This is a training document written for SEs about the optimality criteria (D, I, A, and Alias) used in JMP's Custom DOE. The document also contains discussion around contrasting optimal designs with classical designs and when you would consider something else other than an optimal design.
Authors: Tom Donnelly and Olivia Lippincott
Summary:
Sections 1 and 2 will introduce optimality criteria (D, I, A, and Alias) and discusses how model choice influences the type of optimality used and makes the case (when factors are continuous) for moving from 1st order to full 2nd order (including squared terms) rather than stopping at an interaction model.
Section 3 will explain via case studies the various optimality criteria (D, I, A, and Alias) used in JMP’s Custom Design platform.
Section 4 is provided for those interested in comparing Classical to Optimal designs and understanding many of the real-world issues that force users to drop their classical designs and turn to optimal designs to solve their DOE problems.
Section 5 focusses on the cases where screening designs like DSDs and GO-SSDs should be considered. It includes a discussion about augmenting DSDs to support efficient sequential experimentation.
Section 6 discusses making the analysis easy and the use of the Effects Summary in the Fit Model report.
Section 7 shares some quotes and lessons learned about models and running and analyzing designs.
Thank you both! This is wonderful
I echo @mia_stephens, so useful. Thank you for putting this together.
Thank you for this fabulous resource!
