During the summer school, participants and experts work together in projects on a particular research topic with specific CSS methods. Some of our experts already developed a project idea on which they will elaborate together with participants during the summer school. Please find the list and summaries of projects offered at our CSS summer school on conflicts 2018 @ Jacobs University Bremen below. Participants may pick up a pre-designed project or propose their own ideas in their application.
Projects of the summer school will relate to the following fundamental research questions:
- When do societal conflicts escalate and turn violent?
- What is the role of economic prosperity, distributive inequality, oppression of freedom and human rights, and ideological attitudes and beliefs in escalation?
- Can one detect escalating processes early on and are there early warning signals?
- How does violence spread spatially?
- How is social cohesion affected by non-violent conflicts or their oppression?
The projects offered at the summer school:
- Values in everyday language and inter-group conflict
- Socio-ecological tipping points. Quantifying the link between environmental and climate change and conflict
- Simulating the Economic Impacts of Mafia
- The role of homophily in the emergence of social norms within social networks
- Analyzing conflict dynamics at the event-level
- The timing and execution of terrorist attacks. The impact of city dynamics, inter-group learning and state opposition
- Between Cooperation and Conflict. Modeling Fine-Grained Textual Revision Changes as Social Interactions
Values in everyday language and inter-group conflict (PDF)
Inter-group conflicts often revolve around cultural differences between groups (Dobewall, & Strack, 2011). We argue that people’s perceptions of their own cultural values as well as of other groups’ cultural values are accessible by means of an analysis of short informal statements on social media (for example, internet postings; Holtz, Kronberger, & Wagner, 2012). Within our project, we will relate value attributions to social conflicts such as debates on migration in Germany and other European countries.
We want to construct a value dictionary based on the work of Christen and colleagues (2016) and use it to analyze discourse on social conflicts such as migration related debates. We will collect data from popular channels such as political party web pages and programs, open comment sections in online newspapers, Facebook, Twitter, and LinkedIn and use several software tools for linguistic analysis, such as word counting (e.g., LIWC), topic models (using the statistical programming environment R), and comparative keyword analysis (using software tools such as AntConc) to relate patterns of linguistic expressions to value expressions and value expressions consequently to conflict related discourse patterns.
Christen, M., Narvaez, D., Tanner, C., & Ott, T. (2016). Using thesauruses as a heuristics for mapping values. Cognitive Systems Research, 40, 59-74.
Dobewall, H., & Strack, M. (2011). Cultural value differences, value stereotypes, and diverging identities in intergroup conflicts: The Estonian example. International Journal of Conflict and Violence, 5(1), 211-223.
Holtz, P., Kronberger, N., & Wagner, W. (2012). Analyzing internet forums: A practical guide. Journal of Media Psychology, 24(2), 55-66.
Socio-ecological tipping points. Quantifying the link between environmental and climate change and conflict (PDF)
Expert: Davide Natalini
Research focussing on the link between climate and environmental change, and human conflict, found a significant, positive relationship between the two (Homer-Dixon 2001, Hsiang et al. 2013). However, how the effects of climate and environmental change are transmitted through the system and translate in increased conflict remains unknown. Unsustainable environmental change is chaotic and difficult to predict, often characterised by nonlinear behaviours and tipping points. When these are surpassed the system generates a shock that reverberates through the system and impacts society in complex ways, mainly through conflict and general loss of wellbeing. An example of this is the 2008 multisystem crisis which was environmentally driven and resulted in national food and fuel riots.
This project will further develop the Food and Fuel ABM in different possible directions. The final option will be selected with the students that will join the project:
- Option 1: Identify additional environmental variables to find a connection between environmental scarcity and food and fuel riots.
- Option 2: Test the effect of other socio-economic variables at different scales on food and fuel riots, to find additional multi-scalar, socio-economic drivers of riots (e.g. volatility in international prices of natural resources).
- Option 3: Extend the original quantitative analysis on drivers of food and fuel riots from Natalini (2016) to different types of environmentally-driven conflict, e.g. civil wars.
Homer-Dixon, T.B. 2001. Environment, scarcity, and violence. Princeton University Press.
Natalini, D. 2016. Estimating dynamics that lead to food and fuel riots: a quantitative and agent-based modelling approach. PhD thesis.
