This book will be useful to anyone who will be working in probability and is already acquainted with measure theory. It includes results that are not standard material for a real analysis text but that are of interest in probability. The author, who needs no introduction, uses simple examples (coin tossing, card shuffling) to illustrate how the "interesting, but imprecise real world" is represented in the "duller but more precise mathematical world", as well as the importance of keeping them apart. I know of no other publication where this is done as effectively and as beautifully as it is presented here. Each proof in this book is actually a sketch of its proof. This means that the reader does not get lost in details and sees the concepts transparently. The author also does a good job of limiting himself to what belongs to the title, as opposed to developing what would belong in a probability text.

I recommend this book to anyone interested in measure theory, whether or not their interest extends to probability.

Ray Carroll and David Ruppert are well known research statisticians who have published many joint articles on regression, weighted regression and transformation and they have also written an excellent book together on this research topic. Stefanski has recently published several papers on measurement error models with Carroll. Here they have teamed up to write a statistics text on a unique topic. Measurement error models are common and practical when dealing with covariates that have measurement error. Least squares estimation in linear regression is based on the assumption that the predictor variables are measured without error. There are many articles and an excellent text by Fuller "Measurement Error Models", published by Wiley in 1988 that deals with the linear case. Also look at a section in Chapter 5 of Miller's "Beyond ANOVA, Basics of Applied Statistics" that refers to the problem as the error in variables problem. For the nonlinear case this is the first treatment. Well written and well documented, this text provides an up-to-date account of the theory and methods and provides real applications (e.g. the Framingham Heart Study). This is a great reference as are many of the other monographs in this series by Chapman and Hall/CRC Press. Includes bootstrap approaches in the chapter on fitting methods and models.

Ram Gnanadesikan wrote the first edition of this book in 1977. It was unique then as a practical text on multivariate analysis based on his experience at Bell Labs. There is a nice mix of good theory and practice in the book. Also it is not tied to the theory of the multivariate normal distribution and it emphasizes graphical representations, robust methods, outlier detection and dimensionality reduction techniques. Clustering and classification methods are also covered.

Gnandesikan and his colleagues were the first to think of using Hampel's influence function for detecting outliers in multivariate data. Their research is covered in this book. I used it in my work at Oak Ridge National Laboratory in the late 1970s to detect multivariate outliers as part of our energy data validation effort. I also applied these ideas to time series analysis.

Twenty years after the publication of the first edition, Gnanadesikan decided to produce the long overdue revision. He is currently retired from Bell Labs and is employed as a Professor of Statistics at Rutgers University. The second edition incorporates advances from the last 20 years and emphasizes the newly available software.

The new computer-intensive methods including the bootstrap are not covered.

This book is great! Some derivations and occasional proofs are included in the text, but it has clearly been written with the applied researcher in mind. In the author's words he has provided "careful intuitive explanations of the concepts and [has] included many insights typically available only in journal articles or in the minds of practitioners" with his primary objective being "clarity of exposition." I have found it to be very easy to read with numerous helpful examples (the data sets and SAS command files for which are available via an ftp site).

This brand new version includes new chapters on cluster analysis, multidimensional scaling, correspondence analysis, and biplots.

Formally speaking, this is the second edition of a set of Paris lecture notes published by Gromov two decades ago in the French language. However, such a wealth of entirely new material has been added that in essence we are talking of a new book.

Among the additions, the bulky new chapter 3 1/2+ stands out, dealing with the phenomenon of concentration of measure on high-dimensional structures. This is a relatively recent discovery of modern analysis and geometry, tracing its origin to the work of Paul Levy and especially Vitali Milman. The essence of the phenomenon is that on many multidimensional structures, every `nice' function is constant with high probability. The manifestations of the phenomenon are many - from geometric functional analysis (Dvoretzky theorem) through information theory (blowing-up lemma) and probability (law of large numbers) to graph theory (superconcentrators) and topological dynamics. As Gromov stresses in his book, even deeper aspects of the concentration phenomenon have been long since discovered and are constantly explored in statistical physics in the context of phase transitions of various kind, and some of the first known examples where phase transitions appear in the context of geometry have been discovered by Gromov himself, e.g. for hyperbolic groups. Finding and exploring more instances of phase transitions in mathematics might well become a unifying heuristic principle across a large number of disciplines.

The mathematical setting for dealing with concentration and related issues is the concept of a metric space equipped with finite measure, what Gromov calls an mm-space. Apart from concrete objects (such as for instance spheres and cubes), there are `higher-level' examples of mm-spaces, for instance those whose elements are isomorphism classes of mathematical objects themselves (e.g. Riemanning manifolds or finitely generated groups). This leads to a probabilistic treatment of such objects. Of course Gromov's strength is that his treatment is always concrete and he never theorizes without having particular objects and applications in mind.

It is quite safe to claim that the full range and power of applications of the interaction between metric and measure are yet to be discovered, which is what makes this book so important. It is rich in open questions and suggested new research directions, but more than that, it helps the reader to develop a good intuitive feeling of where things are going these days, what things ought to be done, and what constitutes proper mathematics.

Even though I unexpectedly found myself among the privileged ones who received a copy of the book as a gift from the author, I would have certainly purchased it otherwise, as I firmly believe that every mathematical library in the world, be it that of a top-class University or just a modest, lovingly selected office collection of a humble mathematician, will be wanting without a copy of the monograph under review, which might well become one of the most important books in mathematical sciences for the early XXIst century.

The probability and statistics for engineeting book provides a very good first book on probability and statistics. It is very useful to engineers and scientists that need to analyse and interpret data. It has a very good material on statistical inference and on quality and reliability. I strongly recommend this book.

A pity this text has gone out of print if you are a student
or practitioner of geostatistics this book is a must have as a comprehensive guide to the subject written by the world leaders in this field. Buy now!

As usual, Paul Allison has produced an accessible and practical treatment of conceptual and methodological issues that commonly confound social scientists. His discussion of the meaning, effects, and remedies for missing data is thorough and clear. In particular, the section on multiple imputation is extremely well-done.

This is a reference work that will improve the scholarship of even the most rigorous researcher, and yet can serve as a wonderful introductory text on the subject of missing data for students at many levels.

Mixed effects linear models are very useful particularly in medical research (e.g. device or drug trials). Pinheiro and Bates provide comprehensive cover of both linear and nonlinear mixed effects models with many applications. Implementation is illustrated using the S programming language and the software package SPlus.

Bates is an expert on nonlinear regression and hence the emphasis on the nonlinear models as well as the linear ones.

Excellent, but not easy... Sennett succeeds in convincingly showing that Plantinga's ontological arguement fail, as well as his arguement for reformed epistemology and proper function. However Sennett shows that Plantinga's contribution to the problem of evil is valid.
Plantinga wrote an afterword on the cover, and seems to agree with Sennett.

An excellent book, valuable for anyone interested in philosophy of religion and who has an academic level.