.K. Ord and S. Arnold,
COMPARATIVE STATISTICAL INFERENCE, 3rd edition.	V. Barnett.
REVEALING STATISTICAL PRINCIPLES.	J.K. Lindsey.
INTRODUCING SOCIAL NETWORKS.	A. Degenne and M. Forsé.
TAKING CHANCES: WINNING WITH PROBABILITY.	J. Haigh.
A PROBABILITY PATH.	S. Resnick.
NUMERICAL ANALYSIS FOR STATISTICIANS.	K. Lange.
MATHEMATICAL STATISTICS.	J. Shao.
RESAMPLING METHODS. A Practical Guide to Data Analysis.	P.I. Good.
SAMPLING OF POPULATIONS: METHODS AND APPLICATIONS, 3rd edition.	P.S. Levy and S. Lemeshow.
MATHEMATICAL METHODS IN SAMPLE SURVEYS.	H.G. Tucker.
FUNDAMENTALS IN THE DESIGN AND ANALYSIS OF EXPERIMENTS AND SURVEYS. GRUNDLAGEN DER PLANUNG UND AUSWERTUNG VON VERSUCHEN UND ERHEBUNGEN.	D. Rasch, L.R. Verdooren and J.J. Gowers.
DESIGN AND ANALYSIS OF EXPERIMENTS.	A. Dean and D. Voss.
FRACTIONAL FACTORIAL PLANS.	A. Dey and R. Mukerjee.
RECURSIVE PARTITIONING IN THE HEALTH SCIENCES.	H. Zhang and B. Singer.
NORMAL APPROXIMATION: NEW RESULTS, METHODS AND PROBLEMS.	V.V. Senatov.
THEORY OF RANK TESTS, 2nd edition.	J. Hájek, Z. Šidák and P.K. Sen.
STATISTICAL ANALYSIS IN CLIMATE RESEARCH.	H. von Storch and F.W. Zwiers.
RELIABILITY MODELLING: A Statistical Approach.	L.C. Wolstenholme.
QUEUEING NETWORKS: CUSTOMERS, SIGNALS AND PRODUCT FORM SOLUTIONS.	X. Chao, M. Miyazawa and M. Pinedo.
NEURAL NETWORKS: AN INTRODUCTORY GUIDE FOR SOCIAL SCIENTISTS.	G.D. Garson.
STOCHASTIC MODELS IN RELIABILITY.	T. Aven and U. Jensen.
RISK-NEUTRAL VALUATION. Pricing and Hedging of Financial Derivatives.	N.H. Bingham and R. Kiesel.
ESSENTIALS OF STOCHASTIC FINANCE: FACTS, MODELS, THEORY.	A.N. Shiryaev. Translated from the Russian by N. Kruzhilin.
RISK MODELING, ASSESSMENT, AND MANAGEMENT.	Y.Y. Haimes.
STOCHASTIC METHODS IN HYDROLOGY. Rain, Landforms and Floods.	O.E. Barndorff-Nielsen, V.K. Gupta, V. Perez-Abreu and E. Waymire (Eds.).
ACTIVE CONTOURS. The Application of Techniques from Graphics, Vision, Control Theory and Statistics to Visual Tracking of Shapes in Motion.	A. Blake and M. Isard.
STATISTICAL SHAPE ANALYSIS.	I.L. Dryden and K.V. Mardia.
MEASUREMENT AND CALIBRATION REQUIREMENTS FOR QUALITY ASSURANCE.	A.S. Morris.
ACHIEVING QUALITY THROUGH CONTINUAL IMPROVEMENT.	C.W. Burrill and J. Ledolter.
STATISTICAL QUALITY CONTROL: STRATEGIES AND TOOLS FOR CONTINUAL IMPROVEMENT.	J. Ledolter and C.W. Burrill.
MULTIVARIATE REDUCED-RANK REGRESSION. Theory and Application.	G.C. Reinsel and R.P. Velu.
STATISTICAL INFERENCE FOR SPATIAL POISSON PROCESSES.	Y.A. Kutoyants,
APPLIED REGRESSION ANALYSIS, 3rd edition.	N.R. Draper and H. Smith.
EXPONENTIAL FAMILY NONLINEAR MODELS.	B.C. Wei.
COMPARISONS OF STOCHASTIC MATRICES WITH APPLICATIONS IN INFORMATION THEORY, STATISTICS, ECONOMICS AND POPULATION SCIENCES.	J.E. Cohen, J.H.B. Kemperman and Gh. Zbaganu.
EXPERIMENTAL DESIGN TECHNIQUES IN STATISTICAL PRACTICE: A Practical Software Based Approach.	W.P. Gardiner and G. Gettingby.
DESIGN AND ANALYSIS OF CLINICAL TRIALS: Concepts and Methodologies.	S.C. Chow and J.P. Liu.
APPLIED CATEGORICAL DATA ANALYSIS.	C.T. Lee.
STATISTICAL METHODS FOR RELIABILITY DATA.	W.Q. Meeker and L.A. Escobar.
FRACTALS AND SCALING IN FINANCE: Discontinuity, Concentration, Risk.	B.B. Mandelbrot, with a foreword by R.E. Gomory.
METHODS OF MATHEMATICAL FINANCE.	I. Karatzas and S.E. Shreve.
THE INVERSE GAUSSIAN DISTRIBUTION. Statistical Theory and Applications.	V. Seshadri.
RANDOM NUMBER GENERATION AND MONTE CARLO METHODS.	J.E. Gentle.

Contents:
1. Introduction
2. Definitions and basic properties
3. Methods of constructing copulas
4. Archimedean copulas
5. Dependence
6. Group differences and bias in assessment
7. Additional topics

Interest in copulas has increased markedly in the last decade; hence this monograph is a welcome addition to the literature. Copulas may be viewed as multivariate distribution functions with uniform (0,1) marginals or as functions that join multivariate distribution functions to their one-dimensional marginals. Their main interest to statisticians lies in the construction and simulation of multivariate distributions and in the study of scale-free measures of dependence. The monograph provides a systematic mathematical development of the fundamental properties of copulas and their main applications. Bivariate situations dominate the exposition but each of Chapters 2-5 ends with a multivariate section. There are roughly 100 examples, 150 exercises, and a bibliography of nearly 180 items. This book is introductory but not elementary. In order to benefit from it the reader needs a thorough knowledge of upper-level undergraduate probability and mathematical statistics (measure-theoretic probability is not a requisite). Some familiarity with nonparametric statistics is also helpful.

Contents:
1. The wavelet transform
2. Multiresolution support and filtering
3. Deconvolution
4. 1D signals and Euclidean data analysis
5. Geometric registration
6. Disparity analysis in remote sensing
7. Image compression
8. Object detection and point clustering
9. Multiscale vision models

Readership: Graduate students and researchers in astronomy, electrical engineering, physics, geophysics and medical imaging

This book gives a convincing demonstration of the power of wavelet and multiscale methods in image processing, and will be instrumental in furthering the use of this methodology. The illustrations are taken from astronomy, electrical engineering, remote sensing and medicine. A software package has been implemented but does not accompany the book. It may be ordered through the Internet for £850.00.
The book will be of most value to researchers already experienced in image processing. Novices will not find it easy to follow without the computational support.
The sections on noise modeling would have benefited from a statistician's eye prior to publication. Inconsistencies in the notation makes it difficult to distinguish between random variables and their values. Probability density functions are called probabilities; bias is the estimated mean divided by the true mean. Yet, for all this lack of clarity in the writing, the book leaves no doubt that the authors understand the statistical techniques being described.

Contents:
1. Brownian motion and stochastic integrals
2. Stochastic differential equations
3. Linear stochastic differential equations
4. Stability of stochastic differential equations
5. Stochastic functional differential equations
6. Stochastic equations of neural type
7. Backward stochastic differential equations
8. Stochastic oscillators
9. Applications in economics and finance
10. Stochastic neural networks

The book contains the Brownian motion theory of stochastic integration. In this respect it does not differ from the many textbooks on the topic. This book also contains the usually omitted, backward stochastic equations and quasi-linear partial differential equations. The applications to finance are the standard ones; optional stopping time problems are part of the same chapter. The reader will find the book interesting, especially because of the less known applications to stability and neural networks.

Contents:
PART I : General Theorems on Ergodicity and Stability
1. General ergodicity and stability theorems for Harris irreducible Markov chains
2. Ergodicity and stability conditions for Markov chains not related to Harris irreducibility
3. Stochastically recursive sequences and their generalizations (Markov chains in random environments)
4. Ergodicity of stochastic processes in continuous and discrete time
PART II : Ergodicity and Stability of Multi-Dimensional Markov Chains and Markov Processes
5. Conditions of positive recurrence and ergodicity of multi-dimensional Markov chains and the method of Lyapunov functions
6. A description of multi-dimensional processes to be studied. Ergodicity, stability, and probabilities of large deviations of one-dimensional Markov chains
7. Ergodicity and stability of two-dimensional Markov chains and the method of approaching times
8. Markov chains in positive octants of three and more dimensions and the method of approaching times
9. Ergodicity and stability of multi-dimensional diffusions and jump Markov processes
10. Transition phenomena for one-dimensional Markov chains: Approximation of stationary distributions
PART III: Auxiliary Propositions. Ergodicity and Stability of Queuing and Communication Networks
11. Estimates of moments and probabilities of large deviations for certain random walks
12. Ergodicity and stability of queuing and communication networks

Ergodicity theorems deal with convergence, when time tends to infinity, to a stationary limit that is independent of the initial conditions of the stochastic process. If this stationary limit is not very sensitive to small changes of the local characteristics of the process, we talk about stability. In this book, ergodicity is studied for various classes of stochastic processes such as Markov chains, stochastically recursive sequences, Markov chains in random environment, vector-valued Markov processes, etc. The applications are in the area of queuing and communication networks. This book provides a second treatment of the subject which will be appreciated by the researchers in this field.

Contents:
1. Introduction
2. Basic distributional results
3. Moment relations, bounds and approximations
4. Characterizations
5. Inference
6. General record models
7. Random and point process record models
8. Higher dimensional problems

Readership: Postgraduate students in statistics and mathematics; research workers in meteorology, hydrology, market analysis, and sports analysts

Here is a lively survey of the theory underlying sequences of record values and times, presented in the form of a textbook for a one-term postgraduate course but suitable also for self-study. It follows on from the authors' First Course in Order Statistics [Short Book Reviews, Vol. 13, p. 5]. Familiarity with the earlier book would be useful but it is not necessary—the essential background demanded by the authors is a one-year course in introductory mathematical statistics, though some understanding of stochastic processes would be useful. There are plenty of exercises but no solutions: helpful references are given for some of the harder and the more thought-provoking exercises. The students' attention and their understanding of the relationships between topics is greatly assisted by the user-friendly chapter introductions. For the exponential and geometric distributions, and their close relatives with lack-of-memory type properties, there are generally very simple results; a subset of the material in the book could very easily be adapted for the use of advanced undergraduates. In general, however, nice results are less easy to come by for records than for order statistics. A plan of feasible routes through the book would be useful in a subsequent printing.

Contents:
1. Quality control with multivariate data
2. The multivariate normal distribution in quality control
3. Quality control with externally assigned targets
4. Quality control with internal targets—multivariate process capability studies
5. Quality control with targets from a reference sample
6. Analyzing data with multivariate control charts
7. Detection of out-of-control characteristics
8. The statistical tolerance regions approach
9. Multivariate quality control with units in batches
10. Applications of principal components
11. Additional graphical techniques for multivariate quality control
12. Implementing multivariate quality control
APPENDIX 1: MINITAB™ Macros for Multivariate Quality Control
APPENDIX 2: The Data from the Case Studies
APPENDIX 3: Review of Matrix Algebra for Statistics with MINITAB™ Computations

Readership: Industrial practitioners of statistics, quality control, reliability, and manufacturing; and upper-level undergraduate and graduate students in these fields

This monograph is a practical introduction to the concepts and execution of multivariate quality control in industrial settings. Manufacturing data are naturally multivariate—yet to this day most reference books on quality control deal with only one variable at a time. Rapid advances in software have finally all but removed the computational barrier, the historical excuse for manufacturers' reluctance to employ multivariate methods. The authors have included MINITAB™ macros that perform the methods described in the book. The book covers both standard and novel approaches, the latter including an interesting tolerance region approach for assessing individual multivariate observations. The book's presentation is compact, and will require the reader to have some maturity in matrix algebra, statistical inference, and basic quality control. The book will make a nice supplementary text for advanced students, or a reference for seasoned practitioners.

Contents:
PART I : Modelling
1. Modelling with primal problem
2. Modelling with dual problem
PART II : Linear Semi-infinite Systems
3. Alternative theorems
4. Consistency
5. Geometry
6. Stability
PART III: Theory of Linear Semi-infinite Programming
7. Optimality
8. Duality
9. Extremality and boundedness
10. Stability and well-posedness
PART IV: Methods of Linear Semi-infinite Programming
11. Local reduction and discretization methods
12. Simplex-like and exchange methods

A linear semi-infinite optimization problem has a linear objective function and linear constraints, however either the number of variables or the number of constraints but not both is infinite. This book is a research text—it comprehensively presents the theory of semi-infinite programming. The numerical methods to solve problems are given for general problems in conceptual form. There are no worked numerical examples. Every chapter has detailed bibliographical notes and a set of exercises with significant examples and counter-examples. Even the examples of modelling are presented in the most general fashion. Semi-infinite programming is a fascinating area with many challenges. If you want to be up-to-date on the theory in this field, then you will value this text highly; otherwise, if you want to formulate and solve problems, then access to a library will be sufficient.

Contents:
1. Introduction
PART I: The Conceptual Framework
2. Issues in assessing the contribution of research and   development to productivity growth [1979]
PART II : R&D and Productivity at the Firm Level: The Evidence
3. Returns to research and development expenditures in   the private sector [1980]
4. Productivity, R&D and basic research at the firm level   in the 1970s [1986]
5. Productivity and R&D at the firm level [1984]
6. Productivity growth and R&D at the business level:   Results from the PIMS database [1984]
7. Comparing productivity growth: An exploration of French and US industrial and firm data [1983]
8. R&D and productivity growth: Comparing Japanese and US manufacturing firms [1990]
PART III: R&D and Productivity at the Industrial Level
9. R&D productivity growth at the industry level: Is there still a relationship? [1984]
10. Interindustry technology flows and productivity growth: A re-examination [1984]
11. The search for R&D spillovers [1992]
12. R&D and productivity: The unfinished business [1996]
PART IV: Patent Statistics
13. Patent statistics as economic indicators: A survey   [1990]
PART V: Interim Conclusions
14. Productivity, R&D, and the data constraint [1994]

Readership: Econometricians, productivity analysts, official statisticians and economic historians

This is a collection of papers [with dates in brackets] written by Zvi Griliches, and with colleagues, on the problem of relating productivity growth to research and development (R&D) expenditure in industry. With the Introduction, it spans almost twenty years of work and draws upon work going back over forty years. The references provide an insight into the school that grew out of this work, and which continues to contribute. In keeping with its title, the papers deal with econometric estimates of production functions containing an R&D variable. However, like all good research problems, the book makes clear that, while privately funded basic R&D contributes to improved productivity, there is still work to be done. Some of that work is on patent statistics, which is surveyed.
A recurring theme is data and the difficulty of working with data from official statistical organizations. Quality adjustment in price indices is raised, with a plea that other technologically advanced industries have a price index based on hedonic regression methods like that introduced in the United States for the computer industry in 1986. The debate goes on, about price indices, about productivity and measuring it for service industries, and, about the role of R&D. This book provides a good introduction to the literature and is a reminder to researchers early in their career that one person can make a difference.

