Do Online Ads Sway Elections? Evidence from a Field Experiment

• In Event: Campaigns 2 - Advertising and Online Politics

Abstract

Do online ads sway elections? We provide evidence from an unusually large online field experiment conducted with a major European party during a recent national election. Randomly assigning 2,346 postal districts to Facebook and Google campaign ads, we estimate that the ads increased the party’s vote share by 0.5 percentage points. The persuasion rate was 2.7 percent with a cost per vote of $4. To parse out causal mechanisms, postal districts were randomly assigned to either candidate or programmatic ads. We find that programmatic ads were twice as effective as candidate ads, questioning a personality-centered model of voting.

Author

• Anselm Hager, University of Konstanz