Many socially relevant problems and research questions are related to the societal dynamics of conflicts, migration, and social cohesion. Societal dynamics result from a multitude of interdependent individual decisions. For instance, individuals do not form opinions, beliefs, and decisions in isolation but are exposed to social influence through social networks and physical, ideological, cultural, or emotional proximity to others. Individuals discuss about, protest against, conform with, or accept their social and cultural context. They may decide to relocate or to protest. They may help, reject, or ignore others. All these activities leave digital footprints in official records, through carried devices (e.g., smartphones), and (social) media on the Internet.
The exploration and merging of traditional and new data sources promises to reveal new insights about societal dynamics, as different data sources come with different strengths and weaknesses. Traditional sources of data, such as surveys, make use of sophisticated measures of opinions, attitudes, and beliefs, but they are very expensive, time-consuming, and in most cases provide only a snapshot of society. What is more, surveys fail to map social networks in society due to the complexity of network measurement techniques and data-privacy issues. New sources of data (smartphones, online) usually provide only rough measures of opinions, beliefs, and attitudes, but generate precise pictures of social networks and the dynamics in and of networks. It is a big challenge to merge these complementary data sources without violating privacy rights.
Data-driven modeling allows to merge traditional and new data in a meaningful and ethically unproblematic way. In a nutshell, data-driven modeling integrates empirical information into formalized models of sociological theories. Data are used to calibrate the theoretical model and to tailor it to the social setting that is being studied (cf., Hedström, P., 2005. Dissecting the Social: On the Principles of Analytical Sociology. Cambridge: Cambridge University Press, Chapter 6). For instance, information about the social network gathered on the Internet can be fed into the model, or data about the spatial distribution of socio-demographic and ethnic groups in a city. Next, data about individuals’ opinions and behavior gathered with traditional methods can be used to calibrate crucial parameters of the formal model. The empirically calibrated model can be used to develop hypotheses and predictions about societal processes and their outcomes.
All ingredients of data-driven modeling — data gathering, data merging, data exploration, theorizing and modeling, simulation, testing, validation, and calibration of models — demand high levels of expertise and experience. However, researchers are usually skilled in only some of these steps, based on their disciplinary background and methods training. In BIGSSS Summer Schools on Computational Social Sciences they will gather the necessary expertise, working in small teams on research projects within a stimulating environment of researchers with diverse backgrounds. Each team will be guided by two experts. The goal is to prepare a manuscript and submit it for publication. Participants can bring their own ideas and data to the school, but they can also pick up on topics and data provided by the expert guides.