ABSTRACT
The yoked control design involves pairs of subjects randomly assigned to experimental and control conditions. An experimental subject and its yoked control subject receive consequences as a function of the responding of the experimental subject. This chapter summarizes three articles that claim that, with this design, sources of random error (between-subject variability, within-subject variability, or variable effects of a consequence) can produce systematic differences between an experimental group and its yoked control. Such effects can appear in studies of punishment, avoidance, and positive reinforcement. Reactions to these claims are then reviewed. Despite some opinion to the contrary, the conclusion is that the design is flawed for making inferences about the influence of the contingency of the response and consequence, that the flaws are serious, and that they cannot be corrected by modifications of the yoked control design.
