What’s the value in DOE if we don’t use a model to analyze the data?
How do we design experiments that are robust to different modes of analysis?
Optimal experimental designs require a model for optimality, is the same person that designed the experiment is not the one that models the data, optimal designs are … sub-optimal!
Thoughts like these creep in all the time.
Is it possible to have a designed experiment that is both efficient and robust?