To make it a little more attractive to general audiences the editors have reprinted one of George Box's articles on scientific learning. Box's article was originally published in the proceedings in computational statistics in 1996 but certainly fits well into a volume like this. Box's philosophy on the iterative nature of scientific learning is always interesting and thought-provoking.

There are a total of 60 authors and 35 articles. A variety of topics are included and most of the papers are oriented toward applications with real examples and case studies.

Part I is on statistics and quality and includes Box's paper, a conceptual paper on variance reduction in manufacturing, roles of academic statisticians as consultants to industry and understanding the prevention focus of QS-9000 contrasted to the control focus of ISO-9000. It includes a total of 5 papers.

Part II deal with statistical process control. In addition to papers on the latest advances in control charting including automatic feedback control type methods there are papers on process capability indices with a thorough and accurate treatment of the subject. Carlsson's article is a very scholarly treatment that points out the pitfalls of naive use of process control indices and he refers to the very nice elementary articles by Gunter. I also referenced Gunter's articles in my book on the bootstrap where I applied the bootstrap to an example of estimating Cpk. Carlsson refers to the vast literature on these indices including the book by Kotz and Johnson that summarizes advances up to 1993 including the bootstrap approaches. The work on Vannman is also cited.
Carlsson's contribution is to generalize the Cpm index for unstable processes (due to a moving process mean) where there are assignable causes for the shift that can be modeled through an ANOVA model.

Kocherlakota and Kocherlakota provide a very technical paper on the statistical behavior of certain process capability index estimates when the observations come from a contaminated bivariate normal distribution (i.e. a mixture of two bivariate normals with a common covariance matrix but a shift in the mean vector). Part II consists of 11 papers

Part III covers the design and analysis of experiments and includes 9 papers including coverage on robust design, process optimization, aspects of TQM and some practical issues. The paper by Ghosh and Lopez covers how probability plots in factor screening experiments can be misleading. They also deal with methods for handling missing data.

Part IV contains articles on statistical methods for determining reliability and consists of 6 articles. It includes an article by Ratnaparkhi and Park on the application of generalized linear models in reliability studies for composite materials.

Last but not least Part V deals with statistical methods for quality improvement and consists of 4 papers. Some of these papers are a little unusual. Murdoch introduces a variation of Gibbs sampling in order to construct confidence regions for measurement uncertainty efficiently when up to 10 model parameters are involved. Gupta and Iannuzzi use logit and probit analysis methods to determine if devices used to detect antibiotics in milk meet regulatory requirements on sensitivity.