Noise, bias, and expertise in political communication networks

Abstract A central focus in the study of social networks and politics centers on the dynamics of diffusion and persuasion, as well as the manner in which these processes are affected by expert “opinion leaders.” The role of experts is particularly important in communication processes characterized by noisy, biased information – processes in which people with variable levels of expertise and strength of preference select informants, as well as being influenced by them. We employ an experimental approach that addresses these problems at multiple levels of observation in a highly dynamic context – small groups of individuals communicating with one another in real time. The analysis shows that participants formulate candidate judgments that decay in time, but the decay occurs at a significantly lower rate among the better informed. Moreover, the better informed are less affected by socially communicated messages regarding the candidates. Hence the influence of experts is not only due to their powers of persuasion, but also to the durability of their own privately formulated opinions. Their role in the communication process is further heightened by the higher value placed by participants on expert opinion, which in turn exposes the recipient to a heterogeneous and hence potentially influential stream of information.