Identifying the Effect of Political Rumor Diffusion Using Variations in Survey Timing

Abstract

Despite growing concerns about the diffusion of political rumors, researchers often lack the means to estimate their effects. Field experiments seem infeasible due to ethical issues. Survey experiments typically invoke strong assumptions about homogeneous treatment effects across subjects and settings. We argue that exploiting temporal overlap between rumor circulations and survey interviews can be a useful alternative. We focus on an accidental and sudden spread of “Obama-is-a-Muslim” myths in September 2008. Using a difference-in-difference strategy that compares over-time belief changes of those interviewed for the September wave of the 2008–2009 American National Election Studies surveys before the rumor circulation and afterwards, we find that this event increased people’s belief that Barak Obama is a Muslim by 4 to 8 percentage points. To rule out various alternative explanations, we show that the treatment and control groups changed in parallel across waves in terms of an extensive set of placebo variables including political knowledge, other political misperception, and general attitudes toward Obama.