Curious Cat Management Improvement Library - Dictionary

Design of Experiments (DoE) Designed Experiments - technique to analyze the effect of varying several variables simultaneously in order to get the most data with the fewest runs (each run generates the result from an the set values of the variables being studied). When not using DoE, experiments often vary only one variable in each run. Design of Experiments greatly reduces runs needed when varying only one variable, and even more importantly, captures interaction effects between the variables being studied.
Designed experiments rely on random test runs. The runs must be in a random order to avoid introducing bias into the results. However, when that is difficult or expensive (for example, if changing one variable requires shutting down the factory for 4 hours to accomplish the change) various methods can be used to use a non random pattern - though a risk is introduced that the results will be misleading.
Six Sigma has greatly expanded the use of Designed Experiments outside the scientific community.

Related Terms:

• Evolutionary Operations (EVOP) - Using DoE to model the "topography" of results. The experimenter designs each successive experiment based on the results thusfar.
• Randomized Order - assuring the order of the runs does not introduce bias into the results. Often the results can be effected by the current state. For example, if certain mechanical components wore out over a two week period and reduced the yield as time went by the results would be decreased over time. By randomizing the order the effect of types or bias can be reduced.
• 2³ Factorial - The variables are studied at two levels each resulting in a cube with results in each corner (see example). The simplest form of Factorial Design.
• interaction - extra effect on the result from the combined effect of variables in combination.

Online Resources:

• Curious Cat Library - DoE articles
• Three Romeos and a Juliet: - Our early brush with Design of Experiments by Ravindra Khare, Jul 2002. Very easy to understand explanation of 3 factorial designed experiment. No math or statistics knowledge needed.
• Linking High School Math and Science with Statistical Design of Experiments, Macomb Intermediate School District, Michigan, USA.
• Eight Keys to Successful DoE by Mark J. Anderson and Shari L. Kraber, Quality Digest, July 1999.
• 101 Ways to Design an Experiment, or Some Ideas About Teaching Design of Experiments by William G. Hunter, Jun 1975. A great article sharing ideas on teaching and learning about Design of Experiments (DoE).
• Strategy for Design of Experiments
• What is Design of Experiments? - C.F. Kavanaugh & Associates

Management Dictionary: Affinity Diagram, Kaizen, PDSA, Poka-Yoke, Six Sigma, Statistical Process Control (SPC), Variation

Curious Cat Directories of online resources: Investing - Management Improvement - Overuse of Antibiotics - Excessive Medical Costs - Identity Theft articles