> They don't have "measurement errors" Sure. I am not arguing that a median is not a median. By measurement error I really meant noise and disturbance, as def

> They don't have "measurement errors" Sure. I am not arguing that a median is not a median. By measurement error I really meant noise and disturbance, as defined in control theory. E.g. it is perfectly possible to have a value "average tax in county" with zero noise as in perfect value to the cent, but year to year variance due to disturbances.