ABSTRACT
Any researcher, in the course of investigating the effects of a particular set of independent variables, must demonstrate that the observed changes in behavior are functionally related to the presence of those variables. That is, there must be a correlated change in the level of responding in an experimental subject with the introduction and, if a return to baseline conditions is possible, the withdrawal of experimental procedures. Experimental control is built on at least two concepts. First, baseline logic, as presented by Sidman (1960), suggests that some estimate of the pretreatment level of behavior must be incorporated in the design in order for one to have a benchmark against which to compare subsequent behavioral changes. In other words, what the level of responding would look like if the experimental procedures had not been attempted. The spirit of the second concept is captured by Baer, Wolf, and Risley (1968) in the statement “… replication is the essence of believability.” [p. 95] This would suggest that it is not merely enough to effect a behavioral change upon a single occasion. The researchers must show that this change may be repeated, again and again, either within subjects or in a number of subjects.
