On resolving ambiguities of the multiple-baseline design: Problems and recommendations

The multiple-baseline design demonstrates the effect of an intervention by showing changes across separate behaviors (individuals or situations), when, and only when, a treatment intervention is introduced. The design attempts to control for the effect of extraneous events (history) by showing that specific changes are associated with the intervention at different points in time. When the onset of the intervention for one behavior produces general rather than specific changes (i.e., simultaneous changes across different baselines), the causal effect of the intervention cannot be unambiguously inferred. There is an inconsistency in the assumptions of the multiple-baseline design, namely, that changes due to treatment will be evident in specific changes (i.e., across specific behaviors with the onset of the intervention) and that changes due to extraneous or uncontrolled factors are evident in generalized changes. In fact, both treatment and extraneous factors can produce specific or generalized changes with interdependence of behaviors as a major determinant. The generalized changes need not interfere with drawing valid inferences about the effect of the intervention. Three recommendations are made for maximizing the likelihood of demonstrating a causal relationship between the intervention and behavior change when generalized changes across baselines are likely to occur.