Scribd
Upload
47 views2 pages

Applied Technology: by John J. Flaig, PH.D

The document discusses issues that can arise when experimental design is not properly planned and executed. It notes that just a few data points are not enough to establish a trend or process, and that experiments should not be abandoned too quickly without fully exploring the design and results. Careful experimental design and avoiding preconceived notions are important to develop robust processes.

Uploaded by

Ed
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
47 views2 pages

Applied Technology: by John J. Flaig, PH.D

The document discusses issues that can arise when experimental design is not properly planned and executed. It notes that just a few data points are not enough to establish a trend or process, and that experiments should not be abandoned too quickly without fully exploring the design and results. Careful experimental design and avoiding preconceived notions are important to develop robust processes.

Uploaded by

Ed
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 2

Quality Technology Corner

Applied Technology

www.e-AT-USA.com

Experimental Design Gone Awry


by John J. Flaig, Ph.D.
Consider the following DOE process development rule:
One point - is a Theory,
Two points - is a Trend,
Three points - is a Process!

Guess what? I wish this was a joke, but Ive seen it many times. The experiment gets started, a few
run results come in and they conform to someone's preconceived notion and BANG we stop the
experiment and roll the "improved" process into production only to get a nasty wakeup call later when
it fails. Or the experimental results are different from the preconceived notion (or they seem to point in
a new direction) and BANG every thing gets changed. Lets modify the experiment and explore a
different region of space. Now we generate a few more runs and the results indicate that the desired
response can be achieved by adjusting certain variables. This sounds good and conforms to those
preconceived notions about what should happen. Therefore, the "improved" process is rolled out, only
to find that the results in actual production differ significantly from the experimental model. Of course
this is explained by the "fact" that those ill-trained production operators can not follow instructions.

Reality check Yes, doing experimentation is a sequential process, but it is supposed to be a rational
sequential process. First, more effort should probably have been expended in design of the first
experiment. Second, the first experiment should not have been abandoned so quickly, because the
data from it might prove insightful. Third, when the second experiment is formulated it requires a

Quality Technology Corner

Applied Technology

www.e-AT-USA.com

significant amount of thought in the design phase. Also, it needs to be a complete experiment not just
a few runs.

Remember, three points dont make a process unless you are very lucky! Have you won the lottery
lately?

John J. Flaig, Ph.D.


Fellow of the American Society for Quality
Managing Director
Applied Technology
Tel: 408-266-5174
E-mail: [email protected]
Web: www.e-AT-USA.com

2004 Applied Technology. All rights reserved.

You might also like