I don't think that somebody who is observing or predicting behavior should also be participating in the 'experiment.'