Modeling Social Networks in Geographic Space: Approach and Empirical Application

Abstract

Social activities are responsible for a large proportion of travel demands of individuals. Modeling of the social network of a studied population offers a basis to predict social travel in a more comprehensive way than currently is possible. In this paper we develop a method to generate a whole social network for a given population focusing on friendship relationships. The core of the method is a friendship-formation model that predicts for any two given persons from a population the probability that a friendship relationship exists between the persons. The model takes into account the degree of similarity in attributes, geographic distance between the persons as well as threshold values representing the persons' opportunities and base preferences for engagement in a friendship relationship. We show how the model can be estimated on the basis of observed personal social networks from a sample of individuals. We estimate and test the model using data collected in a recent survey in Eindhoven, The Netherlands. The results indicate that the model is able to generate networks that display the same structural properties as we find in the sample data. A synthetic network generated in this way can be used to microsimulate social interactions in a population in geographic space