ABSTRACT
Determination of the proper design of the experiment to be used by a research worker depends upon the inference he desires and the randomization procedures available to him. If the experimenter wants to investigate the effects of a factor in a narrow area, say within his laboratory, the design may require only simple randomization and the completely randomized design is appropriate. If, on the other hand, he wishes the results to be applicable over an extremely wide range, say over the entire United States, the design may need to be quite elaborate and the completely randomized design may not be appropriate. References on these concepts include Cox (1958, Chapter 7), Cochran and Cox (1957, Chapter 4), and John (1971, Chapter 3).
