Tips for Analyzing Nonregular Fractional Factorial Experiments

Abstract

Plackett-Burman designs and other two-level nonregular fractional factorial designs are popular for identifying the active factors from a large list of experimental factors. Data analysis for these designs is straightforward if interactions can be safely ignored. However, rather than viewing interactions as a hazard, methods have recently been proposed to identify and estimate active interaction effects via nonregular factorial designs. The purpose of this article is to suggest approaches that the author has found useful and to caution against unrealistic attempts for finding many interactions. If interactions are anticipated, the use of designs with more runs than double the number of factors is strongly encouraged.

• All-Subsets Regression
• Effect Heredity
• Forward Selection
• Generalized Alias Length Pattern
• LARS
• Orthogonal Array
• Partial Aliasing
• Plackett–Burman Design
• Regression Tree
• SAMS

Additional information

Notes on contributors

Robert W. Mee

Dr. Mee is the William and Sara Clark Professor of Business in the Department of Statistics, Operations, and Management Science at the University of Tennessee. His email address is [email protected].