Berlin 2015 – wissenschaftliches Programm
Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe
SOE: Fachverband Physik sozio-ökonomischer Systeme
SOE 13: Opinion Formation, Segregation, and Language Dynamics
SOE 13.1: Vortrag
Mittwoch, 18. März 2015, 15:00–15:15, MA 001
Data-driven modeling in continuous opinion dynamics — •Jan Lorenz1 and Thomas Metz2 — 1Jacobs University Bremen, Germany — 2Albert-Ludwigs-Universität Freiburg, Germany
Models of continuous opinion dynamics as the bounded confidence model or related models with continuous opinions show interesting transitions and bifurcation patterns regarding their stable opinion distributions. The consensus transition which divides stable consensus from stable bipolarization at a critical bound of confidence is of course an interesting phenomena with analogies to real-world phenomena, but full consensus (polarization) on one (two) single point(s) in a continuum is probably not a state reached very often in groups and societies. A major model improvement overcoming this unrealistic convergence was the introduction of noise in the sense that agents sometimes start from scratch with a new random opinion. Nevertheless, all these models of continuous opinion dynamics models have rarely been compared with opinion landscapes from real-world processes of opinion dynamics. We present results from data-driven modeling of continuous opinion dynamics which start with the identification of stylized facts of opinion landscapes from large scale representative surveys and panels. Opinion landscapes indeed often show some clustering which bounded confidence models are able to produce, but the models typically do not reproduce the structure of real-world clustering patterns. We present simple modeling ideas which could bring models and data closer together.