North Carolina Republicans last year passed a sweeping and restrictive voting law, which is currently
being challenged by the U.S. Justice Department. The lawās voter ID provision would likely have done nothing to stop the double voting being alleged here, but solid evidence of illegal voting could still bolster the stateās case that the measure is justified. It could also make it easier for the state to remove from the rolls voters who are thought to be registered in two statesāraising concerns that legitimate voters could wrongly be purged.
But Republicans and conservative media, predictably, arenāt waiting for the results of the probe. Instead, theyāre already shouting voter fraud. āN.C. Board of Elections audit finds up to 35,750 instances of ādouble votingā (voter fraud) in the 2012 election,ā
tweeted Republican National Committee spokesman Sean Spicer Wednesday. A
headline at National Review made the same claim. In a
statement, state Sen. Thom Tillis, the frontrunner for the GOP Senate nomination, pointed to āalarming evidence of voter error [and] fraud.ā
The notion that the board found over 35,000 cases of voter fraudāor even one caseāis flatly false. With the investigation not yet even underway, the board, headed by Kim Strach, hasnāt come close to concluding that any specific case involved double voting.* And there are very good reasons why itās held off.
First, it helps to understand statistics. The political scientist Michael McDonald and election law scholar Justin Levitt have shown in a
detailed statistical study that the number of people who share a name and birthdate is much higher than it might at first appear. (Just for fun, take the RNCās Spicer. Though his name is less common than many,
online records show 20 different Sean Spicers who were born on September 23rd, his birthday.) That statistical reality, McDonald and Levitt conclude, has big implications for how to treat potential cases of illegal voting.
āI would be very interested indeed in how many of the 35K alleged double voters are the results of mistakes or mistaken assumptions,ā Levitt wrote Wednesday in an email to a group of election lawyers. āIām going to bet on the vast majority evaporating upon closer scrutiny.ā
But that still leaves those 765 casesānot as eye-popping a number as 35,000, but still significantāin which the last four digits of a voterās Social Security number also matched that of someone who voted in another state. Statistically, the chances of a false positive are much, much smaller under this scenario.
Even here, though, there are plenty of explanations beyond deliberate fraud. Election experts point to the high frequency of data errors by poll-workers, a possibility that doubles, of course, when matching voters across two states.
Consider the recent experience of North Carolinaās southern neighbor. Last year, South Carolinaās DMV used Social Security matches to help find more than 900 people listed as dead who had voted in recent years, setting off a spate of hand-wringing about fraud. Attorney General Alan Wilson, a Republican,
used the findings to argue for the stateās strict voter ID lawāwhich was later softened after the Justice Department objected. But state law enforcement
ultimately found not a single person who deliberately cast a ballot in the name of a dead person.
Nearly half the cases were the result of clerical errors by poll-workers. Others were attributed to DMV officials finding that Social Security numbers matched but not making sure that names did, among several problems. (About
45,370 people have been assigned by the Social Security Administration to each four-digit combination of numbers.)