Contents:
1.Images and quantities
2.Visual and statistical thinking: Displays of evidence for making decisions
3.Explaining magic: Pictorial instructions and disinformation
4.The smallest effective difference
5.Parallelism: Repetition and change, comparison and surprise
6.Multiples in space and time
7.Visual confections: Juxtapositions from the ocean of the stream of the story
Readership: Statisticians, students, public policy-makers and laymen

This is the third in a series of beautiful, colourful and informative books by the author of The Visual Display of Quantitative Information [Short Book Reviews, Vol. 4, p. 1] and Envisioning Information [Short Book Reviews, Vol. 10, p. 41] . As with the other volumes, this one is a valuable source of information on interesting illustrations and points out what are good ones, cluttered ones, exceptional ones, etc. The second paragraph of the book sums up the contents very well: "This book describes design strategies—the proper arrangement in space and time of images, words, and numbers—for presenting information about motion, process, mechanism, cause and effect. These strategies are found again and again in portrayals of explanations, quite independent of the particular substantive content or technology of display."
The present book, as the others, is full of illustrations. Two interesting ones, given at length in Chapter 2, describe the solution to the cholera epidemic in 1854 in London by John Snow, and the investigation into the explosion of the space shuttle, Challenger, in 1986. A profound statement by physicist Richard Feynman is quoted: "For a successful technology, reality must take precedence over public relations, for Nature cannot be fooled," of which policy-makers should take heed. Tufte ends this section in summarizing the various graphical illustrations of these cases by saying "it also helps to have an endless commitment to finding, telling, and showing the truth." These two statements should be remembered along with the fact that "thoughtful designs may [also] skilfully present false information."
Because during his research for these books, Tufte became "enchanted by the elegant and precise beauty of the best displays of information", he wrote, designed and published the books himself. This is, like the other books, full of interesting ideas; it should be owned by all who are interested in the statistical presentation of data and who admire beautiful and intriguing pictures and ideas.

Contents:
1. Introduction
2. The methods of Boscovich and Mayer
3. Laplace's work on the methods of Boscovich and Mayer
4. Laplace's minimax procedure
5. The method of least squares
6. Statistical foundations of the method of least squares
7. Adrain's work on the normal law
8. Gauss's most probable values
9. Laplace's most advantageous method
10. Gauss's most plausible values
11. Gauss's method of adjustment by correlates
12. Mechanical analogies for the method of least squares
13. Orthogonalization procedures
14. Thiele's derivation on the method of least squares
15. Later work on the method of least situation
16. Concluding remarks

Before there were modern computers, when a reference to the word 'computer' meant a human working with pencil and paper, the difficulties of computation were a serious barrier to the advancement of statistical technology. Farebrother's history of the calculus of observations (as that portion of statistics encompassing linear models was once called) covers more than computation, but it particularly shines in explaining how some of the greats of the past (such as Gauss, Laplace, Cauchy) and many lesser figures (such as Boscovich, Donkin, Thiele) confronted major difficulties in statistical analysis, and how almost in passing they developed much of modern matrix algebra and even precursors to linear programming. Farebrother's story is told in lucid prose and with detailed algebraic development, and it reflects well the decades of painstaking scholarship that lie behind it. The work requires an investment of effort by the reader, in coping with the variety of notation employed and carefully following long algebraic arguments. Some portions, such as those on Donkin and Thiele, are heavy going indeed. But the reader who makes the investment will reap the rich reward of a much fuller appreciation of 18th and 19th century statistical work, and consequentially of modern statistics. We learn about the development of orthogonalization methods, of both graphical and analytical methods for the minimization of total absolute deviations, and about the important problems that gave rise to these developments. In the process the author cannot resist numerous and interesting short side trips, for example where we learn what the explorer David Livingstone's Problem of the Nile had to do with statistics.

Contents:
1. Overview
2. Economics and other sciences
3. The dominant paradigm: RAP
4. The Keynesian paradigm: KP
5. Of debt and taxes: KP versus RAP
6. Some other paradigms
7. The criteria of validity in economics
8. "Successes" of positive economics: Two examples
9. Welfare economics
10. RAP and the ideology of laissez-faire
11. What is economics good for?
12. Prizes, establishment, and heroes
13. The boundaries of economics

Readership: Academics of most disciplines, government bureaucrats, journalists, all others interested in public affairs, and economists

This book presents the reflections of an old-timer in the discipline on the nature of economics and the way it has evolved over the past fifty years or so. The author is a prominent economist who has made contributions over that whole period. His are mature, sophisticated and well-considered reflections. His intention is to reveal what economics is all about and he succeeds quite well. The book says a lot about methodology but it is not a treatise on methodology. Instead it probes the inner nature of economics; it reviews how economics is practised by the most reputable practitioners. No use of advanced mathematics is made, but the discussion is at a quite high level of sophistication. The points the author makes are insightful, the writing is lucid, and the style carries the reader's interest well.
For the non-economist, this is probably not an easy read but well worth the effort. It would reward such a reader with a close, inside look at the subject, at the issues that are disputed by its professionals, and at the way economists go about their arguments. For the economist it offers thought-provoking, interesting, and quite comprehensive collective self-examination. For both sets of readers, this is a worthwhile book which I recommend strongly. I would especially commend it to serious graduate students and young practitioners just starting their careers as professional economists. Nothing at all comparable can be found. It is a unique effort and one to be appreciated.

Contents:
PART I : Small Area Studies and Disease Mapping
PART II : Atmospheric Pollution Studies
PART III: Disease Risks and Social Effects
PART IV: Effects of Radiation
PART V: Agriculture and the Food Chain

This is the fourth volume in the sequence on Statistics for the Environment [Short Book Reviews, Vol. 14, p. 2; Vol. 15, p. 3; Vol. 18, p. 3]. It contains a summary of four introductory talks by scientists from Dutch national research institutes, plus twenty other papers, all presented at the SPRUCE Conference held at Enschede, The Netherlands, in September 1997. The third talk refers to the now widely recognized anomaly that despite the increasing income, health and welfare of the world population, we see an ever-increasing concern for the future. Many of these anxieties have been addressed by the SPRUCE sequence of conferences and the associated books. As with the other three volumes, this is a well-edited package, with an integrated and readable set of papers presented in a consistent style. It should be of interest to all statisticians studying spatial and spatio-temporal modelling, and also to environmentalists looking for assistance from statistics.

Contents:
1. Computer storage and manipulation of data
2. Basic vector/matrix computations
3. Solutions of linear systems
4. Computation of eigenvectors and eigenvalues and the singular value decomposition
5. Software for numerical linear algebra
6. Applications in statistics

Readership: Undergraduate or graduate students interested in statistical computing

The solution of linear systems of equations and the calculation of eigenvalues and eigenvectors is of immense importance not only in statistics but also in a wide variety of other areas. This account of the subject starts with a discussion of the manner in which data are stored—essential for an understanding of the accuracy with which data are represented, to appreciate the errors that can arise in numerical calculations and to understand various techniques for minimizing such errors. There follows a standard account of vector and matrix results germane to the subsequent chapters. The chapter on solutions of linear systems discusses the commonly used methods and in some cases relevant algorithms are provided. There is a useful account of some of the software available for performing numerical linear algebra. The material throughout is clearly presented with plenty of exercises and motivating discussion. It would make an excellent text to accompany a course on statistical computing.

Contents:
1. Fundamental concepts
2. Examples of probability functions
3. The Monte Carlo method
4. Statistical tests
5. General concepts of parameter estimation
6. The method of maximum likelihood
7. The method of least squares
8. The method of moments
9. Statistical errors, confidence intervals and limits
10. Characteristic functions and related examples
11. Unfolding

Readership: Graduate and advanced undergraduate students in the physical sciences who need to draw quantitative
information from experimental data

I am always intrigued by the differences in flavour and emphasis that different disciplines bring to even the same statistical tools, not to mention the different choices of statistical tools which play the major roles in different disciplines. Thus this overview of statistical methods in the physical sciences does not refer to analysis of variance, but it does discuss regularization methods. "Unfolding" in this book refers to deconvolution to remove measurement error, a rather different use of the term from the quantitative social and behavioural sciences.
To me, at least, the book presents a rather unusual mix in the way the material is presented. Thus, for example, we have discussions (or, at least, definitions) of conditional probability. Bayes' theorem, frequentist and subjective probability, and likelihood before the notion of a histogram is introduced. The book seems to sit halfway between an elementary primer on how to use basic statistical methods and an introduction to the more mathematical aspects of statistics. The background knowledge assumed includes linear algebra, multivariable calculus, and some knowledge of complex analysis, but no prior knowledge of probability or statistics.
The material presented in this book is dense. In less than two hundred pages, it takes the reader from the basic notions of probability, through neural networks, Monte Carlo methods, and regularization techniques. I would imagine that readers new to the area would find it hard going, and would benefit from some supplementary reading material: the author's description of the book as a "guide" to the practical application of statistics in the area is astute.

Contents:
1. Introduction
2. Sampling distributions
3. Logistic regression
4. Log-linear models
5. Coordinate-free models
6. Additional topics
APPENDIX A : Power for the Chi-Squared Tests
APPENDIX B : Program for Exact Tests
APPENDIX C : The Hypergeometric Distribution

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

There is only one word to describe this book: excellent! The author has achieved an admirable blend of pertinent statistical theory and practical examples. Chapter 2 covers theoretical aspects of the Poisson, binomial and hyper-geometric and multinomial distributions; Fisher's exact test and estimation of sample sizes.
Chapters 3 and 4 cover logistic regression and log-linear models for cross-classified count data. Acknowledging that there are now several good books on logistic regression, the author places rather more emphasis on log-linear models. Complete SAS programs using the GENMOD or LOGISTIC procedures are given and results of these analyses are interpreted in detail. It is good to see Simpson's paradox discussed a number of times.
Chapter 5, adopting a generalized-linear-models approach, covers various incomplete cross-classifications. Conditions for the existence of maximum likelihood estimates are discussed, giving insight into problems that can arise in practice with awkward configurations of data.
The final chapter gives brief introductions to the analysis of longitudinal data, case control studies, sparse data and goodness-of-fit statistics.
The extensive exercises at the end of each chapter are divided into applied and theoretical sections, each of which amplifies the topics discussed. Some forty examples are discussed in the text.
Students completing a course based on this book should be able to analyze and to interpret most of the count or proportional data they are likely to encounter in practice; they would have insight into the mathematical basis of the subject, and after reading Chapters 5 and 6, may even have their appetites whetted for research.

Contents:
PART I : Descriptive Statistics – Compressing Data
1. Why one needs to analyze data
2. Data representation and compression
3. Analytic representation of random experimental data
PART II : Modeling Uncertainty
1. Algorithmic complexity and random strings
2. Statistical independence and Kolmogorov's probability theory
3. Chaos in dynamical systems: how uncertainty arises in scientific and engineering phenomena
PART III: Model Specification – Design of Experiments
1. General principles of statistical analysis
2. Statistical inference for normal populations
3. Analysis of variance

This book has the subtitle "Uncertainty, Complexity and Chaotic Behaviour in Engineering and Science." It is intended for a course in introductory probability and statistics, aimed chiefly at engineering and science students, and taking a unique and particularly contemporary approach to introductory data analysis. (It might require a somewhat adventurous statistics instructor to use this with statistics majors, because some of the material will seem unfamiliar and possibly unnecessary to those accustomed to traditional introductory statistics offerings.) The prerequisites comprise introductory calculus, some differential equations and linear algebra, and a basic computer programming course.
The book is highly data-oriented, with an unusually large collection of real-life examples taken from industry and various scientific disciplines; this includes the natural, life, and social sciences. Indeed, the approach could be described as almost data-driven, and the learning method emphasizes hands-on computer experiments and numerical techniques. Many experiments and exercises programmed in Mathematica are provided, and, as the authors themselves say, using this book without doing the accompanying Mathematica experiments would be like playing Chopin on the accordion.
The book does cover the usual standard introductory probability theory, including axioms and properties, expectation, independence, discrete and continuous distributions, multivariate distributions, moment generating functions, variance and covariance, and the Poisson and Gaussian approximations. (Some of this material is done at a higher level than might sometimes be desired for an introductory course.) Computer experiments are provided to illustrate the Law of Large Numbers and the Central Limit Theorem. There is considerable material on pseudo-random number generation and Monte Carlo methods.
On the statistical side, the text covers types of data, descriptive statistics, data compression, design of experiments, model selection, maximum likelihood and least-squares estimators, regression and correlations, confidence intervals, and hypothesis tests for normal populations, and one-way and two-way analyses of variance. Again, some of these topics might be presented at a higher level than desired for some introductory statistics courses. The material is largely interwoven with the material on probability theory and other topics.
The book departs from the above standard fare, however, by including detailed coverage of such contemporary topics as chaotic dynamical systems, the nature of randomness, computability and Kolmogorov complexity, encryption, ergodicity, entropy, and even fractals. Some of these topics might be more important to students in certain areas of the physical sciences, engineering, and computing. Statisticians might wish to learn more about them. While it should be possible to work around some of this material if it is not desired, and would be missing the whole point of the authors' approach were one to leave too much out. It is an interesting book.

Contents:
1. Introduction
2. The dichotomous data problem
3. The one-sample location problem
4. The two-sample location problem
5. The two-sample dispersion problem and other two-sample problems
6. The one-way layout
7. The two-way layout
8. The independence problem
9. Regression problems
10. Comparing two success probabilities
11. Life distributions and survival analysis

This book provides a comprehensive guide to nonparametric statistics through extensive use of 'real-world' examples. It provides the reader with the opportunity to understand the theoretical aspects of nonparametric techniques as well as to explore the ways in which the most common techniques are applied to real data. It covers areas of nonparametric statistics, such as example estimation, regression, bootstrapping, that many other texts have failed to cover and in that sense is a useful reference for 'up-to-date' methods. Particularly appealing is the way in which each technique is presented in a standard format: a description of the procedure, details of large-sample approximations that can be made, how to deal with tics, an example, some comments from the authors, some details of the theoretical aspects, some exercises.
Despite its applied nature, the book is fairly technical in places and in that sense may be best suited to those who have a reasonably strong quantitative background rather than other professionals wishing to apply the techniques within their discipline.

Contents:
1. Methodological issues in longitudinal research
2. Analysis of longitudinal categorical data using optimal scaling techniques
3. Univariate and multivariate analysis of variance of longitudinal data
4. Structural equation models for longitudinal data
5. Multilevel analysis of longitudinal data
6. Log-linear and Markov modelling of categorical longitudinal data
7. Epilogue

Readership: Statisticians, graduate students of statistics, researchers in the behavioural sciences, education, sociology, medicine and biometry

Measurements made on the same individuals on a number of occasions occur in many fields of research. Methods for the statistical analysis of such data have given rise to a vast literature, but, as the authors point out, books on the subject tend to deal with only one method of analysis. There are texts on ANOVA and MANOVA methods for continuous, normal responses; on structural equation models for investigating causal relationships; on multilevel modelling for hierarchical data; on Markov or log-linear models for discrete data and forms of multiple correspondence analysis for the graphical display of the relationship between variables, subjects and times. These, often very good, texts are informative but leave the reader a little uncertain as to how the whole area of longitudinal data analysis fits together. Until now that is!
Here is a single book which describes all these methods, their areas of applicability, their strengths, their limitations, and how they relate to each other.
The first chapter deals in general terms with the design of longitudinal studies and their subsequent validity and interpretation. It should be required reading for everyone planning such a study, no matter what method of analysis they intend to use.
In Chapters 3 to 6, the methods are described. The purpose of each is clearly stated and is motivated by a simple set of data, which illustrates its main features. Then more complex sets are analyzed and interpreted. Careful attention is given to the exact statement of the null hypothesis of any test described. Complex matrix formulations are avoided by skilful use of diagrams, but sufficient formulae are used to make the text precise. At every point copious references are given to the literature and comparisons with other approaches are described in the book. Computer programs that will perform the analysis are named. Emphasis is always on the practicalities of the analysis.
In the Epilogue, the relationships between all the techniques are further reviewed and recommendations on how to select a method are given. Finally, topics not discussed in the book, such as item-response theory and event-history analysis are mentioned with references.
This is an outstanding book. Its authors have performed a valuable service to the statistical community and to those who need to elucidate data collected on a number of occasions. Read it! Enjoy it! You will learn a lot!

