8.5 Screening Designs
Whenever there are many factors with which to experiment, we often cannot afford to run enough tests to develop a model that includes interactions. As a Rule of Thumb, we recommend that whenever the number of factors (k) to be tested is greater than or equal or equal to 6, we should consider doing a screening experiment first and then follow that with a modeling experiment if needed. In screening experiments we simply desire to partition our factors onto the “vital few” and “trivial many.” The “vital few” are considered to be the most important factors and if subsequent modeling experiment is desired, these “vital few” will be tested in the modeling design. The “trivial many” are considered to be the less important factors which will typically be set based on economics, convenience, etc., and held constant during any future experiments on the vital few.