DOE—It's Not Just for Widgets Anymore

Minitab Blog Editor 30 January, 2012

frankensteinWhen you think of design of experiments (DOE), what types of applications come to mind? Do visions of camshafts, widgets, capacitors, resistors, and other industrial thingamabobs dance in your head?  

If so, that's probably because DOE has such powerful and successful applications in manufacturing. Those experiments often involve changing levels of physical factors, such as temperature or pressure or speed or material, and then identifying the settings that produce the optimal effect. 

So a designed experiment can raise the spectre of Dr. Victor Frankenstein in the laBORatory, madly pulling levers, flipping switches, and turning dials until the “product” comes to life.

It's easy to forget that DOE also has important applications outside of manufacturing.


Experiments to Optimize the Human Response

One area of enormous potential in DOE is evaluating factors that may affect how people respond.  

For example, in a study funded by the National Cancer Institute, investigators used a fractional factorial design to evaluate an online smoking cessation program. They examined how five factors related to the design of their Web site affected the success of the program (measured by the ability of a participant to stop smoking).

One experimental factor was the quit-smoking success story featured on the web site. This factor had two levels, high or low, depending on how much the story was customized to reflect  the user's individual profile.



Low Setting: Success Story 

For this setting, only name and gender used in the success story were customized to match the user.
low setting



High Setting: Success Story 

For this setting, multiple characteristics used in the success story were customized to match the user.

high setting

Why was this important to evaluate? A high degree of customization is more complex and requires more resources to develop and maintain. If it doesn't produce a significant difference in user response, why waste the time and money?
(As it turns out, the study found that the "high" setting of this factor was associated with a statistically significant increase in smoking abstinence for participants in the online program.) 



DOE and the Service Industry

The human response is ultimately what drives business, of course. That's why the service industry is increasingly realizing that DOE can be a powerful tool to help improve quality and performance.



  • A national carpet retailer increased sales 20% by determining the salesperson/customer interactions that optimized customer purchases.
  • A financial services company saved over $3 million applying the results of a designed experiment on methods to collect unpaid debt.
  • A global newspaper evaluated 11 creative and 4 price elements in one mail drop to achieve a 41% increase in net response.


The first step is to broaden your conception of what constitutes a  “factor," a "factor level," and a “response.”  Think outside of the manufacturing box. 

Remember, a designed experiment is simply a highly efficient, scientific way to discover how changing levels of different factors affects a response—and that can be helpful in many contexts, not just in manufacturing or B-grade horror movies.

Tip: To learn more about applications of DOE for service quality, including the examples listed above, read this article co-authored by Minitab trainer Lou Johnson