This is a type of sampling bias which can effect epidemiological studies testing the efficacy of particular therapies or interventions. The problem stems from the fact that the type of people who would volunteer themselves for a clinical trial and actually follow a regime to completion are not representative of the general population.