Contents:
1. Introduction to regression modeling of survival data
2. Descriptive methods for survival data
3. Regression models for survival data
4. Interpretation of a fitted proportional hazards regression model
5. Model development
6. Assessment of model adequacy
7. Extensions of the proportional hazards model
8. Parametric regression models
9. Other models and topics
APPENDIX 1 : The Delta Method
APPENDIX 2 : An Introduction to the Counting Process Approach
APPENDIX 3 : Percentiles for Computation of the Ball and Wellner Confidence Band

Readership: Statisticians, researchers in medicine and biometry, epidemiologists

The previous book by these authors, Applied Logistic Regression [Short Book Reviews, Vol. 10, p. 27], has won a firm place in the literature as one of the best introductory texts on the analysis of proportional data. It has helped to make logistic regression accessible beyond the statistical community, and also offered theoretical statisticians an insight into the issues that must be considered when applying statistical theory to real data.
Their new book is sure to occupy the same niche in the survival analysis literature. The authors have a gift for putting mathematical concepts into words and for interpreting the results of a complex statistical analysis in terms of the background to the data.
The major emphasis is on the proportional hazards model and the discussion follows the usual regression modeling paradigm: preliminary data, description, model selection, examination of fit, interpretation of the estimates. Mathematics at the level of an introductory regression course gives precision and the surrounding discussion a wealth of insight. Statistical packages that offer procedures for the analyses are referenced. Three sets of data, available on the "Web", are used throughout as examples. Exercises follow each chapter.
The material covered in the text is right up to date. Chapter 9 discusses recurrent event models, frailty models and nested case-control studies. A discussion of these important topics at this level is not available elsewhere. References to the counting processes approach to survival analysis are made throughout the text and a brief introduction to this topic is given in an appendix.
This book makes a major contribution to the understanding of "time-to-event" data. It is highly recommended.

Contents:
PART I : Foundations
1. Oracles, norms, and science
2. Modeling
3. Dreams and disappointments
PART II : The Information Cycle
4. A priori influences on the information cycle
5. Measurement of human information
6. Limitations of measurement in the social sciences
7. Information for inferences: What are social science data?
8. Causality
9. Models and policy making

Readership: Researchers and practitioners in the social and decision sciences, decision-makers, and students preparing for these fields

This book examines the nature and role of formal models in the 'human' sciences, seeking to explore the principles which (ought to) guide the appropriate use of data and models. It begins with a discussion of the problems associated with social science data, including measurement error, problems of precise definition, the spurious sense of precision associated with the measurement of social phenomena, and consequences of emphasis on processes, and then goes on to describe principles which should guide social research. The book will be at best of peripheral interest to most practising statisticians. The authors comment, for example, that 'We have examined a sample of several graduate programs in statistics at major American universities. The programs are very strong in the technical aspects of data analysis. Nevertheless, we were not able to find even one example of a course offered by a graduate-level statistics program that addressed underlying philosophical issues, the history of statistics, or policy implications of statistical analysis' (p. 13). On the other hand, new PhD students concerned with developing formal models for social processes would benefit from an awareness of its content.

Table des Matières:
1. Introduction
2. L'analyse canonique
3. Analyse factorielle inter-batteries
4. L'analyse des redondances
5. L'approche SIMPLS
6. L'algorithme NIPALS
7. La régression PLS univariée (PLS1)
8. Propriétés mathématiques de la régression PLS1
9. La régression PLS multivariée (PLS2)
10. Applications de la régression PLS
11. L'analyse canonique PLS
12. Traitement des données qualitatives
13. L'approche PLS

Ce livre donne une présentation théorique et pratique de la méthode de la régression PLS (Partial Least Squares). Il est démontré que cette méthode, qui est principalement connue dans la chimie, s'applique aussi dans beaucoup d'autres domaines. Il s'agit d'une méthode de régression pour décrire les relations entre la réponse et un grand nombre de variables d'entrée en l'absence d'un modèle théorique. L'algorithme de la régression PLS est décrit avec ses propriétés mathématiques. Il y a de nombreux examples et illustrations en utilisant le logiciel SIMCA-P for Windows.

Contents:
1. Introduction
2. Introduction to 2D scatterplots
3. Constructing 3D scatterplots
4. Interpreting 3D scatterplots
5. Binary response variables
6. Dimension-reduction subspaces
7. Graphical regression
8. Getting numerical help
9. Graphical regression studies
10. Inverse regression graphics
11. Sliced inverse regression
12. Principal Hessian directions
13. Studying predictor effects
14. Predictor transformations
15. Graphics for model assessment

Readership: Advanced regression practitioners with a good theoretical background

This is an intriguing and imaginative book. It discusses how graphical analysis can aid the investigation of regression data and perhaps help to reduce the data to a structure of smaller dimensionality. Although a number of data sets are analyzed extensively, the basic ideas are difficult to pick up on an initial reading due to the high technical density of the material. Natural questions are whether, when a reduction in dimension has been achieved, it has practical consequences that can be fully understood and whether the simplification relates to canonical dimensions (in the case of response surface data, for example). Such questions provide ample incentive to study this important book, perhaps via a seminar course. It is an essential library purchase.

Contents:
1. What is it all about?
2. The two basic, time-honored process capability indices: Cp and Cpk
3. First-generation modifications: Cpm and its close relatives
4. The avalanche
5. The benefit (or curse) of non-normality and asymmetry and how to get rid of them
6. A superstructure and unified approach to process capability indices
7. The dangerous but unavoidable area: Multivariate process capability indices
8. Practical issues in capability analysis
9. Just say yes!
APPENDIX: List of Univariate Process Capability Indices

Readership:Statistically sophisticated practitioners who employ process capability indices to characterize manufacturing processes, and statisticians and advanced students who seek a summary of the past fifteen years' research in the field

This book is the best overview yet of the broad field of process capability indices. A process capability index (PCI) is a measure that seeks to represent, in a single number, a manufacturing process's ability to deliver a product within specifications—typically in terms of a ratio of allowable variation (the customer's specifications versus the variation actually achieved by the process). Since the mid-1980's, industrial use of PCI's has polarized practitioners and theoreticians—with wide gulfs developing both within and between these populations. There have been accusations of statistical terrorism, manipulation, and abuse, and a consequent (healthy) explosion of debate and research. The authors aim to bridge the gaps among all parties, but I would bet that practitioners and their managers will be startled on reading this book to discover just how wide the gap has become as research sheds new light on the field.
The authors describe essentially all of the process capability indices in current use, summarizing their assumptions, statistical properties, intended use, and potential weaknesses. PCI's are random variables, after all, and an understanding of their variability and biases is a prerequisite to their respective applications. The book brings some theoretical foundation to these mostly ad hoc measures. The field is maturing but not yet ripe. The authors draw broadly from the research of the originators and other investigators of the various PCI's, and the compilation of references at the end of each chapter is a valuable contribution. Chapter 6 offers a "superstructure" of which several of the existing indices are special cases, and from which other indices can be developed. The authors introduce this approach, tongue-in-cheek: "The time-honored device of generalization utilized in statistics is usually a clever and delicate introduction of additional parameters."
The authors' enthusiasm is evident in their writing. The book includes more references to "current events" than is customary in statistics texts, primarily motivating examples from various manufacturing companies. This will date the book, but for a field that is changing so rapidly, these references give an interesting historical context.
Some bits of statistical theory are given in various appendices, but the reader will need some sophistication in addition to follow the derivations and to appreciate the conclusions in the various chapters. As to whether process capability indices are useful, the authors are positive. Their parting words: "Remember, it is much easier to lie without statistics than with them."

Contents:
1. Introduction
2. Information theory and likelihood models: A basis for model selection and inference
3. Practical use of the information-theoretic approach
4. Model selection uncertainty with examples
5. Monte Carlo and example based insight
6. Statistical theory
7. Summary

Obviously, the authors are enthusiastic disciples of Kullback-Leibler information/distance. The first four chapters are written with inspiration. Although it cannot be considered as a rigorous mathematical text, it is an interesting reading together with occasional historical assays. Other chapters contain a number of examples and derivations of some "practical" formulae and are much more technical but still the mathematical rigour is not their strongest feature. Remarks like "straightforward mathematics ..." and "... is mostly a straightforward exercise" do not help either to understand or to illuminate of the reported results. The use of "the best (approximating) model" instead of "the true model" is an interesting idea, which is frequently addressed in the book. I think that could make the following step they focused on more an empirical observation that "the best (approximating) model" changes when the number of observations increases (p. 69). Actually, this fact has a theoretical background and can be formulated in terms of nested models as dependence of complexity of the best model on the number of available observations. The book would gain a wider audience if the informational vocabulary were more frequently compared with the standard statistical language.

Contents:
11. Introduction to Volume II
12. Main crop with supplementary crops
13. Three or more main crops—Density constant
14. Varying densities for some or all crops in a mixture
15. Mixing ability effects when individual responses are available
16. Mixtures when responses not available
17. Spatial and density arrangements
18. Some analytical variations
19. Intercropping procedures
20. An intercropping bibliography

In an intercropping experiment, two or more crops and/or mixtures of these are grown simultaneously. Responses will depend, amongst other things, on the precise spatial arrangement of the crops and the density level of each crop. There may be multiple goals, for example, both the nutritional value of the food produced and its monetary worth may be of interest.
This volume is a sequel to the first which dealt with mixtures of two crops and sole crops. Chapters in the two books go in parallel so that the topics covered here are similar to those in Volume I [Short Book Reviews, Vol. 14, p.5]. Although the author points out the dangers of making too simple a generalization from two crops to three or more, he gives us an authoritative and extremely thorough account, biased more towards analysis than design, of intercropping experiments. There is a wealth of worked numerical examples, making this book an invaluable reference tool for those working in this area.

Contents:
1. The basic concepts
2. Simple random sampling
3. Stratified random sampling
4. Ratio estimator
5. Difference estimator and regression estimator
6. Systematic sampling
7. Cluster sampling
8. Probability proportional to size with replacement sampling
9. Varying probability without replacement sampling I
10. Varying probability without replacement sampling II
11. Multistage sampling
12. Multiphase sampling
13. Sampling on successive occasions
14. Some problems of inference under a fixed population set-up
15. Inference from a finite population using the prediction-theoretic approach
16. Errors in surveys
17. Randomized response techniques
18. Small area estimation

Readership: Survey statisticians, survey samplers, students taking an introductory course in sampling

The first twelve chapters of this book could be used as a very thorough and comprehensive first course in survey sampling methods. Later chapters are rather more specialized; each chapter has a set of exercises, many theoretical but some numerical, to support the theory covered. Solutions to some of these exercises are included at the end of the book; each chapter has its own extensive list of references, most of them from the early development of the subject with relatively few published in the last twenty years, the exceptions being for some of the more specialized material covered in the later chapters and to some of the author's own recent contributions to these topics. Admissibility of estimators and optimality of sampling strategies when sampling from a superpopulation are discussed in Chapter 14, which is followed in the next chapter by an investigation of model-dependent optimal strategies under alternative models in a superpopulation framework. The final chapters cover randomized response methods for dichotomous and polychotomous populations and some synthetic and composite estimators suitable for small geographical areas.

Contents:
1. Mathematical background
2. Convergence in probability and in law
3. Performance of statistical tests
4. Estimation
5. Multivariate extensions
6. Nonparametric estimation
7. Efficient estimators and tests

The book gives a comprehensive account of first-order large-sample theory for students who have taken two courses of calculus and some linear algebra. It does this by stating the more difficult results without proof, by using convergence in probability rather than almost sure convergence as the mode of probabilistic convergence—and by teaching the mathematics needed beyond the second calculus course. The book took several years to write, and its content has been well tested in courses. Notable features are the concise and thoughtful summaries at the ends of sections, and the extensive sets of problems, which occupy nearly one hundred pages.
There is a good deal more in the book than would be covered in a first graduate course in mathematical statistics, but students are likely to find it a handy and accessible reference. For example, Chapter 5 on multivariate extensions collects together the results they will need for vector-valued statistics. Chapter 6 summarizes the theory of U-statistics, gives an introduction to density estimation and explores briefly some uses of the bootstrap in point estimation of biases and variances. Chapter 7 tackles maximum likelihood estimation and associated tests, illustrating the results with continuous models and with categorical data models. That this chapter comes last underlines the fact, implicit in the title, that the book is more about mathematics of large-sample theory than about its uses. However, there is no extraneous mathematics: the treatment stays admirably close to the fundamentals of the frequentist statistical theory it supports.

Contents:
1. Introduction
2. Stochastic convergence
3. Delta method
4. Moment estimators
5. M- and Z-estimators
6. Contiguity
7. Local asymptotic normality
8. Efficiency of estimators
9. Limits of experiments
10. Bayes procedures
11. Projections
12. U-statistics
13. Rank, sign, and permutation statistics
14. Relative efficiency of tests
15. Efficiency of tests
16. Likelihood ratio tests
17. Chi-square tests
18. Stochastic convergence in metric spaces
19. Empirical processes
20. Functional delta method
21. Quantiles and order statistics
22. L-statistics
23. Bootstrap
24. Nonparametric density estimation
25. Semiparametric models

Readership: Graduate and postgraduate teachers, researchers in mathematical statistics

This well-written book covers limit theorems for the standard special cases of statistics such as M-, L-, R-, and U-statistics. Also likelihood inference and asymptotic efficiency of estimators and tests are treated. Important concepts in the book are Hajek's projection method, weak convergence and the unifying mathematical concept of approximation by limit experiences. There is a thirty-page summary of the theory of empirical processes and via the functional delta method, the asymptotic distribution of quantile estimators is derived from that of the distribution function estimators. The chapters on the bootstrap and on nonparametric density estimation are short and necessarily very incomplete. The last chapter is a seventy-page treatment of semiparametric models, an area which is still in full development. The notes to the history and the bibliography are a bit too selective. Each chapter has a section with a number of nice problems, some of which are hard. This makes the book interesting for teaching projects.

Contents:
1. Sums of independent random variables
2. Randomly stopped processes with independent increments
3. Decoupling of U-statistics and U-processes
4. Limit theorems for U-statistics
5. Limit theorems for U-processes
6. General decoupling inequalities for tangent sequences
7. Conditionally independent sequences
8. Further applications of decoupling

The method of decoupling aims at handling problems with dependent variables by the reduction to problems on related (conditionally) independent variables. This approach grew out of some situations where the traditional method of martingales was not applicable. Very important tools for the method are the so-called decoupling inequalities which compare functionals of the dependent variables to functionals of conditionally independent (decoupled) variables.
This book gives a well-written and very detailed description of the general theory and specific applications. More than half of the book is devoted to the decoupling of U-statistics and U-processes. It is shown how the decoupling inequalities play a crucial role in their asymptotic theory: law of large numbers, central limit theorem and law of the iterated logarithm. Also results for randomly stopped U-statistics are dealt with (extending results of Wald and Anscombe). Applications are given for the empirical median, M-estimators and hazard and distribution function estimators for left truncated data.

