Several classical system dynamics models, such as models of disease spread, and technology adoption, are built under assumption of a homogeneous population. These assumptions have been recently challenged by recent results showing that the degree distributions of many social and natural networks, such as the so-called scale-free networks, exhibit long-tailed degree distributions. This paper adopts a system dynamics approach to replicate preferential attachment, one of the network dynamics mechanisms known to produce power-scale distributions. We then study the diffusion processes on these networks, e.g. epidemics, product adoptions. We consider a basic compartment model (Susceptible- Infected) and apply scale free network topology in place of the random network topology that is traditionally assumed. The resulting model is used to assess the effect of the topology on the diffusion of attributes throughout the network.