By Associate Professor Justin Levitt
Every year, late October brings pumpkin pie and horror movies and reports of deceased voters. The reports are, inevitably, spooky, with a hint of the comic macabre. And they are often followed by proposals to slash the voter rolls, with the equivalent of an electoral chainsaw.
But now that the cobwebs have been brushed away in the clearer light of a November morning, most of the specters have vanished. Turns out that the undead hordes are too busy terrorizing your Tivos to be stealing your elections.
Most allegations of dead voters rely on attempts to match large computerized databases of dead people to large computerized voter rolls. And they fall prey to three basic fallacies.
First, bad data. Sometimes people listed as dead are not really dead. Investigative reporters at Scripps-Howard have discovered that the Social Security's Death Master Index, one of the most-used registries in the country, falsely announces the deaths of nearly 1,200 living Americans per month. Think of these legitimate voters as buried alive.
And sometimes, people listed as voting did not really vote. Election records are imperfect. In the press of a busy day at the polls, people sometimes sign the wrong line of a pollbook. And in the press of a busy canvass after the election, officials sometimes hit errant keystrokes, recording a vote that a deceased individual did not actually cast. Think of these legitimate souls resting peacefully, without braving the lines at the polls.
Second, bad matching. These reports often rely on comparing names and birthdates. But in any large pool of records, it's surprisingly common for two different people to share the same date of birth. Statistics prove that if you've got 460 people named "Michael Myers" in your population, it's virtually guaranteed that two will share the same date of birth. Which means that in millions of computer records, Michael Myers, dearly departed, may not be the Michael Myers casting a ballot. Florida governor Rick Scott was actually purged from the rolls in 2006 by such an error.
Third, bad timing. Though the dead rarely vote, voters do sometimes pass away. There have been reports of voters casting early votes or absentee ballots, and then shuffling off this mortal coil.
Take away the mistakes and misinterpretations, and you take away most of the howling behind the dead voter allegations. When real researchers spend real time following up on the claims, they leave at most a handful of oddities unanswered, almost exclusively in the absentee system. An enterprising reporter dug deep into the St. Louis -- St. Louis! -- rolls in 2007, and every single suggestion of gravesite voting evaporated. Sometimes the follow-up dispels the ghosts entirely.
There is more legitimacy to the notion that registered voters remain on the rolls after they die -- not that they cast ballots, but that the records linger. The volume of deceased registrants is also often overblown, for all the reasons above, but the leftover records do end up amounting to more than a handful. When election officials do their job well, these records are removed from the rolls, slowly and deliberately and in the off-season, with safeguards to ensure that no legitimate voter is caught up in the sweep.
That sort of careful list maintenance is a far cry from the hurried and harried last-minute purges urged by the purveyors of dead voter tales. Overbroad actions in the waning days of an cycle can only undermine the integrity of a election by jeopardizing the rights of real, live, eligible voters. And that is what's really scary about the dead voter stories of late fall.