KNOWLEDGE OF DOE SELF ASSESSMENT FORM
FOR MEASURING SKILL IN DESIGN OF EXPERIMENTS (DOE)
Your name: ___________________________________________________________
Please check Yes or No and write a brief statement.
When you are assigned a project, do you use a well-
defined strategy for finalizing the question(s)
to be answered? Yes ___ No ___
What is your strategy? ___________________________________________
__________________________________________________________________
Do you know what synergy is (in a physical process) Yes ___ No ___
Is discovering synergy beneficial to your company? Yes ___ No ___
What do you do to discover, and benefit from,
synergy? _________________________________________________________
Do you know what antagonism is? Yes ___ No ___
Is it beneficial to your company? Yes ___ No ___
What do you do to discover, and benefit from, antagonism? ________
__________________________________________________________________
Do you use the following experiment strategy?
Hold all controllable factors fixed except one,
and vary that one factor to see what its effect
is on the outcome . . . . . . . . . . . . . . . . . Yes ___ No ___
If not, what strategy do you use? _____________________________________
_______________________________________________________________________
Suppose you have a project in which there are
13 controllable factor settings and 8 measurable
outcomes. Do you know what combinations of the high
and low limits of your settings you would run to
discover how to set the 13 factors to optimize,
as nearly as possible, all 8 outcomes? . . . . . . . . . Yes ___ No ___
What experiment design would you use? _________________________________
(attach an additional page if necessary)
Suppose maximizing one outcome conflicts with
maximizing another;
Do you know how you would deal with that? Yes ___ No ___
How? _____________________________________________________________
__________________________________________________________________
Do you use a square diagram to represent
two-factor systems? Yes ___ No ___
Do you use a cubical diagram for three-factor
systems? Yes ___ No ___
Do you use diagrams when there are 4, 5, 6 and
higher numbers of factors in your system? Yes ___ No ___
What kind of diagrams? ___________________________________________
After you have run experiments at the corners of
a cube, do you know how to use the data to
interpolate predictions between the experiments
you ran? Yes ___ No ___
How? _____________________________________________________________
If in your current project you were to run the
same combination of factor settings over and
over again, say 1000 times as you might during
production, do you know what the graph of the
measured values of your outcome (plotted
horizontally) versus frequency of occurrence in
the group of 1000 (plotted vertically) would
look like? Yes ___ No ___
How would it look? (either sketch or say in words): ______________
__________________________________________________________________
Is it of any use to have an idea of what that
graph would look like? Yes ___ No ___
Do you know what confidence limits are? Yes ___ No ___
Do you know what "95% confidence level" means? Yes ___ No ___
Do you know how to compute confidence limits
. . . for the long term average? Yes ___ No ___
. . . for the next single observation? Yes ___ No ___
. . . for 99% of future production? Yes ___ No ___
Do you know how to calculate standard deviation? Yes ___ No ___
I use "the data function" on my hand calculator Yes ___ No ___
I use this formula: s= _________________________________________
Other: _________________________________________________________
What is the greatest number of factors you have studied in one
experiment? ___________________________________________________________
If you use a hand calculator, please check its accuracy by computing s
for the following group of numbers. Please enter them exactly as they
are listed and exactly in the order listed below. Use the "(n-1)"
option, not the "n" option. If you wish to do it over again in a
different way, please do so in a second pass.
6.1159 What was displayed by your hand calculator
6.1159 when you asked for standard deviation? ________
6.1160
6.1161
6.1158
If you did it a second way, (again using the
(n-1) option)what did you get for s? _________
Does an average get better as the number of
replicates going into the average increases? Yes ___ No ___
Why? _____________________________________________________________
Do you know what an "interaction" is? Yes ___ No ___
Definition: ______________________________________________________
Do you know what "degrees of freedom" are? Yes ___ No ___
Definition: ______________________________________________________
Are you already using contour plots? Yes ___ No ___
What model will a "cube-corners" experiment design
support? "Cube corners" is also called "factorial." ______________
When only a fraction of the corners in a factorial
design are run, money and time are saved; what
opportunities are lost? _______________________________________________
_______________________________________________________________________
Do you know how to interpolate predictions of product
performance,for thousands of experiments you
did not run, from the data from experiments you
did run? Yes ___ No ___
Do you know how to avoid the problems that come from
a poor choice of run order for your experiments? Yes ___ No ___
What problems? ___________________________________________________
Do you know how to prepare contour plots from your data? Yes ___ No ___
If the polynomial model with main effects and
interactions does not adequately mimic your data,
do you know what you can do to get a better model? Yes ___ No ___
If yes, what can you do? _________________________________________
Do you believe you should set the limits on your
factors far apart? Yes ___ No ___
Why: _____________________________________________________________
Do you know the benefit from including many factors
in one experiment? Yes ___ No ___
If yes, what? ____________________________________________________
__________________________________________________________________
Can confidence limits be stated for predictions for
factor level combinations that you did not run? Yes ___ No ___
How? _____________________________________________________________
Do you know what conditions must be satisfied for
these limits to be safe to use to guide
management decisions? Yes ___ No ___
If yes, what conditions? _________________________________________
Do you know what is meant by pooling standard
deviations? Yes ___ No ___
If yes, when is pooling safe to do? ______________________________
__________________________________________________________________
If yes, what benefit is there in pooling? ________________________
__________________________________________________________________
Do you know what is meant by "uniform standard
deviation over the region of the design?" Yes ___ No ___
If yes, how do you test for uniformity of
standard deviation? ______________________________________________
__________________________________________________________________
How many times in a project should you run a
combination of factor levels over again
to get an adequate estimate of the
reproducibility of your process? Number of times: ___
Why is that number of times your choice? ______________________________
_______________________________________________________________________
If your confidence limits are too wide to be
helpful in making a management decision,
what can be done to make them narrower?
1. _______________________________________________________________
2. _______________________________________________________________
3. _______________________________________________________________
Are you already familiar with Plackett-Burman (PB)
main effect screening designs? Yes ___ No ___
In what sizes do they come (that is, numbers
of experiments to be run)?
__________________________________________________________________
How can you eliminate the contamination of the
main effects by the interactions at minimum
extra cost when using Plackett-Burman designs? ___________________
__________________________________________________________________
Are you familiar with the Box-Wilson central-
composite (CC) designs? Yes ___ No ___
What criterion should be used for deciding
whether the axial ("star") points are to be
on the cube faces or on the sphere passing
through the corners of the cube? _________________________________
__________________________________________________________________
Can some axial (star points) be on faces and
others on the sphere and others in between? Yes ___ No ___
Are you already familiar with Hardin-Sloane designs? Yes ___ No ___
Do you know what I-Optimal designs are? Yes ___ No ___
Do you know how to obtain an I-Optimal
design for your project? Yes ___ No ___
Have you been measuring interactions?
. . . 2 factor interactions? Yes ___ No ___
. . . 3 factor interactions? Yes ___ No ___
. . . 4 factor interactions? Yes ___ No ___
. . . 5 factor interactions? Yes ___ No ___
Up beyond 5 to what level? _______________________________________
How do you decide whether an interaction you have measured is having an
important effect on the outcome? ______________________________________
_______________________________________________________________________