An experiment is defined as a random variable which may take some posterior probability distributions according to a marginal probability. Elementary properties of this definition with respect to information value theory are derived as well as their practical implications.