Customer Reviews

Statistics for Experimenters: Design, Innovation, and Discovery , 2nd Edition

5 star:		(5)
4 star:		(2)
3 star:		(1)
2 star:		(1)
1 star:		(1)

 

 

I am a mathematical statistician and I appreciate and understand modern books on experimental designs such as the book by Wu and Hamada. However, the first edition of this book became an immediate classic because George Box is a genius and is from a rare breed of statisticians who have strong theoretical and practical experience in statistical methods and in this case statistical design. Stu Hunter and Bill Hunter are two other statisticians with strong applied backgrounds in engineering and other applications of experimental design. Before you can appreciate the theory you need to understand the theory. The first edition of this book presented the concepts beautifully. This was a great help to me as I had learned the theory and the construction of factorial designs, fractional factorial designs and incomplete block designs, but never had a clear understanding of when to use them until I read this book. Other important simple designs of great practical importance are also covered extremely well.

I wrote a review of the first edition of this text. Justin Hunter appreciated it so much that he wrote a very touching email to me on this and he was very kind to send me a complimentary copy of the second edition. Justin is the son of Bill Hunter. Unfortunately Bill past away before this second edition was conceived. I believe it was partly as a tribute to Bill that George Box and Stu Hunter put together this revised edition. The spirit and philosophy of the first edition has been maintained and since the first edition had appear way back in 1978 the production of an updated edition is welcome and way over due.

Please read the book review by Justin Hunter. He is very upfront about his bias for his father but what he writes is honest and comes from an interesting and unique perspective.

This is the best applied book in any scientific or mathematical subject that I have ever read. The reviewers who are looking for equations and the typical assumptions-theory-proof presentation just picked up the wrong book.

If you're interested in applying experimental design to real-world problems, this book is indispensable. The authors help you get inside the math and really understand the important and often profound issues. It is easy to write a book that regurgitates equations and proofs; it is a major accomplishment to bring to bear decades of practical insights.

I still need to read the 2nd edition in detail and I plan to spend as much time as needed. Based on my brief reading of selected sections, the authors have retained the same style which has made their 1st edition a classic.

A classic updated. The first edition, written in 1978, has 8 ratings of 5 stars and one of 4 stars ("due to it's age") on Amazon.

As Six Sigma grew in popularity many more organizations discovered the power of design of experiments, as developed by Sir Ronald Fisher. George Box, was his student. He built on the ideas of Fisher and together with the other authors created a book that had already become the classic reference for designed experiments and how to apply statistics to improve results before the term "Six Sigma" was coined. The new edition retains the clear and friendly writing style of the original and adds some new tools to the experimenters toolbox (as mentioned in the publishers text above).

I am biased as my father was one of the 3 authors of the original book and the second edition is dedicated to him. Luckily, you don't have to take my word for it, many others regard this book as the primary resource for design of experiments.

Statistics for Experimenters is a tool that is used frequently (not a book gathering dust on the shelf) by many experimenters, engineers, Six Sigma experts and others as they use experiments to learn, innovate and improve.

Superb! If you are involved, in any way, with science or engineering, you need this book on your shelf (after you have carefully read it twice). My only complaint is that I found out about it circuitously reading Prof. Box's "Improving Almost Anything"; I was curious what the often cited BHH reference was. I think someone should have a word with the publisher's marketing department; if we don't know about it, how are we supposed to buy it?

First: the BOOK is very well written. Box is an true expert on his field, and a good writer. How well his style and pedagogic layout 'connects' with you will depend on your humour (if any), whether you will use it for study or reference, and your background (specifically this book is written with an eye to manufacturing industry, not just chemo/bio disciplines).

Second: the TOPIC of experiment design (and strategy) is extremely useful to understand. If you are involved in industrial experiments (as opposed to data analysis) this is far more useful than the statistics you learned at university.

In my opinion the 'how' (the various design types, tests, plots etc) is actually secondary, the most useful is to understand *why* and *when* to design experiments. The authors do a good job here too, it's possible to get the essentials without getting bogged down in the statistics.

Conclusion: if you want a solid book on this topic (and you do!), then I recommend this one.
[...]

While it seems odd to describe a very good book on statistics as a great read versus merely as a great reference book, this latest edition by George Box et al., is very much both. Useful new topics were added in this edition, e.g., experimentation using surface response methods, graphical analysis of variance, computer analysis of complex designs, simplification by transformation, etc. It stays on my reference shelf in easy reach.

I got this book to help me prepare for the CQE. This is book is a good reference for the CQE and also it helped me know more on DOE.

I have found it very difficult to identify a good book on study design for the biomedical industry as most are too focused on engineering/manufacturing applications. Though this book does discuss many different study designs and their theoretical underpinnings well, it does not do a very good job of explaining the issues of sample size and power. These issues are critical to designing studies in biologic settings.

If one is planning experiments on animals, it is mandatory to design studies that minimize the number of animals required to answer the scientific question at hand--given pre-specified levels of statistical power and effect size. Similarly, when experiments are conducted on humans (i.e. clinical trials) it is important to include enough patients so that adequate precision is obtained. I wish this book discussed better how to find the appropriate number of replications required to optimize experimental conclusions (like the interactions terms in a factorial experiment, for instance).

This book is supposed to be a great text book desigend for the course--EXPERIMENT DESIGN. However, after using it for the first 2 weeks, we found it is very obscure in telling the basic concepts and the examples are long but distractive. I have to read other books to learn the basic concepts to understand what this book is telling. My reference books are 'PROBABILITY AND STATISTCS FOR ENGINEERS AND SCIENTISTS', by Ronald E. Walpole 7th edtion and 'Probability, random variables and stochastic processes', by Athanasios Papoulis, 4th edtion.

What's more, this book is unique in that it has the most typos I have ever found in an textbook.

For example:

Chap 2. Problem 14. line 3 pp 65, the "probablilty shoulde be about 1 'change' in 500".

Chap3. Ex3.6 line3 pp84," the corresponding 'scabbed' t distributions".

Mistakes are almost in every chapters. When you read, you will found more, enjoy them.

As a text book, this book will drive students crazy. Despite the typos which can be found not occasionally, I found the authors wrote most of the examples using large paragraphs but hard to comprehend.Pretty often the results they generated comes from nowhere. No clear deduction, some times even no equations, just tell you some meanlingless number. The authors must have treated students as experienced as they are, therefore they omit most of the important details for a student to follow. Obviously, by using computer software, they can easily get answers from problems they made without bothering write down how they can get it by pencil and paper.

My suggestion is that if ever you want to learn some statistics, get some books which are clearly written and well illustrated. This book is obviously overrated.