Although the title seems to target statisticians, i find that this book is a blessing for any scientist involved in data analysis. Chatfield being a statistician focuses on his preferred audience, but all scientists having to analyze data, no matter what the objective is, should read this book. Data analysis is an art that deserves that one spends some serious time on thinking about methodology. This book is short, easy to read, and will provide you with a fresh view about what data analysis should be.

It is divided into three parts :
- first a pragmatic discussion of all aspects of data analysis as problem solving,
- then the second part deals with exercices aimed at showing the "artistic" nature of data analysis,
- and finally a refresher in statistics along with pragmatic comments.

Both the statistician and the scientist will benefit from all three parts.

The objectives of this text are very similar to the objectives in the Kotz and Johnson treatise of 1993. Written 5 years later this text can be looked at in part as an update to that volume.

However, that book was more of a reference manual for process capability whereas this book reads more like a course text including many more illustrations and examples. Also there is an attempt here to include both the theoretical and applied aspects of capability indices.

It covers the available distribution theory results for processes with normal distributions and non-normal as well. The effect of non-normality is carefully analyzed and alternative methods including the bootstrap are considered. This has all the attributes of Kotz and Johnson book but is more current, more detailed and has more references.

The authors are experts in the law and the use of statistics in legal settings. This book gives an excellent overview of how statistics is applied in the courtroom. Many interesting and important cases are presented. Although not professional statisticians the authors have a good understanding of statistical methods. The book gives one of the best presentations of the importance of randomized controlled experiments that I have seen.

Sally Morton and John Rolph edited and contributed to this collection of case studies of public policy issues that were conducted by the Economics and Statistics Group at the RAND Corporation. Sally is the current head of the group and John is a former head who is now a department chairman at the University of Southern California. The statistics group at RAND has a celebrated history for excellent and unbiased analyses of public policy data. As an organization they have existed since 1976 but individuals from the group had an impact even early. For example, Bill Rogers was part of the famous Princeton robustness study (published in 1972) while he was employed at RAND.

Many of the leaders of this group have participated in writing these studies including several who have moved on to careers elsewhere (e.g. Bob Bell, now at AT&T Labs - Research; John Rolph, now at USC; Jim Hodges, now at University of Minnesota; and Carl Morris, now at Harvard University).

Those currently at RAND who have contributed include Allan Abrahamse, John Adams, Phyllis Ellickson, Lionel Galway, Catherine Jackson, Dan McCaffrey, Sally Morton and Dan Relles. This is a mix of very seasoned RAND statisticians along with some junior members and colleagues. Several members of the group did not contribute to the case studies but well could have. I was particularly surprised at the absence of Naihua Duan who is an ASA fellow and has contributed to major health studies at RAND. Naihua has also been responsible for innovations while at RAND including contributions to sliced inverse regression and transformation "smearing" methods.

Nevertheless, the collection of studies are both interesting and important to the public and in several cases the findings go counter to the popular information in the media. Well-known Stanford Statistics Professor Brad Efron calls the media statements "misinformation" and "disinformation" and claims that RAND gets it right in his foreward to the book. Many Stanford students and colleagues of Efron had careers at RAND including Bob Bell, Naihua Duan, Carl Morris, Bill Rogers and Sally Morton.

There are a total of 10 case studies included. The first three are categorized as primarily addressing the collection of data (addressing issues in the design phase). The next three are considered to be primarily addressing the detection of effects (estimation or hypothesis testing aspects of statistical analysis) and the last four are considered to emphasize the understanding of relationships.

I have skimmed through all ten case studies and have read case numbers 3, 4, 5 and 7 in detail. The topics are as follows: 1. School-Based Drug Prevention by Bell and Ellickson. 2. The Health Insurance Experiment by Morris and Hill. 3. Counting the Homeless by Abrahamse. 4. Periodicity in the Global Mean Temperature Series? by Adams, Hammitt and Hodges. 5. Racial Bias in Death Sentencing by Morton and Rolph. 6. Malpractice and the Impaired Physician by McGuigan and Rolph. 7. Supply Delays for F-14 Jet Engine Repair Parts by Galway. 8. Hospital Mortality Rates by Thomas and Rolph. 9. Eye-Care Supply and Need by Relles, Jackson and Lee. 10. Modeling Block Grant Formulas for Substance Abuse Treatment by McCaffrey and Adams.

Analysis in #3 indicates that there are only about 400 homeless in Orange County as opposed to public estimates and claims of 4000 or more. Results in #4 indicate that the data are inconclusive regarding a global warming effect. In #5 both logistic regression and tree classification methods are used to show no clear bias in death sentencing based on the race of the victim. In #7 careful analysis of the data reveal that transporting supplies is the key factor in delays for getting repair parts for the engines and not the slow procurement process.

