Applying Schelling’s Dynamic Models of Segregation to Psychological Decision Making

Agent-Based Models (ABM) have become accepted as an experimental technique in social psychology research. However, ABMs are rarely validated externally. This research attempted to experimentally validate, in particular the Schelling’s Dynamic Models of segregation. This model is an agent-based proximity model implemented with a new parameter, the tolerance level, which is based on perceived similarities of the agents. The introduction of this new parameter has made ABM applicable to social sciences. Five parameters (tolerance threshold, neighbourhood ranges, environment types, group size and proportions) of the model have been chosen to be tested experimentally and evaluated theoretically. Therefore, relevant chapters include an experimental and a theoretical section. In the experimental part group of participants played the linear public good game with altruistic punishment. The level of cooperation was measured through contributions scores. In the theoretical part the outcome of the simulation models was evaluated using the segregation factor. The experimental results and the simulation outcomes have been compared in each chapter separately. Five chapters inspected the five parameters of the Schelling model considering additional prospects. The tolerance threshold was tested using groups with different wealth origin. The neighbourhood ranges were tested using groups with different geographical proximities. The environment type parameter was tested using different gender groups. The group sizes parameter was tested using different community sizes. Finally the group proportions were tested by allocating the participants into groups with different ethnicity proportions. Additionally, the effects of these conditions were examined on altruistic punishment. The findings show that the contributions scores alter between the different scenarios. However, altruistic punishment is not always influenced. This thesis provides a partial validation of five parameters of the Schelling model and contributes to an effective development of all future ABMs. The research suggests unique recommendations for further research in both the experimental and theoretical parts.