33 Pages Posted:
Date Written: April 14, 2019
I study opinion dynamics in a social network to assess the effects of interventions. The combination of discrete opinions, stochastic updates, and stubborn agents leads to a non-degenerate steady-state. For arbitrary graphs, I characterize the expectation and variance of public opinion, as well as the rate of convergence to steady state. We obtain a new measure of individual influence and can assess the correlation between the views of any two agents. I provide analogous limit results for a model of random graphs, which enables a comparative statics analysis. Homophily and polarization result in smaller public opinion swings, which should make elections more predictable. Under standard voting mechanisms, integration of separate groups benefits the minority view. An extension shows how to evaluate the impact of targeted propaganda.
Keywords: Networks, Influence Campaigns, Voter Model, Stochastic Block Model
JEL Classification: D85
Suggested Citation: Suggested Citation