The Effect of Issue-Oriented Facebook Advertisements on Voter Turnout

• In Event: Zoo Station: Social Media and Political Engagement

Abstract

Facebook advertisements were used by foreign and domestic entities in their efforts to interfere with the 2016 U.S. Presidential Election. However, empirical evidence as to the effect of such ads on voter behavior has been limited. This paper presents the results of an experiment conducted during the 2018 U.S. Midterm elections that measures the impact of longitudinal (7-week) exposure to issue-oriented Facebook advertisements on voter turnout on 870,000 unlikely voters in Texas. Conducted in partnership with a progressive c3 organization in Texas, the experiment tested exposure to one of four different advertising streams: ads pertaining women's health, health care, gun control, or immigration. Facebook's Custom Audience feature was used to micro-target ads to individual voters, using voter file data; turnout was measured using the public voter file.

Results show that assignment to the women's health messaging stream increased turnout by a statistically significant 0.47% relative to the control group. None of the other message conditions had an effect. An ITT analysis suggests that based on an estimated 40% treatment rate, the effect of treatment on treated subjects is approximately 1.18%. The effect of treatment was moderated by sex, in that the ads were only effective on women, and county, in that they were only effective in Republican-stronghold counties. This experiment provides empirical evidence that issue-oriented Facebook ads can impact turnout, but that a) the message of the ads matters, and b) effects are limited to individuals who are susceptible to such appeals based on identity and/or a dearth of other campaign mobilization efforts. Contributions to theory about the effects of micro-targeted advertisements are discussed.

Author

• Katherine Haenschen, Northeastern University