Occurs when the behavior that is measured is in some way different from the behavior of interest. This is considered less valid than direct measurements.