Design and Analyse Your Experiment with Minitab . Tony Greenfield and Andrew Metcalfe (Eds). 2007. Hodder Arnold: UK. 303 pages

Jim Morrison has been campaigning for many years for engineers to be taught statistics in ways that would be useful in the real world of variation in practical engineering, as opposed to the tidy, pure world of mathematical statistics. In this respect he has not only followed in the footsteps of the first great engineering statistician, W.A. Shewhart, but he has also developed new methods. This is the first time that he has put his knowledge and ideas together in a book. I am happy to welcome it, and I am joined in this by George Box, who has written a generous foreword. The preface outlines the philosophy of the book clearly. To quote: ‘Engineers using statistical methods need not concern themselves with profound issues of statistical inference or the subtleties of statistical mathematics. They require only familiarity with relevant statistical methods, an understanding of how they work and how to use them safely.’ And: ‘The engineer who is lacking in statistical skills is less than competent to handle variability.’ The first two chapters describe the nature of variation and basic statistical methods, clearly and concisely. These are followed by chapters on production, design and research. The production chapter describes the common methods, including sampling, control charts, analysis of variance and regression analysis. Morrison’s technique of ‘variance synthesis,’ which when introduced in the 1950s was a precursor to Taguchi’s ideas on parameter and tolerance design, is described in the design chapter. The chapter on research covers statistical design of experiments, evolutionary operation and multiple regression. Taguchi’s methods are touched on, but only very briefly. These three chapters are excellent, and form the practical heart of the book. My only reservation is that nowhere does the author discuss the fact, first explained by Shewhart but hardly ever mentioned in texts on practical statistics applications, that most real-life variation (particularly in engineering) does not follow the normal distribution. In the final three chapters the book explores wider issues, such as statistical computing, management and training. These make some good points, notably that all engineering undergraduate courses should include practical statistics. However, the discussion of computing and software is mostly historical, and topics like powerful, easy-to-use modern software such as spreadsheets and Minitab are not discussed. There is no mention of ISO9000 or Six Sigma in the chapter on management. The lists of references are far too long for a book that emphasizes practical application. These reservations aside, I strongly recommend the book as an excellent introduction for training and for reference by engineers.