Contents:
1. Some history
2. Deterministic models
3. Stochastic models in continuous time
4. Stochastic models in discrete time
5. Rumours: Modelling spread and its cessation
6. Fitting epidemic data
7. The control of epidemics

Readership: Anyone with an interest in the mathematical aspects of epidemic models

This monograph provides an account of the development of mathematical epidemic models, from their foundations more than three hundred years ago with the work of John Graunt, and systematically covering the advances made this century up to about the mid-1970s. The coverage of work of the last twenty years is much less complete (though the spread of HIV gets a substantial treatment) and many major contributions to the literature over that period are not mentioned. However, the authors do not claim to provide a comprehensive account of recent developments, but rather aim to give a suitable background to the current literature, and in this they are wholly successful.
The book will be accessible, at least in part, and its study highly rewarding, to anyone with an interest in epidemic models and a good undergraduate degree in mathematics (including some basic applied probability). As is to be expected from these authors, standard models are discussed with great insight and the book is as much about the mathematical tools that can be brought to bear on the models as about the epidemic models themselves. Thus the organization of the book owes as much to the mathematical methods as to the models themselves. In these days, when too many modellers use numerical tools as a substitute for thought, the careful exposition of techniques is refreshing. This monograph will be much appreciated by the expert, as well as by the novice modeller.
The epidemics discussed are mostly of infections spread by direct contact. There is a fairly brief but useful discussion of model fitting and of the evaluation of control strategies. The spread of HIV is discussed extensively and other infections treated include measles and influenza. Exercises and Complements are provided at the end of each Chapter, providing a welcome opportunity for the reader to develop intuition and additional insights into the models discussed.

Contents:
1. Introduction and overview
2. Background and fundamental results
3. Processes with general state space
4. Processes with countable state space
5. Sample path stability
6. Little's formula and extensions
7. Insensitivity of queueing networks
8. Sample-path approach to Palm calculus

Readership: All those interested in theory and applications of stochastic processes

The standard approach to the study of stochastic processes is to formulate a particular class, say the M/M/k queue, and then to deduce its properties. However, its is well-known that many of these will not require the full strength of assumptions made, and it is especially beneficial in terms of developing intuition and understanding to consider what conditions are actually necessary for the particular property to hold. A classic example is Little's formula, which relates the average queue length to the arrival rate and average waiting time for very broad classes of queues.
Sample-path analysis concentrates on looking at the properties of single realizations of a stochastic process, thus requiring few stochastic assumptions (for example, the existence of long-term averages along sample paths), and then explores the additional assumptions needed for the results to extend to general classes of processes. Considerable insight can be gained from this approach.
This book, which focuses mainly on queueing systems and other input-output processes, will be a valuable resource to all those interested in applied probability and stochastic processes. While it is not a textbook, it contains a wealth of useful material for those giving courses in stochastic processes at all levels, and its extremely readable style makes it suitable for anyone with previous exposure to basic probability and stochastic processes.

Contents:
1. Introduction
2. LDP for finite dimensional spaces
3. Applications - The finite dimensional case
4. General principles
5. Sample path large deviations
6. The LDP for abstract empirical measures
7. Applications of empirical measures LDP

Readership: Mathematicians, physicists, engineers interested in a deeper understanding of the large deviation principle (LDP)

The first edition of this book was overall received as offering a very sound overview of large deviation techniques. The second edition is a largely updated version which, besides keeping the quality of the first edition, contains a lot of updated and/or added material. Moreover, new exercises have been added and the references have seen more than one hundred new additions. I strongly encourage those that already have the first edition to make a bit of extra shelf-space for this one. For those who are contemplating a text on the large deviation principle, this is an excellent text to have.

Table des Matières:
1.Le choix d'une méthode d'analyse statistique
2.Les conditions d'application des méthodes statistiques et l'examen initial des données
3.Les tests d'ajustement et de normalité et les observations aberrantes
4.Les transformations de variables
5.Les méthodes relatives à une ou deux proportions ou un ou deux pourcentages
6.Les tableaux de contingence
7.Les méthodes relatives à la dispersion
8.Les méthodes relatives à une ou deux moyennes
9.L'analyse de la variance à un critère de classification
10.L'analyse de la variance à deux critères de classification
11.L'analyse de la variance à trois et plus de trois critères de classification
12.Les comparaisons particulières et multiples de moyennes
13.Les méthodes relatives à la corrélation simple
14.Les méthodes relatives à la régression linéaire simple
15.La régression non linéaire simple et la modélisation
16.La régression multiple et la modèle linéaire général
17.L'analyse de la covariance
Lecteurs: Etudiants et enseignants dans de domaine et l'agronomie et de la biologie

Ce Tome 2 est une remise à jour du Volume 2 de l'ouvrage classique Théorie et méthodes statistique: applications agronomiques (1970) du même auteur. Les principales méthodes d'inférence statistique à une et à deux dimensions sont présentées dans une manière fondamentalement différente. Ce livre peut être utilisé à differents 'plans de lecture' ou 'niveaux d'étude'. Il y a des centaines de références, ce qui permet d'aller plus loin dans la littérature. Aussi intéressant pour l'enseignement sont le grand nombre d'exercises avec solutions et l'existence d'une adresse sur l'Internet afin d'obtenir d'autres informations.

Contents:
1.Introduction
2.Setting the stage: Categories of misuses
3.Know the subject matter
4.Definitions
5.Quality of basic data
6.Graphics and presentations
7.Methodology: A brief overview
8.Faulty interpretation
9.Surveys and polls: Getting the data
10.Surveys and polls: Analyzing the data
11.The law of parsimony: Ockham's razor
12.Thinking: Lack of forethought, lack of afterthought
13.Ectoplastistics
14.The body politic: Governments and politicians
15.Afterword

This is a collection of examples of defective statistical work arising in general from the misunderstandings of a naive or inexpert user. As might be predicted, the journalistic profession is responsible for many of the stupider and more obvious gaffes. A nice feature of the book is that the authors have managed to track down the original sources of some widely quoted but seriously inaccurate statistics. The second edition updates much of the earlier material with contemporary examples and includes new misuses that have appeared in the interim. Although there are many revealing misuses quoted, the text is vitiated by a number of examples that lack punch; one sometimes has the feeling of anticlimax: the text leads the reader to expect some compelling transgression only to find the instance given is limp. This reviewer found the rather folksy style distracting, although it may well appeal to others. Also credibility was not promoted by including characters such as Dr. K. Nowall and his nephew, a manager of a fast food store, Kenneth F. Capon.

Contents:
Introductions
The benefits of cases, by W.C. Parr
Partnering for the future of the statistics profession, by R.L. Iman
1.Are the fish safe to eat? Assessing mercury levels in fish in Maine Lakes, by J.A. Hoeting and A.R. Olsen
2.Chemical array validation, by R. Reeve and F. Giesbrech
3.Automating a manual telephone process, by M. Batcher, K. Cecco and D. Lin
4.Dissolution method equivalence, by R. Reeve and F. Giesbrecht
5.Comparison of hospital length of stay between two insurers for patients with paediatric asthma, by R.L. Houchens and N. Schoeps
6.Comparing nonsteroidal anti-inflammatory drugs with respect to stomach damage, by T. Filloon and J. Tubbs. Validating an assay of viral contamination, by L.I. Lin and W.R. Stephenson
8.Control charts for quality characteristics under nonnormal distributions, by Y.M. Chou, G.D. Halverson and S.T. Mandraccia
9.Evaluation of sound to improve customer value, by J.R. Voit and E. Walker
10.Improving integrated circuit manufacture using a designed experiment, by V. Czitrom, J. Sniegowski and L.D. Haugh
11.Evaluating the effects of nonresponse and the number of response levels on survey samples, by R.K. Smidt and R. Tortora
12.Designing an experiment to obtain a target value in the chemical process industry, by M.C. Morrow, T. Kuczek and M.L. Abate
13.Investigating flight response of Pacific Brant to helicopters at Izembek Lagoon, Alaska by using logistic regression, by W.P. Erickson, T.G. Nick and D.H. Ward
14.Estimating the biomass of forage fishes in Alaska's Prince William Sound following the Exxon Valdez oil spill, by W. Taam, L. McDonald, K. Coyle and L. Halderson
15.A simplified simulation of the impact of environmental interference on measurement systems in an electrical components testing laboratory, by D.A. Fluharty, Y. Wang and J.D. Lynch
16.Cerebral blood flow cycling: Anaesthesia and arterial blood pressure, by M.H. Kutner, K.A. Easley, S.C. Jones and G.R. Bryce
17.Modeling circuit board yields, by L. Denby, K. Kafadar and T. Land
18.Experimental design for process settings in aircraft manufacturing, by R.M. Sauter and R.V. Lenth
19.An evaluation of process capability for fuel injector process using Monte Carlo simulation, by C. Lee and G.A.D. Matzo
20.Data fusion and maintenance policies for continuous production processes, by N.D. Singpurwalla and J.N. Skwish

Readership: Teachers of statistics, especially of statistical consulting courses, students of statistics, statisticians, practitioners of statistics in business, industry and government

The case studies in this book are the work of twenty-two pairs of participants in the Collaboration Project, launched in 1995 with funding from the U.S. National Science Foundation. Each pair comprised an academic and someone from business, industry or government. The introductory essays by Parr and Iman argue effectively for the industry-academe collaboration and for the value of case studies in statistical education.
Three tables index the cases by the statistical methods which are applicable (twenty-four such), area of application, and levels of difficulty. While the format of case presentation is largely uniform throughout (introduction, background, questions of interest, data, analysis, references, notes to the instructor, and biographies of the authors), the level of detail within the sections varies widely from case to case. The quality of the writing is variable as well. The best cases are models of good statistical practice and of good pedagogy. On the other side of the ledger, in a case where the central issue is testing goodness of fit and where one of the stated objectives is "To provide students with a fundamental understanding of the statistical techniques necessary to deal with practical problems", a test of normality is relegated to the press of a button in a statistical package. This does not lead to a "fundamental understanding" especially when the obvious approach (estimate the mean and variance, group, and calculate a chi-square statistic) produces the wrong P-value! The authors of this case neither describe the algorithm for computing the test statistic nor comment on whether the package uses the correct procedure. The reader must use a critical eye.
A comment for editors of future volumes of case studies (and I hope they will appear): the format of the present volume makes large-scale photocopying by the instructor almost unavoidable in a course where the students are expected to analyze the data and present results without having seen the sections on "Analysis" and "Notes to the Instructor". Possible solutions might entail publishing separate student and instructor versions or including case background and instructions for students on the accompanying disks.
Despite the criticisms, the book provides a rich resource for teachers of statisticians and their students.
I will be using it this term in a course on statistical consulting.

Contents:
1. An introduction to forecasting
2. First principles
3. Evaluating forecast accuracy
4. Forecasting univariate processes
5. Monte Carlo techniques
6. Forecasting in cointegrated systems
7. Forecasting with large-scale macroeconometric models
8. A theory of intercept corrections: beyond mechanistic forecasts
9. Forecasting using leading indicators
10. Combining forecasts
11. Multi-step estimation
12. Parsimony
13. Testing forecast accuracy
14. Postscript

The impetus for this book came in 1991, when the second author acted as adviser to a House of Commons Select Committee on the Enquiry into Official Economic Forecasting in the U.K. Encountering little theory that could explain the systematic misforecasting of the 1980s, the authors took up the challenge of developing one. The result is a deep and thoughtful investigation of the nature of econometric forecasting.
A central observation is the inadequacy of constant-parameter stationary models to provide useful forecasts in the face of a reality which is non-constant, continually evolving and subject to structural breaks. The authors identify six critical elements of a forecasting framework: the nature of the data-generating process; the knowledge level concerning this process; the dimensionality of the system; whether the analysis is to be based on asymptotic or finite sample theory; the forecast horizon; the linearity or non-linearity of the system. In this framework they provide an exhaustive analysis of forecasting practice and forecasting error. The book presupposes experience in econometric modelling or time series forecasting and is written in an erudite style that will reward the advanced reader with new insights in these areas.

Contents:
1. Introduction and overview
2. Key features of economic time series
3. Useful concepts in univariate time series analysis
4. Trends
5. Seasonality
6. Aberrant observations
7. Conditional heteroskedasticity
8. Non-linearity
9. Multivariate time series
10. Common features

This book is intended as an introduction for time series modelling and time series forecasting with particular emphasis to economics and business. It contains all the standard time series models, ARMA models, ARCH models and GARCH models with discussion on tests for unit roots, a technique widely used in the econometric literature. A chapter on multivariate (linear) time series is included. Statistical tests for common features in two or more series is included in the last chapter. The techniques are well illustrated with several real time series, and the series analyzed are included in the book.
The book is well written and the professional time series forecasters will no doubt find the book useful.

Contents:
1. Concepts from insurance and finance
2. Probability distributions
3. Premiums and ordering of risks
4. Distributions of aggregate claim amount
5. Risk processes
6. Renewal processes and random walks
7. Markov chains
8. Continuous-time Markov models
9. Martingale techniques I
10. Martingale techniques II
11. Piecewise deterministic Markov processes
12. Point processes
13. Diffusion models

Readership: Graduate students in stochastic modelling, probability theory, statistics, actuarial science and financial mathematics

This book aims to make the basic concepts of stochastic modelling and insurance accessible in a comprehensive (and comprehensible) manner. It does this well. The opening chapter, which is worth reading on its own, provides a clear and concise overview of the book, and also motivates the necessity for the mathematical depth in the rest of the book. The last point is indicative of the fact that, even apart from its length, the book is not light reading.
In the preface the authors caution that "This book is not covering [sic] the statistical aspects of stochastic models in insurance and finance", but it does include some real examples and some numerical and algorithmic details. Proofs are given for all except well-known results. There are helpful detailed bibliographical notes at the ends of most sections. There are no exercises, but the authors stress that the book "has been conceived as a course text", and note that a Teacher's Manual is forthcoming. In my view the clarity and accessibility of the book will make it successful as a course text for advanced students.
I thought I detected some differences in style between the chapters (presumably written by different authors) but, apart from some idiosyncrasies in places, such as that illustrated above, the book is fluently written and a pleasure to read. I recommend it.

Contents:
1. Pricing by arbitrage
2. Martingale measures
3. The fundamental theorem of asset pricing
4. Complete markets and martingale representation
5. Stopping times and American options
6. A review of continuous-time stochastic calculus
7. European options in continuous time
8. The American option
9. Bonds and term structure
10. Consumption. Investment strategies

In the preface to the book, the authors note that the field of mathematical finance is rapidly expanding. Under that circumstance, it is not surprising that besides the appearance of many new journals and different societies in this field. In many countries actively working researchers and practitioners are trying to give an exposition of their vision of the mathematics of finance. A presentation of the corresponding material is necessary in order to get control over the subject connected with analysis and decisions on financial markets.
Authors of the book strictly divide their text into two parts: discrete-time framework (the first five chapters), and continuous-time framework (the second five chapters).
Elliott and Kopp are known as experts in stochastic calculus by their books Stochastic Calculus and Applications by Elliott [Short Book Reviews, Vol. 3, p. 7], and Martingales and Stochastic Integrals by Kopp [Short Book Reviews, Vol. 4, p. 41]. So, readers of their present joint book have a guarantee to obtain a qualified presentation of mathematics of the stochastic financial markets.

Contents:
1. Dynamics, stochastic models and uncertainty
2. Modelling: Markov chains and Markov processes
3. Random walk and stochastic differential equations
4. Jump processes and special problems
5. Memory, volatility models and the range process
6. Dynamic optimization
7. Numerical and optimization techniques

The long title of the book points at a wide circle of the presented topics and problems. The main aim of the author, C.S. Tapiero, who is well known by his papers and books as a specialist on stochastic optimization, was 'not providing one more book on stochastic control' (as A. Bensoussan writes in his Foreword to the book), but writing a book about 'Applications' of stochastic control theory to a booming field of finance and insurance engineering.
The author did not spend a lot of space and time on the basic concepts and theories in finance and insurance. His goal is another—to demonstrate how different concrete stochastic models and optimization ideas 'work' in an economical, financial and business environment, and in insurance, where dynamics and uncertainty play an essential role.
I found the book interesting for its many explanations, examples of different models, and concrete recommendations for control and management. It will be useful to students and also practitioners.

