Quantifying the Returns to Persuasive Message-Targeting

• In Event: Media Campaigns and Persuasion

Abstract

Popular culture and the media have portrayed political campaigns as sophisticated users of persuasive message-targeting, defined as communicating different messages to different voter subgroups to maximize the campaign’s persuasive impact. In this paper, we first present a theoretical analysis that clearly identifies the key parameters that govern the conditions under which a targeting strategy will be effective. We then estimate these parameters using a large archive of campaigns’ own advertising experiments, comprising a total of 617 persuasive video ads created by 51 US Democratic Party campaigns, tested across 146 experiments and ~500,000 survey respondents. Finally, we present a simulation study to quantify the returns to persuasive message-targeting, taking into account the above-estimated parameter values, the financial cost of searching for subgroup-optimal ads, and the risk of mistargeting. We conclude that in our setting, targeting based on age, gender, education, or race/ethnicity would yield little to no benefit; targeting based on party identification could be beneficial, though we emphasize that this marginal gain is small relative to the boost from choosing strong overall ads.

Authors

• Ben Tappin

• Alexander Coppock, Yale University