Abstract
Recent studies on subjective probability suggest that subjects systematically underestimate the value of information in a hypothesis testing situation. In the usual experimental setting of these studies, subjects are asked to estimate subjective probability about two hypotheses on the basis of incoming information.