Demand characteristics and quality improvement: who is fooling whom?

Since some of the data used for quality assurance purposes (i.e. rating scales) requires the active participation of clinicians, administrators use various mandates or incentives to insure its collection. However, although improving clinician compliance these measures may bias clinician responses. It is suggested that the concept of 'demand characteristics' originally developed by Orne to describe how non-specific aspects of the experimental setting shape what the human subject does may be applicable. For example a measure that might increase clinician compliance with completing GAF ratings on an inpatient unit might also influence the scores to make them coincide with the expectation that all patients are very ill when admitted and improved when discharged. Why such a phenomenon would be difficult to detect and what it might say about the relationship between managers and those they manage is also explored.