Moveon experiments with low-cost digital persuasion
During the November 2018 US midterm elections, Moveon conducted an experiment to test whether it could cheaply and quickly maximise the effectiveness of digital persuasion. The project created a Facebook app called MO Research, and recruited people to answer survey questions about current issues via targeted ads; 400,000 respondents answered an average of five questions each via Facebook Messenger, along with providing information about their hometown, gender, and age that allowed MoveOn to build demographic profiles and match them to their voter file records. After that, the researchers randomly divided the survey respondents into groups and targeted them with 2,500 ads of real voters explaining why they supported a particular candidate. Finally, they sent another batch of questions to find out how likely the targets were to vote for the candidates discussed in those ads and determine which were most persuadable. Initial results suggested that the advertised candidates' margins were 0.4% higher in the precincts where the ads were shown - enough to make a difference in close elections.
Writer: Issie Lapowsky