Social Media and Democracy: Experimental Results

Abstract

Social media have become a main source of information for many voters. Political interest groups on social media platforms have the ability to (i) microtarget news based on individual-level voter data and (ii) obfuscate their identities, which can be exploited to spread disinformation. Two proposed interventions to prevent election manipulation by disinformation are a microtargeting ban and disclosure requirements. An empirical foundation for these interventions is missing. In a first set of treatments of our laboratory experiment, we study the effects of the implementation of a microtargeting ban and disclosure of interests in a social media environment on voting behavior. We find that disclosure of interests is essential to reduce the frequency of inefficient election outcomes. We also show that a microtargeting ban is redundant for ‘mainstream’ voters but is needed to improve voter decision-making of a smaller, ‘niche’ voter group. These findings are consistent with the result of our second set of treatments, in which we show that the transition from a traditional media environment to a social media environment makes all voters worse off. Lastly, we show that ‘niche’ voters are equally well off as ‘mainstream’ voters in games with public communication (one message for all voters) but stay behind in games with a microtargeted communication technology (one message for each voter group).