Evidence from the 2016 US Presidential Elections

Through social media, politicians can personalize their campaigns and target specific groups of voters with an unprecedented precision. We assess the effects of such political micro-targeting by exploiting daily advertising prices on Facebook during the 2016 US presidential campaign. We measure the intensity of online campaigns using variation in ad prices charged to reach certain audiences, defined by political orientation, location, and demographic characteristics. We address two fundamental questions: How intensively did social media political campaigns target each audience? How large were any effects on voters? We find that micro-targeted political ads on social media had significant effects when based on geographical location, ideology, ethnicity, and gender. Exposure to these ads made individuals less likely to change their initial voting intentions, particularly among those who had expressed an intention to vote for Donald Trump. We also find that micro-targeted ads reduced turnout among targeted liberals, whereas they increased turnout and support for Trump among targeted moderates. JEL-Codes: D720, M370, D910.