Diffusion Through Networks of Heterogeneous Nodes in a Population Characterized by Homophily

We examine the impact that homophily can have on the diffusion of a phenomenon. We identify three mechanisms from the literature by which homophily can have an effect and model how they can change diffusion that is happening through social influence. By modelling and simulation we vary the size and composition of the initial seed of adopters who start the diffusion process -- the 'critical mass' -- and test this on simulated and real data. We then use real data on personal characteristics to model genuine -- rather than simulated -- homophily. Our main contribution lies in examining the impact that the composition of the critical mass has. When the critical mass group is small, a homophilious group will cause a phenomenon to spread further than a heterophilious group. As the critical mass group grows in size a chaotic period is entered where small variations in the composition will have a huge effect on whether a group of all one type or another will cause more diffusion. As the group size continues to grow a new pattern emerges where a heterophilious group will cause more diffusion than a homophilious group. These results are discussed and avenues for future research are identified