Definitive screening: A new way to DOE things

Screening design of experiment has been an integral part of a scientist’s toolbox for many years now, but there have been some shortcomings with its use.  Classic screening designs have utilized two factor levels, a high and a low, which can cause the main effects to be confounded by two-way interactions, and the two-way interactions to be confounded by each other.  To tease out the unadulterated main effects, further experiments have been generally required which can slow the pace of development.  A new class of screening DOEs has been recently presented that resolves both of the interaction issues mentioned above.  These are known as definitive screens, and their primary difference is the use of three factor levels, low, mid and high.  A further advantage of definitive screening is that the number of experiments required for the design are only 2n+1 where n represents the number of factors under study.  With definitive screening, better data quality can be gained from fewer experiments, certainly a tool for the modern scientist.