Just under 3 million Americans are registered to vote in two places, but the overwhelming majority of them don’t vote twice, according to a new analysis by TargetSmart, the data firm used by the Democratic National Committee for its national voter file.
These findings, discussed in an interview with VICE News, discredit recent assertions by President Donald Trump that there is an epidemic of double-voting and voter fraud.
Since it became clear that Hillary Clinton won the popular vote in November’s presidential election, Trump has repeatedly claimed, without evidence, that he came up 2.9 million ballots short because 3 million to 5 million people voted illegally. (There are 200 million registered voters in the U.S.) Five days into his presidency, Trump called for a “major investigation” into that alleged fraud but then later scaled back his effort to a “commission” led by Vice President Mike Pence to study the problem.
“When you see people that are registered in two states that voted in two states, when you see other things, when you see illegals,” Trump said when pressed to provide evidence to back up his claims in an interview this week with Fox News’ Bill O’Reilly. “A lot of people have come out and said that I am correct,” he added, without specifying who.
A quarter of American voters now believe millions of votes were cast illegally in the 2016 election, according to a Politico/Morning Consult survey released last week (they are divided on whether those votes helped Trump or Clinton).
Each state maintains its own voter database, and it’s true that all of them include a number of dead, moved, and ineligible voters. A 2012 Pew Report found that about 2.8 million people have active registrations in more than one state and more than 1.8 million deceased people had active voter registrations. And states can’t easily check or investigate whether people are mailing in a ballot in one state and voting in another because there is no national voter file kept by the federal government to crosscheck people’s names.
TargetSmart’s researchers found that states tagged about 45,000 voters who could have voted twice in the 2014 election, about 0.0005 percent of the more than 83 million ballots cast that year. Moreover, a near equal number of potential double votes came from registered Republicans as from registered Democrats.
TargetSmart was able to conduct its independent analysis because its engineers spent years painstakingly constructing a national voter file of voting records and party identification, among other information, for the Democratic Party and other Democratic campaigns. The firm specializes in political data technology that it mainly offers to Democratic and corporate clients for communications. (The GOP has also spent years working with its own firm to build out a sophisticated voter file.)
TargetSmart looked at voter registrations in the 2014 midterm elections, the most recent election for which there is national data, and like Pew found that just under 3 million people had multiple registrations. TargetSmart corrected for people with the same name and the same birthday using commercial and government data, such as change of address forms.
The group then culled through its voting data to see if states had “tagged” any of those 2.9 million people as having voted multiple times. But “tagging” is disconnected from the process of actually counting ballots, and there can be errors in tagging both on the local level and on the data management side. Poll workers sometimes mark the wrong box and outdated software sometimes tags both registrations even if just one ballot is submitted. “There are some people you can’t weed out since states don’t have the tools to track those things,” said TargetSmart CEO Tom Bonier.
In order to know for sure whether a person cast two ballots, the state or the federal government would have to go through the actual ballots one by one to find the suspected voting rosters and compare the signatures.
These potential tagging errors are why TargetSmart and other voting experts emphasized that even the 45,000 number, already a tiny fraction of the total number of votes, was likely extremely inflated. More investigation is required to determine whether these voters in fact cast two ballots.
Several experts on elections and voting fraud suggested that while many double-votes probably go unprosecuted, the total number of people committing such fraud was likely closer to between 1,000 and 10,000 out of the 83,262,122 ballots cast in 2014. The number of double-ballots cast each election “is almost certainly higher than the few instances out there based on what we know,” Bonier said, but the total is still “small” — and certainly not on the scale Trump has claimed.
Part of the reason we don’t have a definitive count is because there are limits to what TargetSmart can achieve with public data. Their national voter file is created by splicing together 50 different state voter files along with a combination of commercial and government data to create individual profiles for each voter. Voting experts and TargetSmart itself acknowledge that the file is not complete and is partly limited by poor data collection on the state level.
“The political parties are trying to create a single record in these vendors’ files, but each one is imperfect,” explained David Becker, the executive director and co-founder of the nonpartisan group Center for Election Innovation & Research and author of the 2012 Pew study on the millions of defunct voting registrations.
The voting experts also emphasized that the number of double votes was so small and so evenly divided among Democrats and Republicans that it was almost impossible for these votes to have changed the election outcome in 2014. They were confident double-votes couldn’t have significantly altered the 2016 election, either.
Trump has claimed that “none” of any potential illegal votes were cast for him, painting voter fraud as a Democratic phenomenon. But a few people have been criminally convicted of casting multiple votes for Trump and other members of the GOP. One Trump supporter in Iowa admitted to voting twice in the 2016 election and was charged with first-degree election misconduct, for example. In Gov. Scott Walker’s 2012 recall election in Wisconsin, one devout Republican was found guilty of casting five ballots.
There may not be a crime spree of voter fraud, but America’s voting system remains remarkably disorganized, and that presents a challenge for evaluating the extent of the problem. “The saying ‘If you can’t measure it, you can’t manage it’ applies here, and many jurisdictions are not measuring it,” said MIT professor Charles Stewart III, an expert in election technology and administration.
If states do measure it, they often only have the ability to investigate cases within their own borders because there is not yet a centralized national database of voters. Such intra-state inspections are therefore both time-consuming and inherently incomplete. When states like Texas and Kansas have investigated potential double-voters, many man-hours of research ended in just a handful of convictions. “Most states, in my experience, don’t go through one by one on double-tagging because there’s a question about the use of staff time to do this,” Stewart said.
There are two significant efforts underway aimed at finally solving the problems created by the balkanized and rusty state voting systems. Crosscheck, which is spearheaded by Kansas’ controversial Secretary of State Kris Kobach, has enlisted 30 mostly conservative states to check dual registrations. Crosscheck has become politically divisive, however, as Democrats and some media reports have accused Crosscheck of using the program to remove blacks and Latinos from the voting rolls for political gain.
The other effort is an interstate cooperative pragmatically named the Electronic Registration Information Center. Originally launched by Pew in 2012, ERIC is now independent and includes 20 red and blue states plus the District of Columbia, with hopes to enroll all 50 states.
Currently, ERIC uses IBM data-matching software to cull through regularly updated state voting rolls and flag instances of multiple registrations, dead people, and other errors. Tammy Patrick, who worked for more than a decade as a federal compliance officer in Arizona elections and helped write ERIC’s bylaws, said she believes ERIC is “the only way to truly be assured that folks voting in two places would be identified.”
Becker helped launch ERIC and said the ultimate goal is that “every eligible voter that wants to be registered has one registration file that follows them through life.” People would no longer have multiple voter registrations in different states, and the less-cluttered system would, in Becker’s reckoning, raise public confidence in American elections.
The inefficiencies of the current system allow Trump to obfuscate when presented with evidence that there is no large-scale voter fraud problem in the United States. In addition to double-votes, Trump has focused on voters impersonating dead people who have yet to be removed from voting rolls. But an investigation by Professor Justin Levitt, a widely cited expert in voting law, found only 31 potential cases of voter impersonation in more than a billion ballots cast between 2000 and 2014.
A database put together by News21, the student reporting project, documented only 2,068 alleged cases of voter fraud between 2000 and 2012 out of hundreds of millions of votes cast. The violations, many of which did not result in convictions, included people voting twice, people voting who were too young or not citizens, and people impersonating other voters.
Many such studies have been conducted in the past several years as some Republicans have used the specter of fraud to justify voter ID laws that disproportionately disenfranchise black and Latino voters, who tend to vote Democratic. Like TargetSmart’s analysis, none found definitive evidence that there is pervasive voter fraud.