If you show up to vote on Nov. 8, what are the chances that someone with the same name and date of birth is doing the same in another state? New research suggests that this happened about 30,000 times nationwide in 2012, or about 0.02 percent of votes cast, but it’s not clear how many were actually double votes by the same person.
This scenario, which apparently happened to me in the last presidential election, is often cited as evidence for the widespread existence of voter fraud, but a new study published last week found that many of the apparent double votes likely didn't occur but were the result of incorrectly marking people as having voted when they didn't.
The findings "dispel some of the more frivolous anecdotes of rampant voter fraud," write the researchers, who are from Harvard University, Stanford University, the University of Pennsylvania and Microsoft Research.
The potential for double voting usually arises when people move from one state to another or buy a second home in a different state and then register to vote there, while not canceling their original registration. Using national voter file data compiled by TargetSmart, a Democratic firm, the researchers found more than 2.8 million pairs of 2012 voting records where two people shared a first name, last name and date of birth. Some of these cases actually represent two different people who have these traits in common.
The researchers developed a statistical model to estimate that 30,000 of the 2.8 million pairs were likely be the same person recorded as voting twice. But because post-election audits often find errors in poll books — for example, people who did not actually vote being marked as if they did — it's possible that very few or no actual double votes were cast.
Double voting has been enough of a concern that many states participate in the Interstate Crosscheck Program, which is led by Kris Kobach, the Kansas Secretary of State. States upload their voter files, which are then checked for possible duplicate records that could be used for purging voters who have moved or for investigating potential voter fraud cases.
But using data from Iowa, researchers conclude that purging voters based on matching individual records based on name and date of birth would remove about 200 valid registration records for every one registration used to cast a double vote in another state. "This suggests the policies necessary to stop the relatively small number of double votes that do occur would put many more legitimate votes in jeopardy," they wrote.