Contents:
1. Introduction
2. The binomial model
3. Stochastic integrals
4. Differential equations
5. Portfolio dynamics
6. Arbitrage pricing
7. Completeness and hedging
8. Parity relations and delta hedging
9. Several underlying assets
10. Incomplete markets
11. Dividends
12. Currency derivatives
13. Barrier options
14. Stochastic optimal control
15. Bonds and interest rates
16. Short rate models
17. Martingale models for the short rate
18. Forward rate models
19. Change of numéraire
20. Forwards and futures

Readership: Senior undergraduates and graduates interested in financial mathematics, quantitative analysts, financial
engineers

This book is one of the best of a large number of new books on mathematical and probabilistic models in finance, positioned between the books by Hull and Duffie on a mathematical scale. It is largely self-contained, including a pre-measure theory treatment of Brownian motion and the stochastic integral. The text then proceeds to continuous time no-arbitrage pricing for stock and exchange rate derivatives, bonds, forwards and futures. Although essentially mathematical, the author is generous with intuitive explanations and review; see for example the nice chapter on incomplete markets. This is a highly readable book and strikes a fine balance between mathematical development and intuitive explanation.

Contents:
1. Introduction
2. Optimization criteria
3. Finite horizon optimization
4. Infinite horizon discounted cost optimization
5. An inventory model
6. Average cost optimization for finite state spaces
7. Average cost optimization for countable state spaces
8. Computation of average cost optimal policies for infinite state spaces
9. Optimization under actions at selected epochs
10. Average cost optimization of continuous time processes
APPENDIX A : Results from Analysis
APPENDIX B : Sequences of Stationary Policies
APPENDIX C : Markov Chains techniques

Readership: Applied statisticians, operations research engineers, control engineers, communication engineers,
manufacturing engineers

This book combines a theoretical treatment of the optimal stochastic control of queueing systems with computational support. Source code for computer programmes is available off a website. The book treats finite, countable and infinite state spaces with average and discounted cost functions. The methods and programmes will be of interest to a wide range of readers, including those involved in manufacturing, communications and computing systems. The book is self-contained and contains all necessary background theory and illustrative examples. An excellent book for a wide range of readers.

Contents:
1.Introduction: Problems in environmental monitoring and assessment, by L.H. Cox, D. Nychka and W.W. Piegorsch
2.Modeling ozone in the Chicago urban area, by J.M. Davis, B.K. Eder and P. Bloomfield
3.Regional and temporal models for ozone along the Gulf Coast, by J.M. Davis, B.K. Eder and P. Bloomfield
4.Design of air-quality monitoring networks, by D. Nychka and N. Saltzman
5.Estimating trends in the atmospheric deposition of pollutants, by D. Holland
6.Airborne particles and mortality, by R.L. Smith, J.M. Davis and P. Speckman
7.Categorical exposure-response regression analysis of toxicology experiments, by M. Xie and D. Simpson
8.Workshop: Statistical methods for combining environmental information, by L.H. Cox
APPENDIX A :FUNFITS, Data Analysis and Statistical Tools for Estimating Functions. By D. Nychka, P.D. Haaland, M.A. O'Connell and S. Ellner
APPENDIX B: DI, A Design Interface for Constructing and Analyzing Spatial Designs.
By N. Saltzman and D. Nychka
APPENDIX C: Workshops Sponsored Through the EPA/NISS Cooperative Agreement
APPENDIX D: Participating Scientists in the Cooperative Agreement

The book consists of a few loosely connected papers (case studies) and each of them constitutes a separate chapter. The authors tried to balance a self-consistent description of the statistical techniques with actual environmental problems, and in most cases that resulted in the very sketchy description of both. The first two chapters present several rather mature statistical methods with occasional digression towards applications. Some of these methods are masked by the use of terms like "parametric models", "semi-parametric models", "singular value decomposition" instead of "regression models", "regression models with correlated observations", "principal components analysis". The very substantial chapter on design of monitoring networks gravitates to model-free approaches, which may work well for homogeneous regions. This chapter is complemented with an appendix on the corresponding software tools. The rest of the book has more applied character and describes a number of interesting applications.

Contents:
17. Estimation and sufficiency
18. Estimation: Maximum likelihood and other methods
19. Interval estimation
20. Tests of hypotheses: Simple null hypotheses
21. Tests of hypotheses: Composite hypotheses
22. Likelihood ratio tests and test efficiency
23. Invariance and equivariance
24. Sequential methods
25. Tests of fit
26. Comparative statistical inference
27. Statistical relationship: Linear regression and correlation
28. Partial and multiple correlation
29. The general linear model
30. Fixed effects analysis of variance
31. Other analysis of variance models
32. Analysis and diagnostics for the linear model

In 1943, The Advanced Theory of Statistics, Volume 1, by M.G. Kendall appeared, Volume 2 appearing in 1946. The first volume had five editions, the last appearing in 1952; Volume 2 had three editions, the last appearing in 1951 with second, third and fourth impressions. In 1958, the work was started again with three volumes, the first subtitled Distribution Theory and written jointly with A. Stuart. Three subsequent editions of Volume 1 appeared in 1963, 1969 and 1977. Volume 2, subtitled Classical Inference and Relationship appeared in 1961 with further editions in 1967, 1973 and 1979.
Sir Maurice Kendall passed away in 1983 and Professor Alan Stuart very recently, in 1998. Their association will remain with this influential and mammoth work.
This, the sixth edition of Volume 2, has been very much revised since the previous edition [Short Book Reviews, Vol. 11, p. 21]. The authors state in the preface to this volume:
"[This volume] differs from its predecessors in several key respects. At the beginning of the decade, we decided that The Advanced Theory of Statistics should revert to its earlier two-volume format. Thus, this edition has been retitled Volume 2A: Classical Inference and the Linear Model, to emphasize the traditional emphasis on the frequentist approach. The complementary Volume 2B: Bayesian Inference [Short Book Reviews, Vol. 15, p. 4] written by Tony O'Hagan, was first published in 1994, thereby providing an overall balance to the coverage of modern inference. Volume 3 has been phased out and replaced by a series of monographs forming Kendall's Library of Statistics; five of these titles have already appeared and several more are in preparation."
"Statistical inference appears in the first part of the book (Chapters 17–26) and the coverage of the linear model is in the second part (Chapters 27–32). These changes involved both the redesign of existing chapters and the creation of several new ones. Chapter 18 now covers both maximum likelihood and other approaches to inference, while Chapters 19–22 broadly correspond to Chapters 20–23 and 25 in the fifth edition."
The list of references has been made up-to-date by adding about three hundred new ones and discarding over one hundred. This new edition and further volumes and editions will be used by all statisticians for reference and will be admired and appreciated as they have been since Sir Maurice Kendall's first volume first appeared!

Contents:
1. Introduction: Statistical inference and decision-making
2. An illustration of the different approaches
3. Probability
4. Utility and decision-making
5. Classical inference
6. Bayesian inference
7. Decision theory
8. Other approaches
9. Perspective

The aim of this book remains what it was in previous editions: to explain and compare the different approaches to statistical inference without detailed attention to theory or applications. This aim continues to be achieved very successfully. The chapter headings remain the same, but an important change is made with the references, which are gathered together next to the Index so that it is now easy to use these together. There are fifty-six more pages than in the second edition, and the number of references has grown from about two hundred in the first edition to three hundred in the second and to six hundred in the third. The enormous growth in the number of references reflects the burgeoning interest in the subject areas. It has of necessity forced a shortening of discussions of individual papers.

Contents:
1. Planning a study
2. Sample surveys
3. Experimental trials
4. Data analysis
5. Reporting the results

Readership: People who have to organize or evaluate research studies involving human subjects, but who have little or no statistical knowledge

The aim of this short book is to provide an accessible and non-mathematical introduction to the principles of statistics. It is written in an interesting style, with the reader being instructed to do things: "...you should do this ..., you will need to do that... ."
Although the book is 217 pages long, each page is only 17 cm x 12 cm, so it is in fact very short. This is good, since it means it is more likely to be read by those who need it, and who will not be deterred at the prospect of a large tome.
In terms of the concepts the book seeks to cover, it is quite deep. It includes things such as the accuracy of measuring instruments, multistage samples, equivalence trials, interim analysis, generalized linear models, model selection, likelihood intervals, as well as the usual statistical staples. However, in terms of the depth at which each is covered, the book's length means it is necessarily very superficial. As an overview of the nature of modern statistical concerns and practice, then, the book is quite reasonable. However, as a manual to help those without statistical skills who find themselves responsible for organizing a study, I have reservations about its depth.

Contents:
Introduction: The paradigm of structural analysis
1. Social relationships and networks
2. Personal networks and local circles
3. Graph theory
4. Equivalence and cohesion
5. Social capital
6. Power and centrality
7. Dynamics
8. Multiple affiliations
APPENDIX: Matrix Operations
Hierarchical Clustering Procedures
Basic Relational Algebra
Structural Investigation Sampling
Galois Lattices and Hypergraphs
Software

Who do we know, who are our friends, who are our enemies? How sociable are different groups of people, in different geographical or economic contexts? What are the effects of social relations on our careers? These are the kinds of questions addressed by social network analysis, a research area whose literature has grown rapidly in the last two to three decades. This book is a welcome introduction to the subject, and complements well the larger reference work of Wasserman and Faust (1994); see Winship (1996). It is mostly very readable — an exception is the Introduction which is rather inaccessible to non-sociologists — and has interesting examples. The main methods used come from graph theory, the necessary elements of which are well described in Chapter 3. Statistical methods appear in the use of summary measures calculated from (typically large) relational graphs, in the application of standard cluster analysis techniques to identify groups of individuals with close relations, and importantly in the design of surveys or experiments to obtain data on social relationships.
Wasserman, S. and Faust, K. (1994). Social Network Analysis: Methods and Applications. New York: Cambridge University Press.
Winship, C. (1996). Review of Social Network Analysis and Applications by S. Wasserman and K. Faust. J. Amer. Statist. Assoc. 91, 1373–1374.

Contents:
1. What is probability?
2. Lotteries
3. Football pools, premium bonds
4. One coin, many games
5. Dice
6. Games with few choices
7. Waiting, waiting, waiting
8. Let's play best of three
9. TV games
10. Casino games
11. Bookies, the Tote, spread betting
12. This sporting life
13. Lucky for some – miscellanea
APPENDIX I : Counting
APPENDIX II : Probability
APPENDIX III : Averages and Variability
APPENDIX IV : Goodness-of-Fit Tests
APPENDIX V : The Kelly Strategy

Gambling manuals in English using probability arguments have been around since John Arbuthnot's 1692 translation of and addition to Christian Huygens' De Ratiociniis in Ludo Aleae. Over the past 300 years some of these manuals have been bad and some, like Arbuthnot's, have been very good and readable. The current book falls in the Arbuthnot tradition and is a delightful addition to this genre. The book is very well written in a conversational style with technical material inserted at a level intelligible to the layman. The appendices contain mathematical detail necessary to the probability and statistical methods used within the text. The core of the text is devoted to analyses of games in which there is an element of chance. In addition to covering the usual casino and lottery games, the author adds board games as well as game shows from television. The focus is on practical advice related to the strategy of play in a wide variety of these games. For example, there is easy-to-follow advice to improve a player's chances of winning in the board game Monopoly. And the advice comes with a well-argued justification. The focus on strategy is the reason why this book will be more appealing than most books on the analysis of games involving chance. It is the same reason why Hoyle's Short Treatise on the Game of Whist in the eighteenth century was very popular. For those in the academic lines of business, the book contains a wealth of examples that could be used in the classroom to illustrate elementary probability theory.

Contents:
1. Sets and events
2. Probability spaces
3. Random variables, elements, and measurable maps
4. Independence
5. Integration and expectation
6. Convergence concepts
7. Laws of large numbers and sums of independent random variables
8. Convergence in distribution
9. Characteristic functions and the central limit theorem
10. Martingales

This book is different from the classical textbooks on probability theory in that it treats the measure theoretic background not as a prerequisite but as an integral part of probability theory. The result is that the reader gets a thorough and well-structured framework needed to understand the deeper concepts of current day advanced probability as it is used in statistics, engineering, biology and finance. As the author states in his preface, the pace of the book is quick and disciplined. Yet there are ample examples sprinkled over the entire book and each chapter finishes with a wealthy section of inspiring problems.

Contents:
1. Recurrence relations
2. Power series expansions
3. Continued fraction expansions
4. Asymptotic expansions
5. Solution on nonlinear equations
6. Vector and matrix norms
7. Linear regression and matrix inversion
8. Eigenvalues and eigenvectors
9. Splines
10. The EM algorithm
11. Newton's method and scoring
12. Variations on the EM theme
13. Convergence of optimization algorithms
14. Constrained optimization
15. Concrete Hilbert spaces
16. Quadrature methods
17. The Fourier transform
18. The finite Fourier transform
19. Wavelets
20. Generating random deviates
21. Independent Monte Carlo
22. Bootstrap calculations
23. Finite-state Markov chains
24. Markov chain Monte Carlo

Although the author's intention was the creation of a textbook for high-level students, he has actually created a collection of very elegantly written essays on various topics from computational statistics. I found that the book is interesting and it makes enjoyable reading. Similar to essays in poetry, to which the author gravitates judging by some of his remarks, some parts of the collection are better written and more enjoyable than others (Chapter 12 on variations on the EM theme), and some of the latter need more mathematical rigour (Chapter 13 on convergence of optimization algorithms). Undoubtedly, the book will be a good complement to any statistical library.

Contents:
1. Probability theory
2. Fundamentals of statistics
3. Unbiased estimation
4. Estimation in parametric models
5. Estimation in nonparametric models
6. Hypothesis tests
7. Confidence sets

Readership: PhD level students requiring a measure-theoretical foundation in statistical theory

This text is intended for a course of two fifteen-week semesters with three lectures and two discussion hours per week. An introduction to measure theory is included, so the book is quite well-contained. Much is devoted to the classical theory of inference/decision found in other books like Lehmann's classics. An attractive feature is its inclusion of some new topics like the bootstrap and generalized estimating equations.
We see little Bayesian theory even though that theory collectively represents a major modern direction. Nor does its very technical nature leave room for discussion of fundamental principles such as conditionality or any of the wealth of associated paradoxes that add such richness to the discipline.
The book seems generally well written with lots of great exercises. However, it is challenging in places. For example, students will struggle to understand the Cornish-Fisher expansion on page 458 from the explanation given there. The reader is referred to another source for details. In another case we learn that loss functions can be determined with the help of utility analysis that is not otherwise mentioned.

Contents:
1. Cause and effect
2. Testing hypotheses
3. When the distribution is known
4. Descriptive statistics
5. Estimation
6. Power of a test
7. Categorical data
8. Experimental design and analysis
9. Multiple variables and multiple hypotheses
10. Classification and discrimination
11. Survival analysis and reliability
12. Which statistic should I use?

Readership: Experimental scientists, industrial statisticians, statistical consultants

Readers with a wide variety of backgrounds and interests should find this book illuminating and as a valuable introduction to data analysis using permutations, cross-validation and the bootstrap. The text provides a guide to the power, simplicity and versatility of resampling methods to characterize, review, report on, test, estimate and classify findings.

