 |
|
February 2002
Volume 1 • Number 2
Guest Editorial
Finding the Best Tools for Six Sigma Applications
by W. H. Woodall, professor of statistics, Virginia Tech, and editor of ASQ’s Journal of Quality Technology
bwoodall@vtu.edu
Six Sigma projects require the collection and analysis of data. Thus, the use of statistical methods is an integral part of Six Sigma initiatives. The best statistical methods depend on the particular circumstances of the applications of interest. In a given application, methods based on more realistic models of variation tend to be better. I would argue the best statistical method from an applied perspective is the simplest one that provides reliable and complete information.
There are situations in which Six Sigma Black Belts (BBs) need more in-depth statistical knowledge than what is usually provided in standard Six Sigma training. Roger Hoerl, manager of the applied statistics group at General Electric Corporate Research and Development, pointed this out in his October 2001 Journal of Quality Technology (JQT) paper (available at www.asq.org/pub/jqt/past/vol33_issue4). For those without access to statisticians within their company, Hoerl provided a list of recommended textbooks on important topics. I endorse his suggestions, although there are many other good choices.
Recommendations
In the statistical quality control area, Hoerl recommends Introduc-tion to Statistical Quality Control by Doug Montgomery (John Wiley & Sons Inc, New York, 2001). This book would benefit BBs trying to find the best statistical tools for applications. The book could also be used to gain a better understanding of Minitab software.
The references in Montgomery’s textbook contain many articles published in ASQ journals including JQT, its predecessor, Industrial Quality Control, and Technometrics.
The authors and the editorial review board members of these journals include many of the most knowledgeable applied statisticians. It is clear these journals have had a major impact on the state of knowledge in statistical quality control and can provide readers with a better understanding of the basic tools.
Recent papers in these three ASQ journals cover aspects of the following useful methods: split-plot experimentation, experiments with multiple responses, experiments with non-normal responses, fractional factorial experiments, mixture experiments, robust design, generalized linear models, measurement system assessment, optimization methods and many aspects of control charting including multivariate control charts and the handling of auto-correlated data.
A number of examples of direct relevance to Six Sigma practitioners can be found in recent issues of JQT. “Response Surface Methodology—Current Status and Future Directions” (January 1999) and “Controversies and Contradictions in Statistical Process Control” (October 2000) both won ASQ’s Brumbaugh Award for making the greatest contribution to the development of industrial applications of quality control. In addition, the January 2002 issue contained “Process Capability Indices—A Review, 1992-2000” with discussion by experts on this important area.
In his discussion of Hoerl’s JQT paper, Doug Montgomery and his co-discussants encouraged companies to use university courses in statistics as part of the continuing education component of training for BBs and Master BBs. These courses are another excellent resource that can provide knowledge of the best statistical methods. They can lead to a fuller understanding of statistical methods than the much faster-paced industrial short courses.
I strongly encourage BBs to further their knowledge of statistics using all available resources, including local expertise, recommended textbooks, journals, short courses and university courses. In my view, this advancement is required to help fuel the continued success of Six Sigma.
