For each response, you need to define a low and a high contour. These
contours should be chosen depending on your goal for the responses. Here
are some examples:
· If
your goal is to minimize (smaller
is better) the response, you may want to set the Low value at the point of diminishing returns,
that is, although you want to minimize the response, going below a certain
value makes little or no difference. If there is no point of diminishing
returns, use a very small number, one that is probably not achievable.
Use your maximum acceptable value in High.
· If
your goal is to target the response,
you probably have upper and lower specification limits for the response.
Because the graph displays the mean values for the response at the corresponding
factor settings, you should set your low and high values well within your
specification limits to ensure that individual observations fall with
the specification limits. If you do not have specification limits, you
may want to use lower and upper points of diminishing returns.
· If
your goal is to maximize (larger
is better) the response, again, you may want to set the High value at the point of diminishing
returns, although now you need a value on the upper end instead of the
lower end of the range. Use your minimum acceptable value in Low.
In all of these cases, the goal is to have the response fall between
these two values.
Defining Contours