The present paper examines the importance of integrating geographic contextual effects into the analysis of social networks. By considering spatial structures as both produced by and productive of social relations, geographic space seems to be more than the extent on which places, actors or events are located and separated by distance. Territoriality, bordering processes, the sense of place, spatial inequalities, scalar relations and spatial connectivity are among the socio-spatial arrangements and practices that are likely to affect social action. The present empirical analysis thus focuses on policy interactions within the cross-border region of Lille because the spatial dimension particularly influences relations in this area. Specifically, we examine three spatial effects, namely, distance, territorial borders and cross-border territoriality, and we use exponential random graph models to model how these contextual variables influence policy interactions. By addressing multiple spatial effects, we develop a specific approach to control for the interactions that occur between these variables in order to elaborate on the complex processes that lead to the formation of social networks. We also explicitly examine how the spatial interaction function is affected by including in the analysis endogenous network effects, exogenous covariates and border factors. In this regard, we use a novel Monte Carlo-based goodness-of-fit summary in order to demonstrate that the predicted spatial interaction function of our model ? net of other effects ? matches the empirical spatial interaction function.
|Number of pages||36|
|Publication status||Published - 2013|
- Policy networks
- exponential random graph model
- spatial effects