10 Sept 2019, 17:00 – University of Zurich, Foyer L West, Rämistrasse 71, 8006 Zurich
Truth, Democracy, and the Social Network
Democracy, in part, depends on the premise that people make choices based on a collectively reasoned understanding of the world. Our understanding of the world, in turn, is mediated through our connections– to people, to institutions, and to sources of information. The logic of our connections – to people, and to various truth-constructing institutions – have been undergoing a remarkably fast evolution the last generation. This presentation will reflect on some of the emerging logics of the information ecosystem of the 21st century, and their implications for democracy.
David Lazer is a computational social scientist, and Distinguished Professor of Political Science and Computer Sciences, Northeastern University, and Co-Director, NULab for Texts, Maps, and Networks. Prior to coming to Northeastern University, he was on the faculty at the Harvard Kennedy School (1998-2009). His research has focused on themes around democracy, collective intelligence, social networks, and technology, and has been published in such journals as Science, Nature, Proceedings of the National Academy of Sciences, the American Political Science Review, Organization Science, and the Administrative Science Quarterly, and has received extensive coverage in the media, including theNew York Times, NPR, the Washington Post, the Wall Street Journal, and CBS Evening News.
Marijte van Duijn
9 Sept 2019, 14:00 – ETH Main Building, Audi Max (HG F 30), Rämistrasse 101, 8092 Zurich
Applied Statistics in Social Network Analysis
With the advance of statistical modeling techniques for the analysis of complex social network data, the complexity of the analysis and – especially – the interpretation of its results has increased as well. This development requires an effort from social network researchers, builders of new models and software and applied researchers alike. Taking the perspective of the statistical consultant/collaborator, I investigate – good and perhaps bad – practices in applying statistical models in social network analysis. The aim of the investigation is to gain more understanding of the process, not per se to find a “best” practice. One set of theoretical guiding principles can be found in statistical pragmatism as coined by Robert Kass, which focuses on the “big picture of statistical inference”. The big picture connects data and scientific and statistical models to be able to arrive at conclusions about research questions. This approach goes beyond the usual inferential approach based on samples from populations, which is difficult to apply in social network analysis. Another set of practical guiding principles can be found in the well-known setup of research papers, defining a research question leading to testable hypotheses, data collection method and description, statistical hypothesis testing and results, with –ideally – a discussion on the validity and generalizability of the research. The intersection of both principles revolves around the “results”, perhaps –statistically- significant, and their interpretation in relation to the research question. In tune with many other researchers, I would like to argue that this provides a too narrow view, especially in social network analysis. Taking a broader perspective, and building on both older and more recent research, the value of visualization becomes clear immediately. Second, a good understanding of the statistical models used is important as well. In view of the rapid developments in the last two decades, this requires a larger effort, both for the applied researcher and for the statistical “expert”. Simulation studies may be an important tool for this purpose. Moreover, the collaboration within a research team deserves some further attention as well, with respect to division of tasks and responsibilities. The talk will be illustrated with “real” applications and experiences.
M.A.J. (Marijtje) van Duijn is Associate Professor of Statistics at the Department of Sociology, Inter-Universitary Center for Social Science Theory and Methodology (ICS), University of Groningen. Her research interests are in social network analysis, statistical modeling, and especially the combination of these topics with a strong emphasis on applications including software. In the field of social network analysis she has worked on the p2 model as a first statistical model to analyze the friendship, advice and collaboration networks of the well known Lazega lawyers. She was a Senior Fulbright Scholar at the Center for Statistics and the Social Sciences at the University of Washington (USA). She is co-editor of Statistica Neerlandica, the internationally published journal of the Dutch Statistical Society.