Evaluating exclusionary interventions

In evaluation researchs, some interventions are designed to affect both the subjects that receive the intervention and those that do not. Preferred provider organizations (PPOs) are an example, because if they are successful, PPOs will direct patients away from non-preferred providers towards preferred providers. When the intervention affects all subjects, the excluded group cannot serve as a control group if one wishes to estimate the experience of subjects in the absence of the intervention. That estimate must come from subjects completely unaffected by the treatment.