ABSTRACT
This chapter focuses on the design of experiments that will produce valid results. It examines how to design experiments so that researcher can get a given amount of information for a lower cost, or a greater amount of information for a given expenditure—that is, experimental efficiency. The chapter discusses randomized-group designs as well as designs that may be appropriate where random assignment of subjects to groups, or random selection of groups, is not feasible. A frequently used variation of the simple randomized-groups design was actually used by both Shevokas and Atkinson. Simple design is very efficient statistically because one can compare each subject’s before-and-after measurements, which holds consistent the original differences among persons. The chapter describes designs in which two or more elements of the experimental situation are systematically altered. The appropriate design experimentally treats the different samples different numbers of times and observes the results over several periods.
