Design of Experiments

from the Perspective of Competitive Advantage

by
Edwin B. Dean

Montgomery (1991) notes that "A designed experiment is a test in which some purposeful changes are made to the input variables of a process or system so that we may observe and identify the reasons for changes in the output response. ... Experimental design methods play an important role in process development and process improvement."

Design of experiments (DOE) is the application of geometric principles to statistical sampling to obtain desired results such as minimizing the number of experiments necessary to obtain the answer to a problem or minimizing the variance of estimated coefficients obtained through regression. The first result above affects quantity of effort and, hence, cost. The second affects quality of result. Cost and quality are the basic elements of value. Thus, DOE is a powerful tool in designing for value.

Ko, Lee, and Queyranne (1995) note that D-optimal designs, the most efficient designs, can be obtained by maximizing the entropy of the covariance matrix of a design. They provide an exact algorithm, using branch and bound, for obtaining D-optimal designs. It allows expansion and contraction of existing D-optimal designs.

Response surface methodology, Taguchi methods, and conjoint analysis all use DOE as part of their process.

Address information for a number of companies providing quality related software can be found in Struebing (1996).

References

Ko, C., J. Lee, and M. Queyranne (1995). "A Exact Algorithm for Maximum Entropy Sampling," Operations Research, Vol. 43, No. 4, July-August.
Montgomery, D. C. (1991). Design and Analysis of Experiments, 3rd. ed., John Wiley and Sons, New York NY.
Struebing, L. (1996). "Quality Progress' 13th Annual QA/QC Software Directory," Quality Progress, April, pp. 31-59

Bibliographies

Design of Experiments

Table of Contents | Mathematical Technologies | Use