While completely randomized experimental designs are always recommended to conduct experiments, there are situations where restrictions on randomization are introduced which may result in split-plot experiments. However, it is observed that experimenters often fail to recognize the importance of such restrictions on randomization and simply conduct split-plot experiments as if they are randomized. This may result in poor optimization of the experimental factors, and hence, the inferential space may be different from that of the actual purpose of the experiment. Examples are given to illustrate the importance of the role of restrictions in randomization in industrial experimentation.
展开▼
