Welfare-improving misreported polls

Abstract

We introduce an electoral pollster in the canonical pivotal voting model and show that the misreporting of pre-election poll results can happen even in the absence of partisan motives, as long as reputational concerns are present. By underreporting the expected number of supporters of the most preferred candidate in society, the pollster can induce an election result more likely to be in line with its report. By doing so, not only victory chances of the most preferred candidate rise above 50%, thus breaking the unrealistic neutrality result of the pivotal voting model, but also total election costs are reduced, thus yielding welfare gains and partially offsetting the expected negative effect of polls on welfare (see Goeree and Großer in Econ Theory 31:51–68, 2007; Taylor and Yildirim in Games Econ Behav 68:353–375, 2010). Our model also allows for the simultaneous accommodation of the underdog effect (a feature of pivotal voting models) and the apparently inconsistent bandwagon effect, in the sense that the latter can actually be understood as an illusion due to the possibility of misreporting being overlooked. All of these results hold even as the electorate size grows without bound.