Skip Navigation

Voter Suppression Has Gone Digital

The public should be able to tell where these messages are coming from and how they are being targeted

November 20, 2018

This month’s elec­tions were fought online to an unpre­ced­en­ted degree, with an estim­ated $900 million in digital ad spend­ing — more than two and a half times the 2014 midterms. Not all of this spend­ing was inten­ded to persuade voters to favor one side or the other, however. Some online activ­ity tried to keep people from voting alto­gether. My research team at the Univer­sity of Wiscon­sin-Madison found hundreds of messages on Face­book and Twit­ter aimed at voter suppres­sion — designed to discour­age or prevent people from voting. 

Online voter suppres­sion in 2018 showed simil­ar­it­ies to Russi­a’s inter­fer­ence in the 2016 pres­id­en­tial elec­tion. We found three categor­ies of messages: decep­tion about how or when to vote, calls to boycott the elec­tion, and attempts to threaten or intim­id­ate poten­tial voters.

Voter suppres­sion messages appeared online despite the plat­forms’ efforts to stop them. In Octo­ber, Face­book broadened its elec­tion integ­rity policies to combat voter suppres­sion and commit­ted to taking down such content. Just a few days prior to Elec­tion Day, Reuters repor­ted that Twit­ter also purged more than 10,000 auto­mated accounts that discour­aged voting. Clearly, the prob­lem has not yet been solved.

Has Voter Suppres­sion Gone Digital? 

Voter suppres­sion has been taking place in vari­ous forms over the years. Piven and colleagues demon­strate that voter suppres­sion tactics took the form of blatant viol­ence and intim­id­a­tion in the Recon­struc­tion era but have since trans­formed into regu­lat­ory devices such as voter ID laws and racial gerry­man­der­ing. More recently, voter suppres­sion takes the form of misin­form­a­tion campaigns and decept­ive prac­tices such as lying about the time, place, and manner of voting. 

Decept­ive prac­tices have tradi­tion­ally been comprised of fliers and phone calls. Now voter suppres­sion has truly gone digital. In the 2016 elec­tions, my research team found that the Inter­net Research Agency, a Krem­lin-linked, inform­a­tion campaign oper­a­tion, ran paid Face­book ads to suppress the turnout of nonwhite voters, espe­cially African Amer­ic­ans. The night before Elec­tion Day, ads appeared urging people to “boycott the elec­tion” because neither of the pres­id­en­tial candid­ates would serve black voters. The ads, sponsored by the Russian-backed duo Willi­ams & Kalvin, specific­ally targeted African Amer­ican voters. In another study, my team found that voter suppres­sion ads target­ing nonwhites resid­ing in minor­ity counties (counties where the propor­tion of nonwhites is more than 50 percent of the popu­la­tion) in battle­ground states (where 2016 vote margins were less than +/- 5 percent) were eight times larger than that of their coun­ter­parts. Consist­ent with the find­ings from our prior research, most of the voter suppres­sion ads were sponsored by groups that had not registered with the FEC and publicly shared no inform­a­tion about who they were. 

Voter Suppres­sion in the 2018 Midterms

My team monitored the midterm elec­tions in real time over the past several weeks by util­iz­ing publicly access­ible Face­book ads and Twit­ter. Unlike our 2016 digital ad track­ing and analysis, which was based on large scale, consen­ted-user-based track­ing inde­pend­ent from digital plat­forms (we collec­ted 87 million ads exposed to 17,000 users who repres­en­ted the U.S. voting age popu­la­tion), the inform­a­tion provided here is anec­dotal. As Face­book and Twit­ter proact­ively removed voter suppres­sion messages in real time, the data provided by the digital plat­forms do not include posts or accounts that were taken down. Still, we found notice­able voter suppres­sion campaigns online, espe­cially on Twit­ter. 

We specific­ally focused on three types of voter suppres­sion campaigns: decep­tion (lying about the time, place, and manner of voting); calls for boycott; and voter intim­id­a­tion or threats. 


The provi­sion of incor­rect inform­a­tion about the elec­tion date was very common in unpaid posts on Twit­ter. Similar to Russian tactics repor­ted in our previ­ous study, some tweets encour­aged people (more specific­ally, anti-Trump voters) to vote via text. 