Contents:
PART I : Basic Concepts
1. Uses of sample surveys
2. The population and the sample
PART II : Major Sampling Designs and Estimation Procedures
3. Simple random sampling
4. Systematic sampling
5. Stratification and stratified random sampling
6. Stratified random sampling: Further issues
7. Ratio estimation
8. Cluster sampling: Introduction and overview
9. Simple one-stage cluster sampling
10. Two-stage cluster sampling: Clusters sampled with equal probability
11. Cluster sampling in which clusters are sampled with unequal probability
12. Variance estimation in complex sample surveys
PART III: Selected Topics in Sample Survey Methodology
13. Nonresponse and missing data in sample surveys
14. Selected topics in sample design and estimation methodology
15. Telephone sampling
16. Strategies for design-based analysis of sample survey data

Readership: Applied statisticians, social scientists, medical researchers, anyone requiring sample survey methods in their research

This is the third edition of the authors' original text called Sampling for Health Professionals. The emphasis remains non-mathematical with an abundance of examples and exercises taken from medical research. The presentation is particularly clear and the material is handled in a way that is simple to follow. Each idea is described in basic terms and illustrated with an example. The main formulae for each section are presented in a summary box that makes identification of an estimator and its standard error easy.
In this edition many new exercises and references have been included. There is an emphasis on illustrations utilizing the software systems Stata and SUDAAN (in conjunction with SAS), both of which are now readily available and have particularly attractive sample survey features. There has been a reorganization of some of the chapters so that more could be included on two-stage clustering methods. Probability proportional to size sampling is now presented in Chapter 11 with clustering methods using unequal probabilities. Part II is completed with an interesting chapter on computationally intensive variance estimation methods for complex surveys. One of the four more specialized chapters, making up Part III, is contributed by R.J. Casady and J.M. Lepkowski, who describe their extensive experiences of telephone sampling.
This is an excellent text for use either as a reference book for a whole range of survey methods or as a course text, where the instructor wishes to have lots of examples but does not require a mathematical development of all the estimation equations.

Contents:
1. Events and probability
2. Random variables
3. Expectation
4. Conditional expectation
5. Limit theorems
6. Simple random sampling
7. Unequal probability sampling
8. Linear relationships
9. Stratified sampling
10. Cluster sampling
11. Two-stage sampling

This book gives a mathematically rigorous treatment of estimation theory for survey sampling. The necessary background in elementary probability theory is developed in the early chapters before launching into derivations of expectations and variances for estimators under some standard sampling designs. Unlike many textbooks, all derivations are explicit and detailed. As a result, fewer topics are covered than appear in most texts, but they are covered thoroughly.
The approach to sampling theory taken in this book follows some of the classical sampling texts. Consequently, it differs from most modern texts. In both approaches there is a finite population of size N with fixed measurements γ1, γ2, ..., γN. In the more classical framework such as used in this book, the sample measurement, Yi obtained on the ith draw, is a random variable with expectation Y¯ , the finite population mean. Under this approach the expectation of the sample mean g¯ is obtained as an average of expectations of the sample measurements. In more modern approaches to sampling theory, what is considered as the random variable is s, the set of units chosen for the sample. The probability function P(s) defines the sampling design. The sample measurements remain explicitly identified as fixed values. My preference is for this latter approach, since it makes a clear distinction between random variation in the measurements and random variation in the selection process.

Contents:
1. Introduction – Einführung
2. Planning of experiments and surveys and the description of simple designs – Planung von Versuchen und Erhebungen und Beschreibung einfacher Anlagen
3. Design and analysis of completely randomised designs – Planung und Auswertung vollständig randomisierter Versuchsanlagen
4. Analysis of variance – Varianzanalyse
5. Regression analysis – Regressionsanalyse
APPENDIX A: Symbols – ANHANG A: Symbolik
APPENDIX B: Fundamentals in Statistics – Overview – ANHANG B: Vorausgesetzte Grundkenntnisse der Statistik
APPENDIX C: Matrices – ANHANG C: Matizen

Readership: Students and scientists with an elementary knowledge of statistics / Leserschaft: Studenten und Wissenschaftler mit Grundkentnissen in Statistik

This is a unique book; in fact it could be considered to be two books. The left-hand pages comprise the English version; the right-hand pages the German version. The authors have written a basic text which covers the subject matter well. The Appendices will be a help to those with an inadequate background in statistics. Besides having a text in the design and analysis of experiments and surveys, this would be an excellent way to learn or practice the two languages and discover the equivalent statistical words.
Dieses ist ein einzigartiges Buch: eigentlich besteht es aus zwei Büchern. Auf der linken Seite steht die englische und auf der rechten Seite die deutsche Version. Die Autoren haben einen grundlegenden Text geschrieben, der das Thema ausführlich behandelt. Die Appendices werden denjenigen nützlich sein, deren Wissen in Statistik noch nicht ausreichend ist. Abgesehen davon, dass man einen Text über Planung und Auswertung von Versuchen und Erhebungen vor sich hat, bietet dieses Buch eine ausgezeichnete Möglichkeit, zwei Sprachen zu lernen und zu üben, sowie das entsprechende statistische Vokabular zu entdecken.

Contents:
1. Principles and techniques
2. Planning experiments
3. Designs with one source of variation
4. Inferences for contrasts and treatment means
5. Checking model assumptions
6. Experiments with two crossed treatment factors
7. Several crossed treatment factors
8. Polynomial regression
9. Analysis of covariance
10. Complete block designs
11. Incomplete block designs
12. Designs with two blocking factors
13. Confounded two-level factorial designs
14. Confounding in general factorial experiments
15. Fractional factorial experiments
16. Response surface methodology
17. Random effects and variance components
18. Nested models
19. Split-plot designs

This book is based on courses in experimental design which the authors have taught for the past ten years. The courses they give must be very good indeed.
Believing that this subject is learnt best by experience, most designs are introduced by a simple experiment, the aim of which can be readily understood and which could be easily performed by any number of the class. This enables the students to consider aspects of the planning of an experiment that go beyond the formal techniques of design and analysis. The text contains a number of suggestions for such simple experIments. Numerous other, more sophisticated experiments are discussed in the text and in the exercises which follow each chapter.
A checklist of points to be considered when planning an experiment is given. The book is characterized by very detailed, clear explanations and comprehensive treatment of the subject matter. Matrix algebra is avoided and mathematics is kept to a minimum, making the material accessible to a wide audience. SAS is used for the data analysis and complete programs are given both for the analysis of variance and model checking.
Taguchi methods are not introduced as a special topic. Instead the authors introduce "noise" factors and analyze the variability of the response. This reviewer would have some reservations about the overuse of multiple comparisons and the testing of main effects when interactions are present, but this is a book from which I would be happy to teach. It is highly recommended.

Contents:
1. Introduction
2. Fractional plans and orthogonal arrays
3. Symmetric orthogonal arrays
4. Asymmetric orthogonal arrays
5. Some results on nonexistence
6. More on optimal fractional plans and related topics
7. Trend-free plans and blocking
8. Some further developments
APPENDIX A.1 : Hadamard Matrices
APPENDIX A.2 : Difference Matrices
APPENDIX A.3 : Selected Orthogonal Arrays

This rigorous and authoritative book is likely to become an important reference for researchers and postgraduate students working on the design of experiments. Its approach requires some background in matrix algebra and linear models. Examples are used to illustrate ideas throughout; exercises are provided at the end of each chapter, but no solutions are given. The authors offer the reader an up-to-date and cohesive account of an area of mathematical fascination and considerable practical importance. An impressive long list of references is included. I have long treasured the monograph by A. Dey (1985) on orthogonal fractional factorials. This new book goes far beyond that text in the range of topics covered and the generality of the results presented.

1. Introduction
2. A practical guide to tree construction
3. Logistic regression
4. Classification trees for a binary response
5. Risk-factor analysis using tree-based stratification
6. Analysis of censored data: Example
7. Analysis of censored data: Concepts and classification methods
8. Analysis of censored data: Survival trees
9. Regression trees and adaptive splines for a continuous response
10. Analysis of longitudinal data
11. Analysis of multiple discrete responses
12. Appendix

Recursive partitioning is a statistical technique that forms the basis for two classes of nonparametric regression methods: classification and regression trees and multivariate adaptive regression splines. This book presents a very good summary of the methodological and theoretical underpinnings of recursive partitioning and illustrates its concepts well using biomedical examples. A free program (RTREE) from a web site was used for the analysis of the data of the examples.

Contents:
1. Introduction
2. Elements of the theory of probability metrics
3. Method of characteristic functions: Berry-Esseen theorem
4. Method of compositions in the one-dimensional case
5. Method of compositions in the multidimensional case
6. Convergence rate estimates in weak metrics
7. Estimation of uniform distances in l2
8. Estimation of derivatives and measures
9. Lower bounds for uniform metrics

The importance of the central limit theorem in probability theory and statistics can hardly be over-estimated. Closely related to it is the topic of normal approximations, which, since its origin around 1900 by Lyapunov, has been the source of some very fruitful research. Writing in the language of probability metrics and multidimensional Euclidean space as well as Hilbert space, the author gives the subject a modern look, many times pointing to the future by formulating new questions and problems.

Contents:
1. Introduction and coverage
2. Preliminaries
3. Elementary theory of rank tests
4. Selected rank tests
5. Computation of null exact distributions
6. Limiting null distributions
7. Limiting non-null distributions
8. Asymptotic optimality and efficiency
9. Rank estimates and asymptotic linearity
10. Miscellaneous topics in regression rank tests

Many of us received part of their advanced statistical education through the classical book Theory of Rank Tests by J. Hájek and Z. Šidák, published in 1967. J. Hájek died seven years later at the age of 48, but the subject of this book continued to generate further research, not only in the Prague School but also world-wide. His co-author Z. Šidák together with P.K. Sen now publish a second edition of this influential book. The original text has been kept as far as possible, but several updates and supplements are now provided, taking into account some major developments during the past thirty-five years. The last two chapters are newly added. Chapter 9 gives introductory ideas on R-estimates of location and regression, asymptotic linearity of rank statistics in regression parameters and rank estimation of regression parameters. Chapter 10 adds some further topics on aligned rank tests, regression rank scores and robustness.
Some of the other chapters of the original text have been updated with (mostly) small new sections on, for instance, martingales (5 pages), statistical functions (6 pages), rank tests for censored data (16 pages), multivariate rank tests (9 pages), weak convergence in D[0,1] (2 pages), functional limit theorems (10 pages), non-contiguous alternatives (5 pages), Bahadur efficiency, Hodges-Lehmann deficiency and adaptive rank tests (28 pages). The updated bibliography at the end covers the original references, but some eight new ones are added now, of which half are authored or co-authored by P.K. Sen.

Contents:
1. Introduction
2. Probability theory
3. Distributions of climate variables
4. Concepts in statistical inference
5. Estimation
6. The statistical test of a hypothesis
7. Analysis of atmospheric circulation problems
8. Regression
9. Analysis of variance
10. Time series and stochastic processes
11. Parameters of univariate and bivariate time series
12. Estimating covariance functions and spectra
13. Empirical orthogonal functions
14. Canonical correlation analysis
15. POP analysis
16. Complex eigentechniques
17. Specific statistical concepts in climate research
18. Forecast quality evaluation

Readership: Climatologists, meteorologists, atmospheric scientists, statisticians, other experimental scientists

Written by a meteorologist and a climatologist, this book aims at explaining the statistical methodology commonly used in applications of climatological research. Despite occasional idiosyncratic use of statistical terminology (e.g. relative likelihood on p. 19), the book has a modern feel; bootstrap, extreme value analysis (including peaks over threshold), threshold autoregressive models (although they invented a new name for them, namely RAMs), outliers, cross-validation and others are featured or at least touched upon. To the statisticians, the most valuable aspects of the book are the climatological/meteorological examples with their interpretations, which are mildly marred by the absence of the data listing/data disk.

Contents:
1. Basic concepts
2. Common lifetime models
3. Model selection
4. Model fitting
5. Repairable systems
6. System reliability
7. Models for functions of random variations
8. Maintenance strategies
9. Life testing and inference
10. Advance models

This book treats the modelling of both component and system lifetimes to a level suitable for masters degrees in engineering or mathematical sciences. It also provides a reference text for engineers with a practical interest in reliability engineering. Some knowledge of undergraduate mathematics is necessary, though there is an Appendix which covers algebraic and calculus methods relevant to reliability analysis. Two short case studies, which illustrate material covered in the book, are given in the final chapter. Only the software package MINITAB is used to illustrate the statistical analysis. There are about seventy references to published books and papers.

Contents:
1. Introduction
2. Fundamentals
3. Quasi-reversibility
4. Networks of quasi-reversible queues
5. Networks with exponential service times
6. Multiple customer classes and arbitrary service times
7. Networks with batch services and negative signals
8. Batch arrivals, batch services and concurrent movements
9. State-dependent and history-dependent transitions
10. Local balance revisited
11. Characterization of product form and stability issues
12. Discrete time networks
APPENDIX A : Basic Concepts in Probability Theory
APPENDIX B : Markov and Poisson Processes
APPENDIX C : Brouwer's Fixed Point Theorem

Readership: Graduate students and researchers in engineering, operational research and probability

The authors consider queueing networks with additional signals. These signals generalize classical queueing networks by triggering simultaneous events at one or more queues. For example, they can be used to model batch arrivals or services, deletion, splitting or coalescence of customers. This book focuses in particular on when such networks are quasi-reversible and so possess product-form stationary distributions. The theory is both mathematically appealing and well motivated by applications. The exposition is refreshingly clear, although the notation sometimes becomes rather dense. The book is written as a textbook for Masters or Ph.D. students who have a background in stochastic processes but not necessarily queueing theory. It would be accessible to such an audience. However, the prospective teacher should note that this is not a textbook on standard queueing theory: it is much more specific than that, and is based in large part on recent work by the authors.
For this reason, I imagine that it will be more useful to researchers.

Contents:
1. Introduction to neural network analysis
2. The terminology of neural network analysis
3. The backpropagation model
4. Alternative network paradigms
5. Methodological considerations
6. Neural network software
7. Example: Analysing census data with neural connection
8. Conclusion

This is more in the nature of an overview of neural networks than an introduction. It is suitable for someone who already has had some exposure to the subject and would like to draw various threads together. The treatment is too sketchy to allow a reader unfamiliar with neural networks to obtain a clear picture of this important topic. A simple example would have been helpful early on to fix ideas, instead the reader is treated to generalities for forty pages (largely unrelieved by diagrams, or other visual aids) before encountering a concrete example that would enable the reader to gain some idea of the essential mechanism of the method.
Later chapters are centred around an SPSS module and its use in applying neural networks in the context of some census data.

Contents:
1. Introduction
2. Basic reliability theory
3. Stochastic failure models
4. Availability analysis of complex systems
5. Maintenance optimization
APPENDIX A : Background in Probability and Stochastic Processes
APPENDIX B : Renewal Processes

Both authors of this monograph are active researchers in this subject and they now present a unifying state-of-the-art in a number of areas of the enormously broad field of reliability theory. Their treatment is based on the modern theory of stochastic processes such as point processes, renewal processes and (semi-)martingale theory. Many readers will appreciate the two Appendices (40 pages in total), containing some of the necessary theoretical background material. This advanced stochastic process background is not needed for the first two chapters which serve as a motivating introduction with examples and two real-life situations from engineering. The main body of the book (Chapters 3 to 5) is meant for researchers in the field. It could also be used for a graduate course, but there are, however, no exercises in the text.

