Can I add another input factor to D Optimal Custom Design after interpreting results?

Question

Hi everybody,

I have executed a study using a RSM methodology with 5 input factors, Custom design: D-Optimal, 28 runs and we realized during the interpretation of results that it would have been great to have included a 6th factor. Is it possible to "augment" my design at this stage and how can I design the extra runs? If you have had similar experience it would be great to hear from you. Thanks

Cristina_G · Answer

Hi Martin, good news since I already have some information on the factor level in the previous 28 runs. Glad to hear doing a reduced version that fills in gaps is possible. Many thanks for the support

Cristina_G · Answer

Hi Mark, the 6th factor was kept constant through-out the 28 runs and the level is known. Thanks for your answer, I will start from there and explore the Augment Design

martindemel · Answer

as Mark stated it is important if you have prior information of this parameter from the previous runs. If you have measured this parameter as additional variable, you can see if the model will tell you already something about its impact. If it was constant, this is not much but still some information.

If you have not measured it, nor information about what level it has during the 28 previous runs, you need to do as many runs as needed to estimate the main effect, 2-factor interaction effects of this 6th parameter with the previous five each, and the quadratic effect, to get a full understanding of its influence. (Of course having few more runs allows to estimate the error as always).

In case you know upfront specific interactions or quadratic effects cannot happen, you might want to adjust the model and reduce therefore the required number of runs.

Summarized, if you have already some information from the previous 28 runs you might need just a few more to fill the gaps, if you know nothing, you need to have at least one additional run for each additional model effect this parameter adds.