On the Value of an Observation