ABSTRACT
We now know a good deal about many crossover designs of interest to behavioral scientists. But, we have ignored many designs used in other fields of research. Furthermore, we have not really asked what makes a good design. From a psychological perspective, we care that we are studying an important problem, such as factors that control retrieval from short-term memory. We care most that the design is clean-that we are not completely confounding variables by manipulating two independent variables in paralllel when we think we are manipulating just one.
