According to the theory of deliberative democracy, deliberation is aimed at producing reasonable and well-informed opinions in which subjects are willing to revise their views in light of discussion, new information and claims made by other subjects, thereby producing optimal collective decisions. At the same time, social psychological research on group processes has revealed that group interaction involves various sources of bias that causes the collective decision to deviate from the optimum. We translate this problem into a group decision experiment on estimating unknown facts. Participants get paid according the distance of their final estimate to the true value. This mimics an expert committee, which is to give a good collective advice. We compare a simple aggregate of individual estimates to the outcome of a deliberation processes. We distinguish three conditions: (i) no required quorum where each individual can comes up with an individual estimate, (ii) majority voting where a majority has to agree on a common value which is the basis for payments for all participants, and (iii) unanimity, where all agents have to agree to receive any payment.
