| Since the deliberative turn in democratic theory began several decades ago, political institutions have frequently used small group discussion to create more deliberative policy. Political institutions based on discussion are also used as a way to aggregate information and produce better decisions. To produce these benefits, both normative and more tangible, small group discussion must protect the equality of participants. The equality of discussion depends on how groups use the information that their members bring to the table, be it new facts, perspectives, or moral and ethical arguments, to make decisions. If deliberating groups do not treat the information contributed by different deliberators equally, then their claims to democratic legitimacy come into question and their epistemic quality---their tendency to make good policy decisions---will be harmed.;Unfortunately, little research in the growing empirical literature on deliberation examines how groups use information to produce decisions, much less whether they do so in a way that treats their members equally. However, two existing research literatures suggest that group discussion is biased in the way that it uses information. First, the game-theoretic literature on strategic information transmission suggests that deliberators who have a strong interest in the discussion's outcome will use information strategically and that as a result groups will undervalue information provided by members whose preference is in the minority with respect to the preferences of other group members. Secondly, the psychological literature on the common knowledge effect suggests that the process of group discussion leads groups to undervalue information that fewer people know before discussion begins, creating a bias against group members who need to justify their positions using novel information and hampering the ability of groups to take this novel information into account when making decisions.;This dissertation tests whether these critiques apply to small group discussion about political issues using two experiments: a laboratory experiment that combines the methodologies of experimental economics and psychology and a field experiment that took place in deliberative forums organized by a school district in New Jersey. In both experiments I manipulate the composition of groups as well as the information that is given to different group members before discussion begins. I use this manipulation to test whether the identity of the person who knows an item of information before discussion begins affects how much influence that item has in discussion and over the decisions made by the groups. These experiments find that, contrary to some earlier evidence, deliberators use information strategically in face-to-face discussion, supporting the predictions of the game-theoretic literature. However, deliberators are able to overcome the tendency to over-emphasize information that is commonly known before discussion begins if they care about the outcome of deliberation, disconfirming predictions based on the psychological literature. This experimental evidence provides useful guidance for designers of deliberative institutions as well as insight on the roles of interests and group dynamics in determining the outcomes of group discussion. |
