2017 Graduate Workshop in Complexity and Computational Social Science

Student Projects

Each student began a research project during the two-week workshop. Below are brief descriptions of these various projects. These projects will form the basis for dissertation chapters and journal articles.

Therese Bennich, System Dynamics and Economics, Stockholm University, (therese.bennich@natgeo.su.se).

Therese is interested in how a more bio-based economy can emerge from existing industrial systems. Bio-based economies have the potential to provide both environmental and economic gains, yet they remain relatively rare. The municipality of Norrkoping in southeast Sweden has transitioned to a bio-based economy, and Therese has begun to model key parts of this region using system dynamics models. By using these models she will identify the key leverage points that allow industrial systems to transition into bio-based regimes with minimal effort and disruption.

Weikai Chen, Political Economy – Micro, Umass Amherst, (weikaichen@umass.edu).

Weikai is analyzing how different social networks can lead to inequality. Weikai set up a model where each direct connection in a social network comes at a cost, while the benefits to an agent are tied to both its direct and indirect connections. Depending on the levels of the direct-link cost and indirect-link discount, and benefit function, different network structures arise in the model---including the possibility of multiple possible structures under some sets of parameters. Given a network structure, he derives insights into the resulting level of inequality, shedding light on the deep history of the human experience.

Jaeho Choi, Management, University of Pennsylvania, (jhchoi@wharton.upenn.edu).

Jaeho wants to understand how redundancies can influence organizational decision making and search. His search space is defined by an NK model, and different units of the organization focus on, sometimes overlapping, parts of the search problem. When units differ in their desires for the state of overlapping choices, a manger steps in to resolve the conflict. Jaeho finds that in simple landscapes, redundancy leads to lower overall performance. However, in more complex landscapes redundancy can improve performance.

Jonas Dalege, Social Psychology, University of Amsterdam, (j.dalege@uva.nl).

Jonas is using an Ising-like model to look at how attitudes influence the spread of opinions. Agents have attitude positions and strengths, and adjust their positions tied to these variables, incoming information, and the influences of their neighbors. As the fundamental parameters change, he finds that the diversity and clustering of attitudes changes in predictable ways, ranging from the majority attitude prevailing across the entire space to "islands" of attitudes buffered by variable width boundaries of the neutral agents.

Nikolos Gurney, Behavioral Economics, Carnegie Mellon University, (nmgurney@gmail.com).

Nik considers a model where sellers can either truthfully reveal the quality of their goods or provide no information, and buyers must decide on an offer price based on the seller's behavior. To explore this model, he uses a simple model of buying and selling agents that adopt their strategies based on the profits they earn trading in a market. The system results in interesting dynamics whereby there are periodic disruptions that result in radical shifts in the strategies of the market participants.

Carolynne Hultquist, Geography and Social Data Analytics, The Pennsylvania State University, (hultquist@psu.edu).

Carolynne is developing models that allow one to understand better how various information feeds can be used to improve disaster relief efforts. There are a variety of new information feeds, such as twitter messages and the efforts of citizen scientists, that could be harnessed to provide more accurate data during natural disasters. To exploit such information, Carolynne is modeling how best to aggregate these sources given various limitations in availability and quality.

Jiin Jung, Social Psychology, Claremont Graduate University, (jiin.jung@cgu.edu).

Jiin generated a model of how minority attitudes can influence social change. Minorities influence the majority via indirect attitude change on non-focal issues that, given a desire for consistency, may ultimately influence more focal attitudes. Agents have multiple attitude states that influence nearby attitudes. Jiin then identifies the conditions necessary for the system to converge to the majority attitude, maintain attitude diversity, or have the minority attitude overthrow the majority one.

Linfeng Li, Behavioral Economics, University of Michigan, (llinfeng@umich.edu).

Linfeng is looking at bilateral games with endogenous interaction networks. In the model, agents must decide whom to connect with (at a cost) to play a game. If both agents agree to connect, they engage in either a coordination or anti-coordination game. Agent's adapt their behavior using a hillclimbing algorithm. He finds that the resulting networks have characteristic clustering and density, and that the agents tend to rapidly adopt strategies that result in outcomes with high total payoffs.

Mario Molina, Sociology, Cornell University, (mm2535@cornell.edu).

Mario is exploring the link between group identity and cooperation. He uses a model of public goods where individuals are differentiated by ability. Each individual must then decide on how much effort to contribute to the public good, as well as vote for an allocation rule that will determine each individual's share of the public good (ranging from equal shares to all to returning your exact contribution). The ultimate allocation rules is determined by the median voter. The resulting system produces interesting dynamics that link the resulting levels of inequality and cooperation to key factors such as the underlying distribution of ability.

Marie Schellens, Peace and Conflict Studies, Stockholm University, (marie.schellens@natgeo.su.se).

Marie is considering the impact of environmental variables on the potential for conflict. Starting with data from the Global Conflict Risk Index (1989-2010), she added key environmental data (such as arable land). She finds that the addition of environmental variables may significantly improve the explanatory power of the basic model. Moreover, using an alternative nonlinear estimation technique based on a three-layer neural network, she finds that the explanatory power of the original model (with and without the environmental variables) is much improved.

John H. Miller , miller@santafe.edu.