5.5.16.2.1 Definitive Screening Design
Contents
Introduction
Definitive screening design is used to efficiently identify key variables and their interactions in a process. It is a resolution IV design, meaning it can distinguish main effects from 2-way interactions without aliasing, providing a clearer understanding of the relationships between variables and the response. Unlike traditional screening designs, definitive screening designs allow for the estimation of some interactions and square terms.
When to Select
- Screening 6 or more continuous factors
- Needing to detect curvature (quadratic effects) in a single experiment
- Wanting to avoid follow-up designs to resolve ambiguities
- Minimizing run count while maintaining model-richness
When Not to Select
- Fewer than 4 factors
- Purely categorical factors (no center levels)
- Requiring precise estimation of all two-factor interactions
How to create the Design
Create Definitive Screening Design
- Select Statistics: Quality Improvement: Design of Experiment from Origin menu
- Select the Create Design icon and choose Definitive Screening Design from the list
Create Custom Screening Design
Please refer to the define custom design page for instructions
|
Topics covered in this section: |