The DeGroot–Friedkin (DF) model is a recently proposed dynamical description of the evolution of individuals’ self-appraisal and social power in a social influence network. Most studies of this system and its variations have so far focused on models with a time-invariant influence network. This paper proposes novel models and analysis results for DF models over switching influence networks, and with or without environment noise. First, for a DF model over switching influence networks, we show that the trajectory of the social power converges to a ball centered at the equilibrium reached by the original DF model. For the DF model with memory on random interactions, we show that the social power converges to the equilibrium of the original DF model almost surely. Additionally, this paper studies a DF model that contains random interactions and environment noise, and has memory on the self-appraisal. We show that such a system converges to an equilibrium or a set almost surely. Finally, as a by-product, we provide novel results on the convergence rates of the original DF model and convergence results for a continuous-time DF model.