Applications of Statistics to Industrial Experimentation

Mr. Daniel's book concerns itself with statistical aids available for the planning of industrial experiments and with the analysis and interpretation of the data collected from such experiments. In it, he questions some standard designs and procedures in analysis, contending that many can be checked only after the data are in. There are many detailed examples. Chapters 1-3: Design of multifactor industrial experiments has become routinized. In addition, analysis of resulting data is often handled by computer programs long-since outdated-programs still being taught on the university level and still part of most statistical texts. Chapter 4: Here the study of data from simple two-factor three-level designs is carried further than in other available works. Spotting and interpretation of irregular results are feasible even in such small data sets. Chapters 5-8: Size of industrial experiments is rarely determined by standard statistical criteria of power or by shortness of confidence intervals. The real criteria are usually time- and budget-restrictions, and nearness to final decision on marketability. Chapter 10: Blocking of factorial and fractional factorial plans is considered. Chapters 11-13: The rationale of fractionation of factorial designs is given in new detail. Chapter 14: Clearing up residual ambiguities after completion of a fractional design is often proposed by way of doubling the amount of work really needed. Daniel points out that much smaller augmentations are often feasible, and their selection is detailed for many cases. Chapter 15: When smooth time trends (linear and quadratic) are known to occur, more efficient and compact plans than the usual blocked designs are possible. Here they are offered. Chapter 16: Serious examples of nested designs require special care-especially in analysis. Here is a volume that should be valuable to experimenters who have some knowledge of elementary statistics (at least one year of undergraduate-level statistics), and to the statistician who seeks simple explanations, detailed examples, and documentation of the many outcomes that can be expected to occur.Daniel, Cuthbert is the author of 'Applications of Statistics to Industrial Experimentation' with ISBN 9780471194699 and ISBN 0471194697.