As an applied statistician who does a fair amount of consulting, I always find good case studies to be enlightening and helpful to me in my practice of statistics. These articles are very good and enlightening and they follow a common format. They start with an executive summary that provides an overview and the bottom line results. This is followed by an introductory section and then a section describing the study design, data collection, data sources and elements. The third section deals with datafile creation, descriptive statistics and exploratory data analysis. The fourth section covers statistical methods and models used. The fifth section gives results. Section 6 is a discussion section which may include summary, possible future extensions of the analysis, and conclusion and recommendations. The final section provides exercises. This last section is excellent for a course based on the case studies as it tests the student knowledge based on material learned in the case study. Sometimes self-contained problems are given but in other cases the reader is referred to the casebook web page at the RAND web site where data sources can be found to do the exercises.

In practical work I have always found that a clear understanding of the problem and good descriptive statistics and/or graphics are far more important than the particular method of analysis (which often times can be very elementary). These studies exhibit this principle well. In many cases good exploratory analysis, good design and clear understanding lead to the key results and the appropriate statistical methods. These methods are usually simple and elementary although some are fairly new tools (e.g. bootstrap, tree classification and empirical Bayes methods).

I felt that this book writen by Dr. Heldt was thoroughly insightful and was full of great ideas.

This book is an excellent introduction to queueing analysis using matrix analytic techiques. The book is very readable and the author provides numerous examples to illustrate crucial points. Analysis of the M/M/1, M/G/1, GI/M/1, M/G/C, and GI/G/1 queues and variants are performed via a consistent framework. Also, the author presents an introduction to busy period analyses which is interesting. The only comparable books are by Marcel Neuts, however, Neut's treatment is geared towards researchers or graduate students in the field.

This book consists of a huge number of examples that can be programmed on a spreadsheet. It illustrates random walks, Poisson processes, the achievement of thermal equilibrium, Boltzmann processes, Galton boards -- tons of examples in a readable and interesting way.

This work was written in 1978 by the Old Man himself, Arpad Elo. Chess players remember Elo as the cantankerous professor of physics from Marquette University, and as the inventor of the rating systems used by the United States Chess Federation (USCF) and the International Chess Federation (FIDE). Who better to explain the mathematics behind the rating system which, according to a letter Elo wrote to Chess Life several years ago, "is, after all, MY system". The explanations are semi-technical, but understandable by anybody with a mathematical inclination, regardless of their education in that field. There is even a chapter in which ratings are calculated retroactively, for grandmasters of bygone days. The book was written B.K. (before Kasparov), so of course Bobby Fischer comes out on top, with a rating of about 2780. His closest competitors are Lasker, Capablanca, and Botvinnik, each of whom peaked at about 2720. The average rating of tournament players in the U.S., by the way, is about 1500, several classes below the stars. Interesting reading.

Like its predecessor, "Reading and Understanding MORE Multivariate Statistics" achieves exactly what its title implies. Geared toward non-statisticians in behavioral and social science fields, this book provides clear and reasonably simple explanations of common multivariate analyses. This book includes special attention to scales of measurement, reliability and generalizability theory, item response theory, and assessing the validity of measurement. In addition, it covers cluster analysis, Q-technique factor analysis, structural equation modeling, canonical correlation analysis, repeated measures analysis, and survival analysis. The authors present the conceptual underpinnings, underlying assumptions, and basic procedures for each analysis with a minimum of equations and many concrete examples. The book not teach you how to perform the analyses but does provide references for those who wish to get more detailed information. As a research scientist who doesn't always remember everything I learned in graduate statistics class, I find this book an invaluable aid keeping up with the current literature in my field and in making the most of statistical consultations. This book is ideal for anyone whose job requires them to be a "consumer" of research; for researchers who wish to further their understanding of data analysis; and as a companion text for graduate statistics classes.

As a professional chemist, I do not look back on the statistics courses I took with particular fondness. While I didn't find them very difficult, neither did I feel that provided me with any deep insights. It was with this, lukewarm attitude that I first picked up Reading Between the Numbers. And then I noticed Chandler's quotation in one of its early chapters. Statistics and Chandler?! This must be interesting. I purchased the book and wasn't disappointed. I found the statistical concepts explained in a way that is both simple and insightful. And the book reminded me that I should have reviewed the topic, which is so important for my work, long ago. But I found the book unique in other ways as well. First, the book is relevant - there is not one concept in it that is not in some way connected to the real world. The author takes special care to show how and why statistics is important in both our professional and everyday lives. The concepts in Reading Between the Numbers are related to applications and vice versa, so that I can now see both sides in a new and more interesting light. Above all, I now feel that the ideas underlying statistics are more accessible to me than before. Second, as in many fields, the more advanced topics are based on earlier ones. And throughout, that author provides reminders of the basics so that I didn't find myself having to return to earlier chapters in order to understand later ones. It seems to me that author was simply sensitive to the needs of nonprofessional readers, which is a rarity in professional books. Finally, Reading Between the Numbers a joy to read. It is written as fiction rather than science and is often very funny. That the author has a rather quirky sense of humor is clear. That he has succeeded in combining it with statistics seems to me no small achievement.