ABSTRACT
In the early stages of investigation, particularly in industrial applications, screening experiments are often used to identify the few important factors affecting a system’s response. To this end, fractional factorial designs (Chapter 7) are common choices in settings that aim to investigate relatively many factors in relatively few trials. In many cases, such experiments are performed as completely randomized designs, and data analysis is conducted via regression and ANOVA approaches (e.g., Dean and Voss 1999; Montgomery 2008; Wu and Hamada 2011) or through the use of half-normal probability plots (Daniel 1959). However, complete randomization of the experiment trials may be impossible due to practical or cost constraints. Consequently, restrictions on the randomization are imposed, thereby complicating both the design and the analysis of the experiment.
