Quality Improvement Through Planned Experimentation
This text is much more than a handbook for planned experiments. The authors Ronald Moen, Thomas Nolan, and Lloyd Provost have created a piece that is equally at home in the classroom as in any Quality professional’s library. "Quality Improvement through Planned Experimentation" is a thorough yet easy to follow guide to experimental design and analysis in the quest for continuous improvement.
While this book will certainly be used as a reference for full factorial and fractional factorial experimental designs, the true value is the application and understanding of sequential study cycles. Students of Deming, Shewhart, Taguchi and other Quality advocates will enjoy the coupling of their teachings with the authors’ methodology for experimental designs to create a systematic model for quality improvement.
Quality novices will appreciate the complete explanation of the components of planned experiments along with the references to earlier works including the Plan, Do, Study, Act "Shewhart Cycle". Seasoned veterans will benefit from the detailed instructions of identifying and isolating response, background, nuisance and factor variables. Of particular interest was the use of "chunk" variables and blocking in conjunction with randomization to increase the degree of belief of an experimenter’s conclusions.
While other texts are deemphasizing the importance of randomization, Moen, Thomas and Provost have appropriately centered their design methodology on randomization and replication. At the same time, they have given alternatives to isolate nuisance variables when it is not feasible to fully randomize studies.
Another central tool for designing studies is the Planning form shown in Fig. 1. This form is the starting documentation for all PDSA cycles regardless of planned experiment type. Beginning with the general approach to One-Factor Experiments, the authors lead us through multiple case studies representing increasingly more complex designs.
Progressing from full factorial to fractional factorial screening experiments at multiple levels (Fig.2,) sequential experimentation is demonstrated to isolate factor and interaction responses while reducing run sizes. The chapter on nested designs is particularly informative to an experimental designer, as is the laboratory study containing both crossed and nested factors that requires a combination of factorial and nested patterns.
Sections of the book on data analysis are as comprehensive as the design of the experiments. Some of the tools explained include run charts, blocked and time variant control charts, dot-frequency and dot diagrams (Figs. 3 and 4), and response curves and surfaces. A download of Study-It software is included with the textbook. The authors incorporate many examples of software outputs anddo an exceptional job in demonstrating the building, use, and interpretation of these analytical tools. These sections may be the most powerful in aiding engineers and statisticians to integrate planned experiment methodologies into their everyday work requirements.
Also of notable interest are the last two chapters which address specifically emerging applications in Health Care and New Product design. The human factor poses unique circumstances when designing experiments in the health care field. Special elements are introduced to facilitate experimental design while maintaining ethical considerations and adhering to the Federal Policy for Experimenting on Human Subjects, also known as the "Common Rule." To facilitate the coordination of many different functional areas integrated with a new design, the authors developed a four phase design approach to the major activities denoting the areas suitable for planned experiment applications. See Fig. 5