Espe­cially, #voten­ov­em­ber­7th, a hashtag that mobil­ized turnout but with incor­rect elec­tion date inform­a­tion, was often paired with other hasht­ags designed for partisan mobil­iz­a­tion such as #blue­wave, #redwave, and such. We do not yet know who these users spread­ing misin­form­a­tion were exactly.

Many messages provided two differ­ent elec­tion dates — the correct one for their own party but the incor­rect one for the oppos­ing party. This type of message was tweeted by support­ers of both parties.

We did not find many paid ads contain­ing incor­rect inform­a­tion about the time, place, and manner of voting on Face­book. However, one ad target­ing Kansas is worth noting. The ad indic­ated that voters would need a birth certi­fic­ate or natur­al­iz­a­tion docu­ment to register. However, over the summer a court struck down the law requir­ing proof of citizen­ship. The ad was sponsored by a Repub­lican candid­ate and primar­ily targeted women. 

While many of these messages might have been circu­lated with innoc­u­ous intent, they are still worry­ing. Given the ampli­fic­a­tion poten­tial of social media messages, they may confuse a large number of voters. 

Demo­bil­iz­a­tion and Calls to Boycott

We also found some voter suppres­sion attempts to deter people from voting by saying that voting is point­less or that there is no differ­ence between the parties. Many messages used hasht­ags like #dont­vote and #dont­beavoter. 

While the target­ing of these calls to boycott was unclear in most cases (due to the limited inform­a­tion we have), some of the messages clearly targeted Latino and African Amer­ican voters. 

Voter Intim­id­a­tion

In the 2018 elec­tions, we found voter intim­id­a­tion and threats attempt­ing to organ­ize viol­ence at polling places.

After Nancy Pelosi’s mention of “collat­eral damage” on Octo­ber 18, 2018, a Twit­ter user named “Illeg­alAlien” said, “if you are GOP, be SURE to BRING YOUR GUNS TO THE POLLS. Let’s show COLLAT­ERAL DAMAGE.” This tweet itself did not have much trac­tion. 

However, after Octo­ber 25, when the Nation Rifle Asso­ci­ation’s spokes­per­son said on NRATV that gun support­ers would need to bring guns to the polls to protect them­selves from left wing mobs, tweets suggest­ing that NRA members or Repub­lic­ans need to bring guns to the polls star­ted trend­ing.

Voter intim­id­a­tion messages were often paired with misin­form­a­tion about voter fraud. We found many conspir­at­orial messages; for example, that George Soros organ­ized migrant cara­vans to rig the elec­tion, and that ICE was show­ing up at the polls to deport immig­rants.

What Should We Do? 

In both the 2016 and 2018 elec­tions, my research team has been observing vari­ous forms of voter suppres­sion campaigns take place on digital plat­forms. Despite tech plat­forms’ proact­ive meas­ures for combat­ing voter suppres­sion in the 2018 midterms, voter suppres­sion messages still appear to be spread­ing rapidly. 

Tech plat­forms’ volun­tary, self-regu­lat­ory policies combat­ing voter suppres­sion are a very import­ant first step. However, we need to consider consist­ent and system­atic regu­la­tions to effect­ively address voter suppres­sion campaigns. The policy response should be two-pronged: Decep­tion about how to vote should be banned, and other messages about voting should be trans­par­ent.

First, Congress should expli­citly define and prohibit decept­ive prac­tices like spread­ing false inform­a­tion about how or when to vote in order to keep people from voting. Although much of this type of conduct is already illegal, we need a federal law to clearly and compre­hens­ively address it. The Decept­ive Prac­tices and Voter Intim­id­a­tion Act, intro­duced this year in Congress, would accom­plish this.

Second, online messages about voting and elec­tions should be more trans­par­ent. The public should be able to tell where these messages are coming from and how they are being targeted. As my research has shown, most voter suppres­sion campaigns online were sponsored by undis­closed groups. And demo­bil­iz­a­tion frequently targets specific segments of the popu­la­tion in terms of race, gender, and income, poten­tially lead­ing to discrim­in­at­ory effects. Disclos­ure would discour­age some from enga­ging in voter suppres­sion and give the govern­ment, journ­al­ists, and civil soci­ety the chance to respond and counter suppress­ive messages. Before 2020, Congress should enact legis­la­tion that would provide clear, consist­ent guides for disclos­ure to address the issues described here.

(Image: Shut­ter­