Estimating spatial dimensions of networks using exponential random graph models

Rapid urbanization in the global south has been accompanied by remarkable changes in residential settlement patterns, with cities as the sites of intense contestation over essential services. Meanwhile, local communities in rich countries are beginning to grapple with their response to the localized impacts of climate change. But how well can neighbors coordinate their political activities in response to common needs? The answer may lie in the interaction between patterns of residential settlement and socio-political contact. Sociological theory posits that the strength of the relationship between proximity and contact should vary by the type of interaction, but much remains to be known about the topographies of local networks and their relationship with the capacity for political mobilization. To test these relationships, I use exponential random graph models (ERGM) to estimate the dyadic spatial parameters of social, economic, and political networks in two very different contexts: Indian urban slums and Ugandan villages. I find that social networks are quite localized, but the spatial patterns of economic and political networks are more complex and context-specific. Moreover, using the spatial nature of the data, I test the relationships between network homophily and ethnic segregation. The results have implications for the capacity of communities to mobilize in response to common shocks.