Value of information to a group of decision-makers with different priors

The expected value of sample information (EVSI) depends on the decision-maker's (DM) prior beliefs. Some public information or advisory services, however, cater to entire populations of decision-makers, who differ in their prior beliefs (e.g., their degree of optimism). Evaluating the potential contribution of such services thus necessitates averaging the DM-specific EVSI's over the population's prior beliefs' distribution. We show how to accomplish that, exemplifying it for two forms of population heterogeneity. We then compare the exact method to an approximation that computes the EVSI for an "average" DM.