Number of experiments for Definitive Screening DesignThe original paper describes the number of runs as 2m+1 runs. However, for a 4-factor screening evaluation, the DSD option on JMP provides an augmented default design with 4 more runs, i.e. 13 runs. What is the rationale?
Re: Number of experiments for Definitive Screening DesignThank you. The response is helpful.Re: Number of experiments for Definitive Screening DesignFirst of all, are you really screening four factors? Do you expect only one or two of them to be active?
Second of all, the power of detecting important effects for up to half the number of factors goes up very quickly with the addition of the four optional runs. The DSD is very economical but counts on the screening assumptions to hold. So if fewer than half of fhe factors are active, if the effects are 3-5 times the response standard deviation, the power becomes dramatically higher with the additional runs.
The additional runs are optional. They are added by default but you can change that behavior. Design identical experiments except include the extra four runs one time and remove them the other time. Compare the Design Evaluation information for both.
