Abstract:
|
It has been emphasized on the variation reduction in industrial quality improvement. Thus, well designed experiments are usually needed to identify important effects so as to control the quality characteristic of interest around a target value with a minimal variation. Several methods have been proposed to analyze both dispersion and location effects from unreplicated factorial designs in the literature. However, the use of unreplicated designs can be inherently difficult in the identification of the truly active effects since location and dispersion effects are possibly confounded. In this study, we would promote compromised designs with partial replications. Based on the concept of generalized p-values, we develop a statistical testing procedure to identify significant dispersion effects from partially replicated designs. Subsequent identification of location effects is then performed using classical Student's t test. The proposed procedure is illustrated with a real data analysis, and evaluated through a detailed simulation study. The results support the use of partially replicated designs in the identification of active factorial effects.
|
ASA Meetings Department
732 North Washington Street, Alexandria, VA 22314
(703) 684-1221 • meetings@amstat.org
Copyright © American Statistical Association.