Research shows social media advertising suppresses voting in targeted communities
A new study is the first to quantify the effect of this kind of microtargeting on voter turnout.
Messages intended to suppress votes can be precisely delivered to particularly vulnerable and consequential groups of people via social media and keep millions of them from casting ballots, according to a new study that is the first to quantify the effect of this kind of microtargeting on voter turnout.
A team led by a researcher at the University of Wisconsin–Madison recruited more than 10,000 people across the United States — a group representative of the country’s voting population — to install an app that captured every ad they viewed for the six weeks leading up to the November election in 2016.
The study participants who saw the ads with vote-suppressing messages on Facebook were 1.9 percent less likely to actually vote in the election than people who did not see the ads, according to a study of the participants’ voting behavior published recently in the journal Proceedings of the National Academy of Sciences.
There were several types of targeted messages, but the most common suggested an election boycott would send the strongest message to politicians. The ads’ creators used Facebook’s microtargeting advertising features to reach mostly non-white, voting-age people in hotly contested states in the presidential election, such as Wisconsin. Those Facebook users received four times as many vote-suppressing ads as their white neighbors.
“In fact, those minority users got the vote-suppression messages nearly 10 times more often than white people in white-majority counties in non-battleground states,” says Young Mie Kim, UW–Madison professor of journalism and mass communication and the lead author of the study.
By obtaining state voter records, the researchers could see whether their study participants actually voted.
“We’re not basing this on surveys or modeling of aggregated ad distribution and expected turnout,” Kim says. “It took some time and hard work. The matching between voter suppression exposure and turnout indicates that those who were exposed to voter suppression were less likely to turnout to vote than those who were never exposed to voter suppression messages.”
While a 1.9 percent shift in the behavior of some voters is small, the researchers say, so were the margins of victory in many states in 2016. And, compared to white voters in white-majority counties in non-battleground states, nonwhites in nonwhite-majority counties in states with tight presidential races were about 14 percent less likely to vote. Extrapolated across the country for the 2016 election, the effect of the ads may have kept about 4.7 million people from voting.
Congressional investigators showed that many of the ads the research team identified as voter suppression ads were purchased by the Internet Research Agency, a Russian digital disinformation operation. They used terms including “Martin Luther King, Jr.,” “African American Civil Rights Movement” and “African American history” to target nonwhite voters on Facebook and discourage them from voting.
“These messages are designed to influence communities that already have a level of distrust of government and elections,” says Kim, whose previous work has examined the power of micro-targeting social media users and the effect of social media algorithms on political behavior. “Much of that is based on historical problems with access to voting. This campaign to suppress voting is a kind of digital-age version of limiting access to the ballot box.”
The source of the ads collected in 2016 would have been unclear to targeted Facebook users.
“None of the vote-suppressing ads were purchased by groups who had disclosed their political activity and filed reports with the Federal Election Commission,” says Kim.
Strengthening and enforcing federal regulations on disclosing the source of political messages could provide important context for targeted social media users, according to the researchers, which included Ross Dahlke and Richard Heinrich of UW–Madison and Hyebin Song of Penn State University.
“Transparency is important to democracy. Awareness is key,” Kim says. “We should be helping voters be more aware of these targeted campaigns. And they should be able to understand who is trying to influence them — especially whether it is foreign influence.”
