We have described three experimental designs, all of which involve the removal of a successful intervention (i.e., one in which observations of target behaviors appear to be clearly moving toward...

We have described three experimental designs, all of which involve the removal of a successful intervention (i.e., one in which observations of target behaviors appear to be clearly moving toward desired levels) followed by a baseline period in order to test the potential causal effects of the intervention’s impact on the target behavior. Variations on this return to baseline make up the three designs: A-B-A, A-B-A-B,and B-A-B. Because of the successively unlikely occurrence that such variations in the target problem could have happened by chance, we have the basis for inferring a causal influence on the part of the planned intervention. This means that not only is the pattern of change documented, but control over that change is to some extent documented as well. This is the basis for a logical analysis of the events in question. This is a very powerful set of designs that you should consider carefully, evaluating each design for its own strengths and limitations. We also discussed the difficulties in using such designs, such as practical problems (there could be carryover effects from one phase to the next phase), ethical considerations (is it right to stop a successful intervention in order to attempt to understand causality?), and collegial problems (who wants to work with a troublesome client, especially after the problem appears to be resolved?). Thus, with logical power comes social dilemmas regarding practicality.