On the Independence of Observations between Experiments