Agent Based Modeling for the Social Sciences

Instructor: John Gasper
Office: 313 Claudia Cohen Hall
Syllabus


An agent-based model is a tool used by social scientists to study how large-scale social dynamics result from micro-level individual behavior. In its basic form, a collection of individuals, or agents, are programmed into a computer, along with a simple behavioral algorithm describing how they act. The computer then allows the agents to interact (possibly in both space and time), and ultimately generates a social dynamic that can then be analyzed in terms of the behavioral rules. For more information please see the syllabus for the course.


In the course you’ll learn how to program your own agent-based model in Netlogo, a java based programming environment. While this course assumes little or no background in computer programming, a modest (minimal) background in statistics or some quantitative reasoning would definitely be helpful for analyzing the data generated by your model. Below you’ll find some of the final projects from Fall 2008.

Fall 2008 Final Projects:

 

Author(s)

Emma Edelman
Jared Miller

 

Microfinance: Developing an Agent-based Model

Abstract:
In the past 25 years, and especially over the last decade, microfinance has become a particularly popular and effective means of raising citizens in developing countries out of extreme poverty. Because the process of microfinance requires the formation and collaboration of groups, it lends itself to a study using agent-based modeling techniques. The goal of this paper is to use agent-based modeling to observe and measure the effects of several variables – mainly social pressure, interest rates, group size and the presence of natural disasters – on both bank and societal welfare in low-income communities when microfinance institutions are introduced.

 

Author(s)

Chris Poliquin

 

Fundamentalist and Chartist Trading Strategies in a Fixed Network Financial Market

Abstract:
This project examines how the properties of the network over which information is shared and confirmation bias influence strategy selection and prices in a two-asset financial market. Confirmation and less connected networks increase the use of trend-following investment strategies, which results in greater movements in price and higher trading volume.

 

Author(s)

Richard Norman
Landon Robinson

 

Agent-Based Model: Rush

Abstract:
This applet is a model of ‘off-campus’ fraternity rush at the University of Pennsylvania. In it there are many freshmen of various characteristic nature and one fraternity 'house'. The house throws different events throughout the course of the semester. At the end of the semester (ticks = 8), a "score" for the rush process is given to the house by the freshmen. Various variables such as number of freshmen, type of freshmen, and group leader rush on or off affect the course and outcome of rush. Though the model was designed specifically to model 'off-campus' rush at the University of Pennsylvania, it is relevant to all types of fraternity rush across campuses in the United States.

 

Author(s)

Tony Wang

 

An Agent-Based Model of Viral Marketing

Abstract:
Viral marketing, is it hot or just hot air? Is the only thing viral about viral marketing the hype it’s receiving from the academic community, the corporate world, and the media? This study presents an agent-based model to determine if viral marketing is significantly more effective than traditional marketing in the context of a closed world where two clothing companies compete with one another for consumer loyalty, one using only online viral ads and the only online traditional ads. The study concludes that viral marketing has a small, yet statistically significant advantage over traditional marketing as a whole. The study also finds that viral marketing does significantly better in settings with high populations and high levels of localization.

 

Author(s)

James Katz

 

Culture Wars: Survival of the Fittest

Abstract:
From a Darwinian perspective, cultures are in a constant battle with one another for attention from human beings in order to be integrated into people’s lives and passed on to the offspring of those people. However, the cultures that have been successful in perpetuating themselves continue to succeed today not only because some person liked it and thought it worthwhile to pass on that culture to their children, but because adhering to specific ideas within those cultures has contributed positively to the Darwinian success of their adherents, and in turn, to the Darwinian success of the culture itself. This agent-based model will show that a) It is theoretically possible for individual preferences to cause a sub-optimal non-Darwinian cultural trait to become the dominant one and that b) Not all Darwinianly-helpful cultural traits will necessarily lead a culture to total dominance.

 

Author(s)

Michael Durkheimer

 

The Hurdle Hypothesis

Abstract:

1. Why is broad relevance a predictor for New York Times article success, and why are other factors alone not predictors of success?
2. How does the social network affect which articles are more spread and which variables are more important to that success?

 

Author(s)

EJ Horlacher

 

Modeling Ethnic Tension: Red vs. Blue

Abstract:

I attempt to model group-on-group violent behavior to study ethnic conflict in developing nations. In my model I include as user-inputted variables the standard of living, groups’ respect for one another, and population spread. Action during the run is all turtle-generated in an assumedly anarchist environment.

My findings were that violence often burst in the beginning and then either tapered off, turned genocidal, or became an endless war. If threshold conditions were not reached, however, the model would stand still. Population spread proved to be relatively unimportant compared to the former two variables, which both had to be low in order for violence to occur.

Results differed depending on whether or not both groups experienced dissatisfaction with the situation. If only one group was unhappy, the violence would either flicker out or turn to genocide. If both groups were unhappy, the result would almost certainly be interminable civil war.