Simulating the Economic Impacts of Mafia (PDF)
The aim is to understand if and how organized crimes, in specific Mafia-type organizations, impact the economy. We usually assume that Mafia only influences negatively the economy, but there are some works, such as van Dijk (2007), showing that sometimes these organizations boost some aspects of the economy giving the impression that they are beneficial to the society. Hence, this model aims to serve as a tool (1) to evaluate different social scenarios and individual reactions to extortion to check these conclusion about the different impacts Mafia has on economy, and (2) to identify measurements/metrics that helps us to check the possible benefits and damages caused by Mafia.
We may use different survey data to define the input parameter values of the simulation model about level of crime, strength of the State, and entrepreneurial behaviors. For instance, use surveys from World Economic Forum (“The global competitiveness report”) and other European Social Survey to inform the level of activity of organized crime (or at least a perception of this type of groups) and the level of Police enforcement against them.
van Dijk, J. (2007) Mafia markers: Assessing organized crime and its impacts upon societies. Trends in Organized Crime 10:39-56. DOI: 10.1007/s12117-007-9013-x
The role of homophily in the emergence of social norms within social networks (PDF)
Expert: Fariba Karimi
In our research project, we aim to study the impact of homophily in the emergence of social norms within networks among actual humans, instead of simulated agents. Participants will be members of societal groups with clearly distinguishable group identities and demographics (e.g. religious Muslims and Christians of varying sex and age). The participants will be placed in a virtual network with each participant representing a node. Importantly, participants will only be able to see the demographic information of nodes connected to them in their intermediate vicinity. With the instruction for participants to maximize connected networks of the same norms (Axelrod, 1986), the network will iterate two steps. First, participants will be asked to choose a behavioral option related to an arbitrary social norm without pre-existing convictions (“What color of shirt should people wear on Fridays? Green Blue or Red?”). Second, people will receive feedback about the choices of their neighboring nodes. Afterwards, people can adapt their choices to react to the feedback from the neighboring nodes. Iteration will stop after connected networks of norms have converged. To test the impact of homophily on the emergence of norms, we manipulate the connectivity among nodes based on their demographics (Lee et al, 2017).
Lee, E., Karimi, F., Jo, H., Strohmaier, M., & Wagner, C. (2017). Homophily explains perception biases in social networks, arXiv preprint arXiv:1710.08601
Analyzing conflict dynamics at the event-level
Contemporary conflict research relies on a wealth of previously inaccessible data. Actors, locations, and events can be described in great numbers at unprecedented levels of detail. The earlier burden of data acquisition has largely been replaced by the increased challenges of data analysis. In the proposed research project, we introduce students to the basics of geographic event data analysis. In a second step, we give an outlook on inferential and predictive methods for working with event data. Finally, we introduce new methodology directly from the research frontier. Karsten Donnay has pioneered and implemented novel approaches for integrating event data sets. Moreover, he has developed software for automatically joining events with geographic variables from other data sets. Sebastian Schutte and Karsten Donnay have co-developed methodology for the analysis of reactive patterns in conflict event data. Students enrolled in the class will be acquainted with these approaches in a step-wise fashion and are encouraged to purse their own substantive or methodological research throughout the class. Weaknesses of existing methodology and areas of future extension are introduced toward the end of the class.
The timing and execution of terrorist attacks
Expert: Adam Robert Pah
The escalation of attacks from terrorist groups, in frequency and/or casualties, in any city or region poses a distinct threat. While it has been hypothesized that there is a dynamical process at play within a single city or region (Johnson et al. 2011) that pits terrorist groups against the state actor that would produce an ‘arms race’, it is unclear how much inter-group learning occurs. Further, it is unknown how much the dynamics of a city and its populace (such as holidays, festivals, protests, etc.) impact the planning and execution of attacks, as opposed to the process between state and terror actors. This work will focus on decoupling and quantifying the extent that natural city dynamics, inter-group learning, and state opposition impact the timing and execution of terrorist attacks.
Between Cooperation and Conflict. Modeling Fine-Grained Textual Revision Changes as Social Interactions
While every minute action in the collaborative creation of digital artifacts on the World Wide Web is recorded, we still know too little about how conflicting stances between collaborators get resolved and what value or harmful effect they have on the produced content or the social dynamics of the system.
In this project we will explore a large, novel corpus of data on several language versions of Wikipedia, containing all changes ever made by editors to content of other editors, complemented by additional interaction data like talk page discussion threads.
The participants will learn how to handle this data and will explore different ways to classify interactions between editors on the spectrum between cooperation and outright antagonism, ending either in a consensus or a “winner-takes-all” result. To this end, participants will expand on existing human assessments of interaction between editors via crowdsourcing and use text mining and machine learning approaches to build models that can explain or predict these interactions.
A certain pre-existing text mining and machine learning expertise is preferable.