Applicability and Implementation of Optimal Strategies in Partial Information Environments
Open Access
- Author:
- Challis, Michael
- Area of Honors:
- Economics
- Degree:
- Bachelor of Science
- Document Type:
- Thesis
- Thesis Supervisors:
- Kalyan Chatterjee, Thesis Supervisor
Sung Jae Jun, Thesis Honors Advisor - Keywords:
- Economics
Game Theory
Behavioral Economics
Mafia
Voting
Political Science
Imperfect Information Games
Race - Abstract:
- Human agents’ interactions in environments where there is an asymmetry in knowledge are of economic interest both theoretically and practically. What are the optimal actions an uninformed majority can take to counter an informed minority? How do actual people play in these situations? In this thesis we use the game “Mafia” as a vehicle to investigate these questions. This thesis introduces a new theory on the Mafia game which improves upon previously proposed “optimal” strategies. Simulations were created which corroborate these theoretical findings. Finally, results of the first laboratory experiments in the Mafia game, and analyses of these results are presented. We used experimental Mafia data to test the play of humans versus the proposed optimal strategies. There were no statistically significant differences between experimental and theoretical win rates (0.05 significance level), however, there was some evidence that the informed minority group won more than theory would predict. We created a second experiment where some of the player’s IDs were names common to a specific race. We use this data set to test if there are differences in people's play when biases are introduced. Once again there were no statistically significant differences between experimental and theoretical win rates. Therefore, there was no evidence that the voting patterns in the data set where we attempted to induce bias showed any difference from theory. This thesis approaches the game Mafia from two distinct directions – theoretically, and experimentally. This thesis presents the first such research on the Mafia game that bridges the gap between existing theory and experimental results.