A report for the state of North Carolina by Interstate Crosscheck - a consortium out of Kansas that crosschecks 101 million voter records in 28 states - reveals thousands of instances of possible double voting by residents of North Carolina who also may have voted in another state.

WRAL:

Strach said North Carolina's check found 765 registered North Carolina voters who appear to match registered voters in other states on their first names, last names, dates of birth and the final four digits of their Social Security numbers. Those voters appear to have voted in North Carolina in 2012 and also voted in another state in 2012.

"Now we have to look individually at each one," Strach said. "Could there have been data error?"

The crosscheck also found 35,570 voters in North Carolina who voted in 2012 whose first names, last names and dates of birth match those of voters who voted in other states in 2012, but whose Social Security numbers were not matched.

"A lot of states don't provide last four SSN, or they don't have that information," Strach explained.

Additionally, the analysis found 155,692 registered North Carolina voters whose first and last names, dates of birth and final four Social Security number digits match voters registered in other states but who most recently registered or voted elsewhere.

That last group, Strach said, was most likely voters who moved out of state without notifying their local boards of elections. "Those may be voters we need to remove because they've left North Carolina."

Strach also said a "10-year death audit" found 13,416 deceased voters who had not been removed from voter rolls as of October 2013. Eighty-one of those individuals, she said, died before an election in which they are recorded as having voted.

Strach cautioned that about 30 of those 81 voters appear to have legally cast their votes early via absentee ballot and then died before Election Day.

However, she said, "There are between 40 and 50 [voters] who had died at a time that that's not possible."

The four largest states - CA, FL, NY, and TX - are not part of the consortium so if there's a problem, it is probably even more widespread than this report shows.

But is there a problem?

It is tempting to jump to the conclusion that voter fraud in North Carolina is massive. But experts point out there's a statistical fluke involved, as well as the practical matter of the same voter casting a ballot in North Carolina and driving hundreds or even thousands of miles to cast a ballot in another state.

There is a problem with the way the program counts double voters:

The program, for instance, appears to count every instance in which someone has moved out of a state, registered to vote in their new state, but has not yet been removed from the old voter rolls, a process that can take several election cycles to happen automatically.

And while the program asks member states to submit 13 items of data for each voter, including the last four digits of his/her social security number and middle name, Kansas state department officials acknowledged in an email that all that’s required for the crosscheck program to generate a “possible duplicate entry,” is for the last name, first name, and date of birth to match.

When told that the crosscheck reports use only last and first names and a birthday, Philadelphia elections commissioner Stephanie Singer audibly gasped.

“There are going to be a lot of David Lees on that list,” she said.

A “hit,” meanwhile, means even less when it comes to detecting the kind of voter fraud the program is supposed to target: double voting, in which a voter registered in two states votes simultaneously in both.

Evidence of such fraud is scant. Only a handful of cases have been documented anywhere, ever. The the logistics alone of a voter’s voting twice, in two different states, can stretch the imagination.

Why should we be skeptical of the notion of fraud rather than clerical or computer error, or just plain statistical anomalies? In 2012, Interstate Crosscheck scanned 84 million records and came up with 5 million possible double voter registrations. After states looked into the specifics, less than two dozen people were charged with fraud.

Further explanation on the issue comes from Justin Levitt of Election Law Blog:

The National Review Online has today’s latest entry into the litany of voter fraud reports in which the story doesn’t actually support the headline.

The headline says: “N.C. State Board Finds More than 35K Incidents of ‘Double Voting’ in 2012.”

The story reports that an attempt to match voter databases to each other revealed “more than 35K” entries for North Carolina voters who apparently share the same first name, last name, and date of birth with voters in other states.

In order to tell whether there’s any difference between the headline and the story, you need to know:

• Whether there were any data entry errors by NC pollworkers, in noting the individuals who voted,
• Whether there were any data entry errors by other states’ pollworkers, in noting the individuals who voted, and
• Statistics.

Turns out that if you actually run the numbers, there are a lot of people who share the same name and birthdate.  Or, if you prefer anecdote to math, just ask Florida Governor Richard Scott.

I would be very interested indeed in how many of the alleged double voters — the people that the headline would have just libeled had they named names — are the results of mistakes or mistaken assumptions.  And I mean that.  I hope that there’s real follow-up.

If there is, I’m going to bet on the vast majority of allegedly fraudulent votes evaporating upon closer scrutiny.  Perhaps one day we’ll learn not to breathlessly credit the first results of a data-matching exercise.  Remember the reports of 900 dead voters in South Carolina?  How’d that turn out?

Recall that those 35,000 records of double voting have different Social Security numbers, suggesting a statistical fluke rather than anything truly sinister.

Still, it is the responsibility of state officials to dive into this issue. At the very least, they can purge the rolls of dead people and duplicate entries. This in itself will improve the integrity of the process.