Contents:
1. Derivative background
2. Probability background
3. Stochastic processes in discrete time
4. Mathematical finance in discrete time
5. Stochastic processes in continuous time
6. Mathematical finance in continuous time
7. Incomplete markets
8. Interest rate theory
APPENDIX A : Hilbert Space
APPENDIX B : Projections and Conditional Expectations
APPENDIX C : The Separating Hyberplane Theorem

Readership: Students and teachers interested in a mathematical introduction to mathematical finance

Almost six years have passed since the appearance of the French edition of Lamberton-Lapeyre, one of the first texts aiming at giving a mathematically rigorous treatment of mathematical finance. Please note: the adjective mathematical is important. Indeed, mathematicians have taken finance on board and transformed it to a part of their (our) field. By now finance "applications" yield interesting examples of the usefulness of modern probability, especially of stochastic calculus. The fact that real applications have wandered off seem to bother only few. For that reason, one has to be critical on "yet another one". As far as finance goes, the present book does not contain surprises. It is, however, written in an extremely smooth and pedagogically sound way. It presents the basic mathematics underlying (a part) of finance in a most readable form. There are enough more applied comments to indicate to the reader that there is something on the other side of the hill.
I personally enjoyed browsing through it. The book yields a very nice text on which to base an introductory course for mathematicians. I am convinced that it will find its rightful place among the many competitors. Those who are looking for a text between the successful Baxter-Rennie [M. Baxter and A. Rennie, Short Book Reviews, Vol. 17, p. 31] and the mathematically more advanced texts of Karatzas-Shreve [I. Karatzas and S.E. Shreve, Short Book Reviews, Vol. 8, p. 45], or Musiela-Rutowski [M. Musiela and M. Rutkowski, Short Book Reviews, Vol. 18, p. 4] may try it. They will like it.

Contents:
PART I : Facts. Models
1. Main concepts, structures, and instruments. Aims and problems of financial theory and financial engineering
2. Stochastic models. Discrete time
3. Stochastic models. Continuous time
4. Statistical analysis of financial data
PART II : Theory
5. Theory of arbitrage in stochastic financial models. Discrete time
6. Theory of pricing in stochastic financial models. Discrete time
7. Theory of arbitrage in stochastic financial models. Continuous time
8. Theory of pricing in stochastic financial models. Continuous time

Readership: Students and researchers interested in the probabilistic and statistical foundation of finance

This is a remarkable text, spanning fairly introductory material on how financial markets are organized and function, to advanced theory on non-standard probabilistic models relevant for financial modelling. The basic properties and availability of statistical data within finance are discussed fairly in detail. Special attention is given to such non-standard topics as non-linear dynamics, chaos, self-similarity, operational time and high-density (tick-by-tick) data. The result is an exciting text containing a huge amount of interesting material on modern stochastic finance. Especially the young (novice) researcher in the field will find it a very useful basis of results essential for further research. The set of references is impressive and the level of writing is clear and pedagogically sound. The content makes the book stand out among its many competitors: there are many more comments/results on the applied modelling aspects of finance than the by now common mathematical introductions, also compared to the more applied texts, a much more in-depth treatment of a very wide and encompassing range of stochastic models is given. In summary: a text to be recommended warmly.

Contents:
1. Introduction
2. The role of modeling in the risk assessment process
3. Identifying risk through hierarchical holographic modeling
4. Decision analysis
5. Multiobjective trade-off analysis
6. Defining uncertainty and sensitivity analysis
7. Bayes' theorem and the prediction of chemical carcinogenicity
8. Risk of extreme events and the fallacy of expected value
9. Multiobjective decision-tree analysis
10. Multiobjective risk-impact analysis method
11. Statistics of extremes: Extension of the PMRM
12. Statistics of extremes: Sensitivity to partitioning
13. Statistics of extremes: Generalized quantification risk
14. Fault trees
15. Multiobjective statistical method
16. Software risk management
APPENDIX: Optimization Techniques

Readership: Management scientists, civil/system engineers, financial engineers, statisticians

Risk management is a topical subject, perhaps due to the recent fascination with the financial market. However, risk study has a much longer history as actuaries, civil engineers, reliability engineers and many others will testify. Before a common core of essential mathematical and statistical tools is identified, it is natural that different authors will interpret risk management differently. This book is essentially a civil/system engineer's approach. However, many of the tools described in the book should be of wider applicabilities. The level of mathematics and statistics is not much higher than what would normally be covered in a good British first-year undergraduate statistics programme.

Contents:
1.Stochastic spatial-temporal models for rain, by D.R. Cox and V.S. Isham
2.On scaling theories of space-time rainfall: Some recent results and open problems, by E. Foufouda-Georgiou
3.Modelling of drop-size distribution and its applications to rainfall measurements from radars, by M. Porra, D. Sempa Torres and J.D. Crutin
4.Spatial channel network models in hydrology, by B.M. Troutman and M.R. Kartinger
5.Some mathematical aspects of rainfall, landforms and floods, by V.K. Gupta and E.C. Waymire
APPENDIX A: Efficient extraction of river networks and hydrologic measurements from digital elevation data, by S.D. Perkham

This book starts with a concise overview of modelling philosophy, parameter estimation and fitting results in hierarchical (Poisson-cluster) approaches to rainfall modelling. The scaling approach is then given a clear presentation focussing upon wavelet transform analysis of time-series and links between spatial scaling characteristics and physical parameterizations. A wide-ranging discussion of approaches to river basin network modelling is followed by an innovative mathematical framework to the modelling of the hydrological cycle based upon a general difference equation of mass conservation and integrating (multi-)scaling approaches to rainfall, networks and peak flows. The measurement problem is also reviewed with a paper addressing the important issue of drop-size distribution underpinning the use of radar data and another explaining the use and scope of digital elevation model data in hydrology. This book thus presents an exciting review of developments in stochastic hydrology (with a helpful index) and includes many useful references.

Contents:
1.Introduction
2.Active shape models
3.Spline curves
4.Shape-space models
5.Image processing techniques for feature location
6.Fitting spline templates
7.Pose recovery
8.Probabilistic models of shape
9.Dynamical models
10.Dynamic contour tracking
11.Learning motion
12.Non-Gaussian models and random sampling algorithms
APPENDIX A: Mathematical
APPENDIX B: Stochastic Dynamical
APPENDIX C: Further Shape-Space Models

Readership: Undergraduate and graduate students, researchers in computer graphics and statistics

The theory of computer vision has set its sights on the ambitious goal of designing machines that can interpret objects within an image in the way that human vision can organize and interpret a retinal image. Because interpreting an image involves the extraction of information from a noisy picture, the use of statistical methods for the fitting of deformable templates has become standard. The "active contours" of the title of this book are simplified curves that can be moved around an image to highlight its important one-dimensional features, such as the line of a mouth or eyebrows. Of primary interest is the encoding of shape information through a shape space, which is not to be identified with the shape spaces of D.G. Kendall and other authors.
This lovely book finds a pleasant balance between exposition and technical detail, and can be read by the advanced undergraduate. A key feature is the addition of a supporting website at
http://www.robots.ox.ac.uk/~contours/
which offers MPEG movies of the dynamic feature of the algorithms. You can watch the active contours track a moving image of a head among other things. Seeing is believing.

Contents:
1.Introduction
2.Preliminaries: Size measures and shape coordinates
3.Preliminaries: Planar Procrustes analysis
4.Shape space and distances
5.General Procrustes methods
6.Shape models for two-dimensional data
7.Tangent space inference
8.Size-and-shape
9.Distributions for higher dimensions
10.Deformations and describing shape change
11.Shape in images
12.Additional topics

Size and shape have long been key concepts in biology, and quantitative developments have ranged from D'Arcy Thompson's classic work on transformation grids to modern methods of allometry and morphometry. Fresh impetus was provided in the 1970s and 1980s through pioneering work by Fred Bookstein and David Kendall which used "landmarks", i.e. key points of correspondence located on each object, as a basis for quantifying shape.
Shape being invariant under translation, rotation and reflection of the object, Procrustes analysis on their landmarks underlies the measurement of "distance" between two shapes. This leads on to the definition of a variety of shape spaces, with their concomitant properties, and shape models for (predominantly two-dimensional) landmark data. Inferential techniques then come from adaptations of multivariate methods to these spaces and models. The resultant methodology involves mathematics of considerable sophistication, including complex analysis and differential geometry, but is of great practical value.
The authors of this book have themselves contributed substantially to the topic, and now provide a comprehensive, logical and lucid exposition of all the main features. It is an invaluable source of information for anyone interested in this fascinating area.

Contents:
1. Introduction
2. Quality systems in manufacturing and service provision
3. Calibration procedures
4. Instrument classification, characteristics, and choice
5. Sources of measurement error
6. Transmission and processing of measurement signals
7. Measurement signal recording and data presentation
8. Practical implementation of measurement and calibration procedures
9. Reliability in manufacturing systems
10. Software quality metrics
11. Statistical process control
12. Product sampling and testing
13. Temperature calibration
14. Pressure calibration
15. Mass, force, and torque calibration
16. Dimension calibration
17. Volume flow rate calibration
18. Calibration of miscellaneous parameters
APPENDIX 1 : Fundamental and Derived SI Units
APPENDIX 2 : Imperial-Metric-SI Conversion Tables
APPENDIX 3 : Dynamic Characteristics of Instruments
ÁPPENDIX 4 : Concepts of Probability
APPENDIX 5 : Curve Fitting by Regression Techniques
APPENDIX 6 : Typical Structure of a Quality Manual

Readership: Engineering personnel concerned with quality systems, and management personnel and company directors who oversee quality systems

Measurement and calibration procedures are an essential component of a quality control system, both at intermediate production stages and for inspecting the final product. This book describes the measurement and calibration requirements necessary to achieve ISO 9000 registration, which is an international standard giving what are regarded as minimum acceptable standards for quality system management. The book is unusual in that it covers both quality control system management and quality management techniques together.
The first three chapters provide a general introduction to quality systems and the ISO 9000 standard. Chapters 4 to 12 provide a high level discussion of the nature of physical measuring instruments, the source of measurement errors, issues of practical implementation and reliability, and statistical quality control and sampling. The remaining chapters then look in detail at different kinds of physical measurement.
This is not a book for the general statistician, but anyone involved in quality control in manufacturing will find it valuable. It is clearly written and, as someone interested in the general principles of measurement, I found it enlightening to read about some of the specifics.

PART I : Introduction
1. Introduction to quality
2. History of the quality movement
3. The value of implementing quality
4. People and quality
5. Products, processes, and quality
6. Exploring the meaning of quality
PART II : A Process View of Quality
7. The production process
8. Creating a production process
9. The specification process
10. The design process
11. The create process
12. The examine process
PART III: Management Issues in Achieving Quality
13. The quality system
14. Establishing a culture for quality
15. Managing quality
16. Quality and people management
PART IV: Stabilizing Quality
17. Stabilizing the quality system
18. Managing by facts: Data gathering and data classification
19. Data analysis and data presentation
20. Statistical techniques for achieving quality
PART V: Improving Quality
21. Quality improvement
22. Quality improvement initiatives
23. Quality innovation
PART VI: Conclusion
24. Optimism for the future: Quality awards and success stories
APPENDIX: Case Studies

Readership: Young students who seek a first exposure to the business principles that underpin the management of quality improvement in industry; practitioners who seek a digest of the entire field in a single volume

Contents:
PART I : Quality Problems and Problem-Solving Strategies
1. Introduction and outline of the book
2. Detecting and prioritizing problems
3. Problem-solving strategies
4. Group-based problem solving
5. The reward structure: The human side of problem-solving
PART II : Management Based on Facts: The Importance of Data and Data Analysis
6. Measurements and their importance for quality
7. Analysis of information: Graphical displays and numerical summaries
8. Modeling variability and uncertainty: An introduction to probability distributions
9. Sample surveys
10. Statistical inference under simple random sampling
11. Acceptance sampling plans
PART III: Process Stabilization: Making Processes Predictable
12. Statistical process control: control charts
13. Process capability and pre-control
PART IV: Improvement Through Designed Experiments
14. Principles of effective experimental design
15. Analysis of data from effective experimental designs and an introduction to factorial experiments
16. Taguchi design methods for product and process improvement
PART V: Other Useful Statistical Techniques
17. Regression analysis: A useful tool for modeling relationships

Readership: Students who seek an introduction to statistical and non-statistical problem-solving tools in current use in industry

Together, these two books are a comprehensive treatment of the twin requirements of quality improvement in business and industry — the managerial infrastructure that nurtures continual improvement on the one hand, and the problem-solving tools that can help achieve it on the other. The books share a common authorship. Dr. Burrill takes the lead in the management book (Achieving Quality Through Continual Improvement), while Professor Ledolter takes the lead in the "tools" book (Statistical Quality Control). The books are aimed at college students who are encountering these topics for the first time, typically in a course on quality in an undergraduate business or engineering curriculum. The books can be used together or separately, depending on the emphasis of the course.
The management book gives a friendly introduction to the vast array of quality principles in the current literature without playing favourites. Coverage of each topic is consequently very brief. Examples and anecdotes help get the point across without pedantry, and many of the classic case studies from the early days of modern quality in the U.S. are included.
The tools book offers a unified problem-solving strategy that employs non-statistical as well as statistical methods. The treatment is elementary, and no prior background in statistics is needed. Nice project ideas and exercise questions are given for each section in both books.
There are surprisingly few graphs in the "business" portions of either book. It is not until the discussion of problem-solving tools that the customary diagrams appear. Students may at first be daunted by the absence of pictures in the first several hundred pages of text, but the authors' narrative style is eminently accessible. The long introductory chapters are in fact vital in conveying the flavor of the business environment to young students, who most likely will have had no personal experience with it.
These two books would be fine core texts for an introduction to the quality business.

Contents:
1. Multivariate linear regression
2. Reduced-rank regression model
3. Reduced-rank regression models with two sets of regressors
4. Reduced-rank regression model with autoregressive errors
5. Multiple time series modeling with reduced ranks
6. The growth curve model and reduced-rank regression methods
7. Seemingly unrelated regression models with reduced ranks
8. Applications of reduced-rank regression in financial economics
9. Alternative procedures for analysis of multivariate regression models

Readership: Those familiar with basic matrix theory, plus at least limited exposure to multivariate statistics

This well-written and well laid-out monograph deals with multivariate (m Y's depending on n X's) linear models which are of reduced rank, that is, use fewer parameters than mn. In fact, for the model
Y = CX + e, rank (C) = r £ min(m,n) is initially assumed. Later, two sets of regressors are considered, one set with reduced rank parameters, one with full rank parameters, as well as the case where both sets have reduced ranks with distinct structures. Applications in the areas of time series, growth curves, economics and finance are subsequently discussed. Several numerical examples are presented to illustrate the analysis of multivariate data sets using reduced-rank methods. A seven-page final chapter sums up other and related approaches. There are two hundred and five references. The cover is soft-back but sturdy. This text is a must for the library and would be excellent for a seminar course.

Contents:
Introduction
1. Auxiliary results
2. First properties of estimators
3. Asymptotic expansions
4. Nonstandard problems
5. The change-point problems
6. Nonparametric estimation

Readership: Researchers and advanced postgraduate students in statistics and probability

Poisson process models are applied in numerous fields; the author discusses examples from nuclear medicine, optical detection, auditory electrophysiology, seismology and other areas of science. The most appropriate audience for this book is the theorist, however, interested in asymptotic properties of estimators. Supposing that the intensity function of an inhomogeneous Poisson process depends on a finite number of parameters, the author focuses on maximum likelihood, Bayes, and minimum distance estimators for these parameters. The asymptotics are considered in terms of expanding sample size, as well as "small sample" asymptotic series expansions of the estimators and their distribution functions. A particularly appealing aspect is the inclusion of results on nonstandard problems. These include misspecification of the parametric family for the intensity measure; nonidentifiability; and intensity functions with jumps. Nonparametric estimates are also discussed. Overall, the material is clear and nicely motivated; the coverage seems thorough and the results valuable.

