ON MAKING STATISTICAL ASSUMPTIONS