by J. Marcus Jobe and Tom Critzer

Journal of Data Science, v.6, no.2, 247-259

Abstract

Complexities involved with identifying the projection for a specific set of k factors (k = 2, ... ,11) from an n-run (n = 12, 20 or 24) Plackett Burman design are described. Once the correct projection is determined, difficulties with selecting the necessary additional runs to complete either the full or half fraction factorial for the respective projection are noted, especially for n = 12, 20 or 24 and k = 4 or 5. Because of these difficulties, a user-friendly computational approach that identifies the projection and corresponding necessary follow-up runs to complete the full or half fraction factorial is given. The method is illustrated with a real data example.

Homepage | Table of Contents | Full Text of This Article