Contents:
0. Basic prerequisite knowledge
1. Fitting a straight line by least squares
2. Checking the straight line fit
3. Fitting straight lines: Special topics
4. Regression in matrix terms: Straight line case
5. The general regression situation
6. Extra sums of squares and tests for several parameters being zero
7. Serial correlation in the residuals and the Durbin-Watson test
8. More on checking fitted models
9. Multiple regression: Special topics
10. Bias in regression estimates, and expected values of mean squares and sums of squares
11. On worthwhile regressions, big F's, and R2
12. Models containing functions of the predictors, including polynomial models
13. Transformation of the response variable
14. "Dummy" variables
15. Selecting the "Best" regression equations
16. Ill-conditioning in regression data
17. Ridge regression
18. Generalized linear models (GLIM)
19. Mixture ingredients as predictor variables
20. The geometry of least squares
21. More geometry of least squares
22. Orthogonal polynomials and summary data
23. Multiple regression applied to analysis of variance problems
24. An introduction to nonlinear estimation
25. Robust regression
26. Resampling procedures (Bootstrapping)

Now in its third edition, this book covers a wide range of regression techniques, leading the student from analyzing a straight-line fit using a pocket calculator through to modern regression techniques, including robust regression and resampling methods.
The book includes a useful diskette containing the data used throughout the text. There is a wide range of exercises with mostly full or partial solutions, in addition to some excellent true/false questions.
For users of regression analysis, either in teaching or for reference, one needs to look no further than this text.

Contents:
1. Exponential family
2. Exponential family nonlinear models
3. Geometric framework
4. Some second order asymptotics
5. Confidence regions
6. Diagnostics and influence analysis
7. Extension

Readership: Those wishing to stretch beyond generalized linear models and
nonlinear least squares model

The author of this inviting volume is based at China's Southeast University in Nanjing. The general models discussed here, which have both a link function and a general model form have generalized linear models and normal nonlinear regression models as special cases. "This is a theoretical book." (Preface) Some data are discussed, however, but (annoyingly) one must look them up elsewhere; see, for example, pp. 27, 28, 111 and 120. The writing style is excellent and the presentation is very clear throughout. The book is a must for the library and a superb resource for seminar-type classes and researchers. This book is recommended.

Contents:
PART I : Comparing Partial Orders Among Stochastic Matrices
1. Introduction
2. Notation and definitions
3. Generalizations of classical channel comparisons
4. Degradation is the same as increasing density
5. Shannon's inclusion implies smaller capacity
6. A simple case: Matrices A and B have only two columns
7. Open problems
PART II : Divergence and Contraction Coefficients
1. Introduction, definitions, and notation
2. A generalization of an inequality of Dobrushin
3. The divergence
4. Divergence between images of measures via Markov kernels. Contraction coefficients
5. A particular case: At most countable spaces
6. Behaviour of φ → ηφ(T) for a fixed Markov kernel T
7. Applications of global divergence to comparison of experiments
8. History of the problem

The assessment of diversity, or mutual distance, between measures in ordered sets of probability measures is the subject of this book. The principal motivation comes from information theory where the sets of measures characterize a communication channel and, in broad terms, greater diversity among the measures leads to more informative outputs from similar inputs. PART I of the book, which is concerned with finite sets of measures, is due to all three of the named authors. PART II, which is concerned with arbitrary measurable spaces, is due to the third-named author.
The book begins with an introductory chapter in which a sketch is given of how the framework discussed in the book is of relevance to various applications in information theory, statistics, economics and population genetics. The remainder of the text focuses on mathematical discussion.

Contents:
1. Introduction
2. Inferential data analysis for simple experiments
3. One factor designs
4. One factor blocking designs
5. Factorial experimental designs
6. Hierarchical designs
7. Two-level factorial designs
8. Two-level fractional factorial designs
9. Two-level orthogonal arrays
10. Taguchi methods
11. Response-surface methods

Readership: Undergraduate students of statistics, engineers, experimental scientists

The layout of this book, which covers what may be termed standard experimental designs, makes it useful both as a textbook and as a reference for the practitioner. Mathematics is kept to a minimum and each design is illustrated by detailed example.
An attractive feature of this book is its emphasis on the complete process of design and analysis of experimental data. Starting from the initial protocol, which includes both the choice of design and sample size calculations, the authors take the reader through each step: the checking of the data before analysis begins; the writing of a SAS or Minitab program to obtain the ANOVA table and estimates of effects and contrasts; the checking of the model assumptions by diagnostic tests; to the final conclusions.
Data transformations or non-parametric tests are suggested to handle cases where the assumption of normality fails. Both fixed and random effects are discussed. A set of problems illustrates and amplifies the subject matter of each chapter.
A thoughtful chapter on Taguchi's methods, while acknowledging the contribution his ideas have made by introducing experimentation into areas where it had never been used before, shows that equivalent and often better results can be achieved by classical methods of experimental design.
All in all, there is much to recommend in this book. However, this reviewer takes serious issue with the recommendation that the SAS "Type III Sums of Squares" is the preferred ANOVA table for the analysis of unbalanced factorial designs. This is an issue that should be debated much more vigorously by the statistical community at large. It is a pity to see such advice being given in a book, which on its other merits, is sure to be widely read and used.

Contents:
1. Introduction
2. Basic statistical concepts
3. Basic design consideration
4. Randomization and blinding
5. Designs for clinical trials
6. Classification of clinical trials
7. Analysis of continuous data
8. Analysis of categorical data
9. Censored data and interim analysis
10. Sample size determination
11. Issues in efficacy evaluation
12. Safety assessment

The back cover of this book describes it as a unique, unifying treatment for statistics and science in clinical trials and indicates that it integrates the statistical and clinical disciplines. The authors would appear to have experience in clinical trials, especially from the pharmaceutical perspective, and there is some reasonable discussion, from this perspective, of issues which arise in clinical trials.
Personally, I wonder if the book achieves one of its aims, that of minimizing the mathematics to make the statistical content accessible. The choice of statistical topics and the amount of material on some topics might also be questioned. The references are not comprehensive. For example, the book length treatment of cross-over trials by S.J. Senn [Cross-Over Trials in Clinical Research, Short Book Reviews, Vol. 13, p. 21] is not referenced, although there is considerable discussion of this type of trial.
A cursory read of the book identified many misprints and grammatical mistakes. For example, a definition of clinical trials given in the excellent book by Piantadosi as "an experiment testing medical treatments on human subjects" is mangled to "an experimental testing medical treatment on human subject" on the first page of this book.
In the section on Cox's Proportional Hazard Model, the partial likelihood is described as being named partial because "It explicitly does not include the probabilities for subjects whose event times are censored", implying that these subjects are ignored. Then there is some confused discussion about the effect of including an "interaction" between time and treatment results because the authors rescale the time axis by subtracting 173 when the latest failure is at time 35.
I wish more care had been taken in the production of this book.

Contents:
1. Introduction
2. Two-way contingency tables
3. Loglinear models
4. Logistic regression models
5. Methods for matched data
6. Methods for count data
7. Transition from categorical to survival data

Readership: Graduate students in epidemiology and public health, biomedical research workers

The chapter headings show the breadth of coverage of this book, the subject matter of each meriting a book of its own. However, here the author's intention is to give the bare outline of each method and show its relationship to others in the field. The book is intended for the training of workers in the health sciences. It is assumed that they will have had a course in basic statistical theory. The author knows exactly what statistical tools they will need in practice. As well as the all important odds ratio, special topics such as the receiver operating characteristic, ROC curve, attributal risks and cross-over designs are discussed in detail.
The text reads like a set of lecture notes, which makes it handy for quick reference. Each chapter begins with data for which a model is suggested, and the likelihood function given. Emphasis is strongly on tests of hypotheses about the parameters of the model and model selection. Surprisingly, apart from goodness-of-fit tests, regression diagnostics are not mentioned at all. Where applicable, the Mantel-Haenszel analysis of the same data is also shown. Fragments of computer code show how the analysis can be implemented with SAS. Numerical examples are given and a set of exercises is included with each chapter.
The book is marred by a large number of typographical errors, some in the computer code and, at times, the verbal interpretation of the analysis is not illuminating.

Contents:
1. Reliability concepts and reliability data
2. Models, censoring, and likelihood for failure-time data
3. Nonparametric estimation
4. Location-scale-based parametric distributions
5. Other parametric distributions
6. Probability plotting
7. Parametric likelihood fitting concepts: Exponential distribution
8. Maximum likelihood for log-location-scale distributions
9. Bootstrap confidence intervals
10. Planning life tests
11. Parametric maximum likelihood: Other models
12. Prediction of future random quantities
13. Degradation data, models, and data analysis
14. Introduction to the use of Bayesian methods for reliability data
15. System reliability concepts and methods
16. Analysis of repairable system and other recurrence data
17. Failure-time regression analysis
18. Accelerated test models
19. Accelerated life tests
20. Planning accelerated life tests
21. Accelerated degradation tests
22. Case studies and further applications

Readership: Engineers and statisticians in industry, engineering and engineering-statistics students

This is a big, well-written, well-organized textbook for engineering statistics. It gives good coverage of the subject overall and is suitable for self-study because of the clarity and detail. Some particular features are as follows: a long list of failure time distributions is dealt with in Chapters 4 and 5; much detail of maximum likelihood, and associated large-sample methodology, is presented throughout; both censored and uncensored samples are treated; at the end of each chapter, except the last, graded exercises are given (but no answers, so they are presumably for class use); bibliographical notes follow each chapter; plenty of graphics appear, accomplished via Splus as are the computations (programs are said to be available on the Wiley website); many datasets are given for the reader to practice on. The elements of the Bayesian approach are described in one chapter, and the influence of the prior is investigated in an example in one other section. The rest of the book is mainly based on standard large-sample maximum likelihood theory.

Contents:
PART I : Nonmathematical Presentations
PART II : Mathematical Presentations
PART III: Personal Incomes and Firm Sizes
PART IV: The M 1963 Model for Price Variation
PART V: Beyond the M 1963

An alternative title for this book would have been "Mandelbrot on Mandelbrot". Let me be clear: I do not mean this in any negative sense. The author no doubt has made fundamental contributions to science in general, and economics in particular. His early ideas on using Lévy-stable and Pareto distributions for the modelling of return data, the importance on scaling and random time change and various other new ideas have been a constant source of inspiration for numerous researchers in economics. The book is mainly organized around various early research papers which are here reproduced and annotated by the author. Some new material is also introduced. The final product makes interesting reading, especially from a historical perspective. Whereas some ideas may be well accepted by now, it is always interesting to hear from one of the early contributors to the field how people reacted to these ideas early on.

Contents:
1. A Brownian model of financial market
2. Contingent claim valuation in a complete market
3. Single-agent consumption and investment
4. Equilibrium in a complete market
5. Contingent claims in incomplete markets
6. Constrained consumption and investment

Readership: Mathematically mature students and researchers interested in finance

This book is the sequel to the very successful Brownian Motion and Stochastic Calculus (BMSC) by the same authors [Short Book Reviews, Vol. 8, p. 45]. This indicates that the mathematical level is rather high. Those who have read BMSC, however, know the authors' excellent pedagogic qualities in presenting difficult material to a broader audience. The present book keeps up with the high standards set in BMSC. As was to be expected from experts in the field, the authors go beyond the classical "plain vanilla" models and questions by introducing portfolio constraints, consumption and investment, transaction costs. This both in an individual agent set-up as well as in an equilibrium model. A special novelty is the detailed discussion in Chapter 5 on incomplete markets. The book contains numerous examples, very detailed notes and up-to-date referencing to over six hundred and fifty items. A scholarly work indeed! No doubt this text deserves its rightful place in the list of classic texts on the subject of mathematical finance.

Contents:
PART I : Statistical Theory
1. Distribution theory
2. Estimation
3. Significance tests
4. Sequential methods
5. Reliability and survival analysis
6. Goodness-of-fit
7. Compound laws and mixtures
PART II : Applications
A. Actuarial science
B. Analysis of reciprocals
C. Demography
D. Histomorphometry
E. Electrical networks
F. Hydrology
G. Life tests
H. Management science
I. Meteorology
J. Mental health
K. Physiology
L. Remote sensing
M. Traffic noise intensity
N. Market research
O. Regression
P. Slug length in pipelines
Q. Ecology
R. Entomology
S. Small area estimation
T. CUSUM
U. Plutonium estimation

Readership: Applied statisticians, research workers in areas such as biology, environmental science, engineering, management science, quality control, survival studies

This monograph in intended as a companion to Professor Seshadri's earlier book, The Inverse Gaussian Distribution—A Case Study in Exponential Families (1993) [Short Book Reviews, Vol. 14, p. 42], not as a replacement. The structure of the book more nearly resembles that of Chhikara and Folks' The Inverse Gaussian Distribution (1989) [Short Book Reviews, Vol. 9, p. 24].
PART I covers the core theory with plenty of illustrative examples; most of this will be accessible to non-specialists in statistics.
PART II is the overwhelming reason why this book should join the two books already mentioned on the bookshelves of all scientific research institutes and university departments. It is very wide-ranging indeed, both in types of application and in areas of application; this reflects the very considerable growth of interest in the distribution in the past ten years.
The author has succeeded in including in the bibliography almost all the latest work on the distribution, theoretical as well as applied.

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

This book investigates Monte Carlo methods from a more statistical viewpoint than many of its competitors. As a result, in addition to the usual chapters on uniform and non-uniform random number generation and variance reduction, there is a refreshing chapter on the design and analysis of Monte Carlo studies in statistics, some discussion of alternatives to crude simulation through pseudo-random numbers such as Latin hypercube sampling and quasirandom numbers, and a bibliography that includes references to the World Wide Web. This is an excellent and readable text for undergraduates in the mathematical or statistical sciences. There is little discussion on validation of simulation models, software (other than Splus and IMSL) or output analysis and as such, it complements but does not replace the more engineering oriented discrete event simulation literature.

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

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

Reviewer:
Institute	ETH-Zürich
Place	Zürich, Switzerland
Name	F. Delbaen

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

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

Title	AN INTRODUCTION TO COPULAS.
Author	R.B. Nelson.
Publisher	New York: Springer-Verlag, 1999, pp. xi + 216, US$39.95 / £30.50.

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

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

Reviewer:
Institute	Statistics Canada
Place	Ottawa, Canada
Name	F.D. Gault

Reviewer:
Institute	Queen's University
Place	Kingston, Canada
Name	A.M. Herzberg

Reviewer:
Institute	University of Chicago
Place	Chicago, U.S.A.
Name	S.M. Stigler

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

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

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

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

Reviewer:
Institute	London School of Hygiene and Tropical Medicine
Place	London, U.K.
Name	C.D. Higgins

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

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

Reviewer:
Institute	SmithKline Beecham Pharmaceuticals
Place	Collegeville, U.S.A.
Name	V.V. Fedorov

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

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

Reviewer:
Institute	Swiss Federal Institute of Technology
Place	Zürich, Switzerland
Name	A.J. McNeil

Reviewer:
Institute	Steklov Mathematical Institute
Place	Moscow, Russia
Name	A.N. Shiryaev

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

Reviewer:
Institute	Oak Ridge National Laboratory
Place	Oak Ridge, U.S.A.
Name	V.V. Fedorov

Reviewer:
Institute	University of Essex
Place	Colchester, U.K.
Name	G.A. Barnard

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

Reviewer:
Institute	University of Western Ontario
Place	London, Canada
Name	D.R. Bellhouse

Reviewer:
Institute	Katholieke Universiteit Leuven
Place	Heverlee, Belgium
Name	J.L. Teugels

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

Reviewer:
Institute	University of British Columbia
Place	Vancouver, Canada
Name	J.V. Zidek

Reviewer:
Institute	University of Hong Kong
Place	Hong Kong, China
Name	H. Tong

Reviewer:
Institute	University of Leicester
Place	Leicester, U.K.
Name	M.J. Phillips

Reviewer:
Institute	University of Cambridge
Place	Cambridge, U.K.
Name	S.R.E. Turner

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