Text

theses

The dissertation examines how individuals behave when facing choices with uncertainty, both within individual and group settings. These questions are probed by examining individuals in experimental situations, with payoffs depending partially upon the choices they make as well as some fundamental uncertainty. In particular, the research seeks to answer, to what extent do individuals consistently evaluate risky gambles, and how do individuals respond to group decisions with either an underlying unknown state and a policy whose payoff depends upon that state, or a coordination task. For both group tasks, the role of communication in overcoming uncertainty is studied. Within group settings, individuals were tasked with a Median game, a coordination problem with efficiency achieved only at an asymmetrical equilibrium, and an aggregation problem where groups would vote on a risky policy whose success depended upon an underlying state of the world that each member privately received signals about. For the Median game, communication results in large gains compared to the sessions with no communication; an effect that grew as play continued. The types of messages and strategies communicated are analyzed to see which were most effective. Within a group setting, individuals were given private signals about the probability of payoff-relevant states of the world with payoffs determined by the policy voted for by the group’s majority. There was a strict incentive to aggregate information, however this was hindered by the inclusion of individual payoff biases, creating competing interests and raising a source for disinformation and thus distrust. The efficacy in achieving aggregation of various exogenously given networks was then tested. Within an individual setting, subjects evaluate scaled lottery questions with consistent aggregate responses that fit well within the established norms of Cumulative Prospect Theory. However, the aggregate data hide considerable individual inconsistency, with the individuals that fit best with a CPT interpretation those that are most variable and for which such a prediction would provide limited if any predictive power.

ETD_4658

Ph.D.

Includes bibliographical references

by Jeffrey Campbell Birchby

http://hdl.rutgers.edu/1782.1/rucore10001600001.ETD.000068821

doi:10.7282/T3TB15H5

ETD doctoral

The author owns the copyright to this work.