Social network analysis has become an important vehicle in studying social phenomena. The trade has developed itself enormously over the last three decades. A common type of network study addresses how the structure of a network affects the actors comprising this network. In this type of study, networks are taken as an independent variable and actor attributes (such as behavior or opinions) are dependent variables. We will call this contagion. An issue far less commonly addressed deals with the question of how actors shape the structure of their network. Network structure now is the dependent variable, changing over time, and actor behavior is assumed constant. This process will be termed "selection". However, in many social situations, both processes will interact: actors will shape their networks and, simultaneously, are influenced by the structure of the network. When this is the case, separate analyses of either contagion or selection processes will be biased. In this paper we present an approach with the help of which it is possible to separate contagion effects from selection effects and estimate various aspects of contagion and selection. Applying simulation techniques, we assess the accuracy of the approach in a number of different situations.
|Title of host publication||Evolution of social networks|
|Publisher||Taylor and Francis Inc.|
|Publication status||Published - 2013|