No abstract

No abstract

No abstract

Reviews[Part 4, fractional replication, confounding, response surface, change-over trials and incomplete block designs. The book is written in a succinct style sometimes reminiscent of lecture notes. Considerable space is devoted to worked examples though no problems are presented for students to work on themselves.Little new material has been added to Anderson and McLean's Design of Experiments: A Realistic Approach. (Marcel Dekker, 1971) which this text supersedes though the order of presentation is different. Most of what is covered appears in greater detail in Cochran and Cox's Experimental Designs. (2nd ed., Wiley, 1957). The author adopts a prescriptive style of teaching which stresses what to do rather than why to do it. A distinctive feature about this book is the inclusion of the "Report of Statistical Analysis" sections. These tell the researcher exactly how to write up the analysis of each design. This emphasis on standardisation makes the analyses easier to follow at the expense of distancing a reader from the phenomena under investigation. The analyses are specified quite independently of how the results turn out. An inflexible approach is also encouraged by not providing possible options. A related problem is that little emphasis is placed on theory, numerical examples being used to clarify inadequate explanations of topics.Particular omissions which might be seen as important include:-An adequate coverage of the random factor models and how they affect significance testing in other than one factor designs, the analysis of contingency tables, the treatment of outliers, the possible use of transformations and the analysis of residuals. In the chapter on change-over trials examples are given where order effects are not examined. This could have unfortunate consequences if the text were being used as a cook book.To sum up this is a text covering traditional topics in experimental design. It should enable the statistically naive to analyse complex experiments but it is not designed to make them understand what they are doing.

No abstract