Abstract:
Understanding the systems and processes that make up international relations and politics is difficult. Outcomes may appear unpredictable and the processes by which decisions get made may seem too complicated to understand or model. Decisions about complex issues like the decision to go to war or what type of government to establish involve many different interconnected parts interacting with one another and leading to emergent outcomes that have great consequences for countries, governments, and individuals. This dissertation uses agent-based modeling to investigate the ramifications of international alliances, on one hand, and the behavior and interaction of competing political parties and voters on the other, using empirical data and existing theoretical frameworks.I extended a well-known model for understanding emerging political actors using historical data on the network of formal defense alliances across the world. Based on a country’s status and network position, individual countries target one another, each utilizing plausible decision rules to decide whether to pay tribute or go to war. The resulting positions of countries in the international system, produced by the Global Tribute model, show broad qualitative agreement with the international system of the late 20th century. This work demonstrates that an agent-based model with simple decision rules can yield historically relevant results.
To investigate election behavior within a country I modify and extend, in various ways, a party competition model. Using empirical data comparing political parties I show that a three-dimensional policy space is sufficient to characterize much of the heterogeneity in ideological positions, and I extend the model from two to three dimensions. Then, based on the movement of parties in the ideological space, I incorporate voter decision rules for assessing which party they will support and if they will choose to vote. The results of this Voter and Election model show plausible results for voter behavior and are another example of how agent-based models with simple rules can be used in concert with empirical data.
This dissertation shows a useful methodological framework to understand how complex systems with many interdependencies work by connecting empirical data to relatively simple models. It demonstrates how abstracting these systems and focusing on their constituent components can help to understand systems’ outcomes and emergent properties.
