Social Connectivity, Media Bias, and Correlation Neglect

Abstract

A biased newspaper aims to persuade voters to vote for the government. Voters are uncertain about the government's competence. Each voter receives the newspaper's report as well as independent private signals about the competence. Voters then exchange messages containing this information on social media and form posterior beliefs, neglecting correlation among messages. We show that greater social connectivity increases the probability of an efficient voting outcome if the prior favours the government; otherwise, efficiency decreases. The probability of an efficient outcome remains strictly below one even when connectivity becomes large, implying a failure of the Condorcet jury theorem.