- Paperback: 636 pages
- Publisher: Cambridge University Press (July 27, 1990)
- Language: English
- ISBN-10: 0521287626
- ISBN-13: 978-0521287623
- Product Dimensions: 6 x 1.4 x 9 inches
- Shipping Weight: 2.2 pounds (View shipping rates and policies)
- Average Customer Review: 2 customer reviews
- Amazon Best Sellers Rank: #2,362,087 in Books (See Top 100 in Books)
Enter your mobile number or email address below and we'll send you a link to download the free Kindle App. Then you can start reading Kindle books on your smartphone, tablet, or computer - no Kindle device required.
To get the free app, enter your mobile phone number.
The Design of Experiments: Statistical Principles for Practical Applications
Use the Amazon App to scan ISBNs and compare prices.
Fulfillment by Amazon (FBA) is a service we offer sellers that lets them store their products in Amazon's fulfillment centers, and we directly pack, ship, and provide customer service for these products. Something we hope you'll especially enjoy: FBA items qualify for FREE Shipping and Amazon Prime.
If you're a seller, Fulfillment by Amazon can help you increase your sales. We invite you to learn more about Fulfillment by Amazon .
Customers who bought this item also bought
What other items do customers buy after viewing this item?
"Simply superlative...concepts and principles are presented with clarity, balance, and thoroughness." Ecology
"A systematic and comprehensive coverage of statistical principles as applied to the design of scientific experiments...The book is well-written and contains very useful information that is relevant to a variety of scientific disciplines." Choice
The statistical principles of good experimental design are explained by employing a minimum of mathematics and emphasizing the logical principles of statistical design.
Top customer reviews
There was a problem filtering reviews right now. Please try again later.
PASTOR BINOCULARIS: NOW WE HAVE NO EXCUSE
Mead, Roger. 1988. The design of experiments: statistical principles for practical applications. Cambridge University Press, New York. xiv + 620 p. $130.00, ISBN: 0521-24512-5.
The news that Professor Mead had written a book on experimental design was exhilarating. There has long been a great need for a thorough, up-to-date, concept-focused textbook appropriate for teaching experimental design. Anyone familiar with Mead's incisive criticism of the neglect of experimental design in one of biologists' more revered statistics books (see Biometrics 38:863-864, 1982) or with his own unusually lucid introductory statistics textbook (Mead, R., and R. Curnow. 1983. Statistical methods in agriculture and experimental biology, Chapman and Hall) will anticipate with pleasure reading his attempt to fill this lacuna.
The book is simply superlative. It is a worthy successor to R. A. Fisher's The design of experiments and D. R. Cox's Planning of experiments and seems likely to be the contemporary classic in its field. It does not, in my opinion, have a close competitor.
The book is strongly focused on concepts and principles. These are presented with clarity, balance, and thoroughness. It is a book on design, not design and analysis, and its success derives in part from its intentionally reduced attention to statistical analysis. Experimental design cannot easily be discussed very deeply, however, without some reference to statistical analysis, and the author gives us just the right amount of it. This includes a 42-page chapter on "General principles of linear models for the analysis of experimental data."
Mead's discussions of statistical matters contain clear-headed advice on several topics that elsewhere are generally treated with much myopia, cant, and carelessness. These include log transformations, multiple comparison tests, and analysis of repeated measures experiments. Among Mead's comments on these topics are: "It should be assumed that data for continuous variables should be transformed to a log scale unless there is good reason to believe that this is unnecessary"; "I recommend ... that multiple comparison methods be avoided; that the idea of experiment-wise error rates be retained, but only as a general principle"; and "[for repeated measures designs] analyses may usefully be performed for each measurement time, [since the purpose is] to discover the important patterns of treatment variation or lack of variation so that a sensible overall interpretation can be achieved."
"What!", you say, "Above item number 1/2/3 is contrary to the advice I was given on my examination/thesis/manuscript last month by a colleague/professor/editor/anonymous reviewer! And it conflicts with much of what I see published in top journals such as E______y!"
Mead's comments on these topics are not as extensive as one would like. But in an understated way he helps bring out an important truth. The application of statistics in biology disconcertingly often reflects the interaction of blind sheep and half-sighted shepherds.
The design topics covered are fairly standard. The clarity and thoroughness of the book, not novelty of subject matter, are its greatest virtue. Nevertheless, there is much original material.
In a thought-provoking chapter on "Computers for analyzing experimental data," Mead discusses the nature and problems of canned programs, cites the need for development of "an interactive, user-friendly, statistically valid and comprehensive general package," and calls for "a complete reappraisal of the principles of experimental design without the restraints imposed by the limits of computational practice" which existed in the past.
In his chapter on "Replication," he devotes several pages to pseudoreplication (though without using that term), notes its frequency in the literature and shows, with examples, various ways in which it can come about. No other book on design does this so well. He does, however, slip into some terminology that, though common, is misleading. Multiple measurements made on a single experimental unit are implied to represent "false," not "useful," or not "proper" replication (pp. 111-113), or, in the equally inappropriate language of others, "pseudoreplicates." Such language is a confusing way to make the point that replicate measurements on a single unit cannot be treated as if they came from separate units. If statistical improprieties occur, they should be attributed to the analyst, not the measurement. Save the cuss words for the former!
The mere fact that the book has a chapter dedicated to "Replication" is itself revolutionary. The topic is so fundamental that one might expect every textbook on experimental design to have such a chapter. Yet of 12 other such books on my shelf not a single one does. Mead's chapters titled "Elementary ideas of treatment structure," "Blocking," and "Randomization" likewise reflect his unusual ability to see, and help the reader to see, the forest despite the trees.
The concepts of response design and the evaluation unit, both fairly new to experimental design (e.g., S. Urquhart. 1981. HortScience 16:621-627) and not presented by Mead, would have helped explicate some key ideas. Both the concepts and the terms are of considerable pedagogic value, and could have kept Mead from wrongly implying that, for example, unless successive repeated measurements are taken on exactly the same evaluation units, they cannotbe considered to have been "taken on exactly the same experimental units" (p. 408).
In his preface Mead states: "Understanding the fundamental concepts of design is essential for all research scientists involved in programmes of experimental work." Most scientists will find the statement unexceptional but also acknowledge that they themselves never had a course, or at least a good one, on the subject. What they did and didn't learn from "on-the-job training" and faculty advisors, however, is now sadly clear in the scientific literature. Lack of an excellent text on experimental design has always been one excuse. We no longer have it. Dismiss your old gurus and come walk with Pastor Binocularis.
Because this is the very best book on experimental design that has ever been written, it is unfortunate, though possibly appropriate, that it is also the most expensive. Regardless, the pit will rarely if ever be ordered in quantity as a course textbook, and sales will be one to two orders of magnitude lower than they might have been with a reasonable price. The rapid exhaustion of the first printing further supports the miscalculation hypothesis. A second printing, and a paperback edition, are promised.
STUART H. HURLBERT, San Diego State University, Department of Biology, San Diego, California 92182-0057