17 September, 2016
Statistical proof of over 750 cases of voter
fraud in the 2012 NC General Election
In 2010 Prof. Lori Minnite wrote that her research indicated that incidents of deliberate voter fraud in the United States are “quite rare,” and her work has was widely cited in support of that conclusion. However, new evidence proves that conclusion was incorrect.
Using Interstate Crosscheck data, the NC State Board of Elections identified 35,750 voters with the same name and Date Of Birth (DOB) as voters in other participating States, in the 2012 general election.
Most were presumably innocent coincidences: people with the same name and DOB as someone else in a different State. But over 750 cases were clearly voter fraud. Here’s how we can tell.
By additionally comparing the voters' entire social security numbers, all cases of innocent coincidence could have been eliminated. That was not done, but election officials did attempt to match the last four digits of the voters' social security numbers (“SSN-4” or “Last4SSN”), in most cases.
SSN-4 is a four-digit number, 0001-to-9999. (Numbers ending in “0000” are not assigned.) There are 9999 numbers in that range, so for two people, randomly chosen, there’s only a 1 in 9999 chance that they will both have the same SSN-4. So, matching SSN-4 is sufficient to eliminate 99.99% (i.e., 9998 out of 9999) of coincidental name+DOB matches.
So if all 35,750 matching name+DOB voters had been innocent coincidences, you'd expect to find only 3 or 4 “false positives,” who coincidentally also matched SSN-4.
Of course, the actual number could be either higher or lower, but a statistical simulation of 25 million simulated elections determined with >99.9998% certainty that it could not have been higher than 15 voters. (Contact me for more details about this work, including the computer source code.)
The NC SBOE didn't find 3, 4, or even 15 cases. They found 765 cases in which NC voters’ name, DOB and SSN-4 all matched voters in other States.
We can say with greater than 99.9998% certainty that at most 15 of those 765 apparent dual-State voters were innocent coincidences. I.e., it is 99.9998% certain that in the 2012 North Carolina general election, Interstate Crosscheck found at least 750 cases of actual voter fraud (765 minus 15).
[Aside: I wrote to Prof. Minnite and asked her, “I don't consider 750 cases of actual, identifiable voter fraud in the 2012 NC general election ‘quite rare,’ do you?” Remarkably, she replied, “Accepting your numbers for argument's sake, yes, I do think that two-hundredths of a percent of all votes cast in the 2012 North Carolina election represents a number that is ‘quite rare.’”]
Note, also, that Interstate Crosscheck cannot identify all cases of election fraud:
1. An impersonator won't be detected unless the person who he impersonates actually votes in another State in the same election.
2. Even then, impersonation won't be detected unless both States participate in the Interstate Crosscheck project. (About half of the States participated in 2012, representing about 78% of the nation's voters.)
3. Even then, it can’t be reliably detected unless SSN-4’s are available from both States.
4. Likewise, cases of the same voter registering and voting in two States won't be detected unless both States participate in the Interstate Crosscheck project, and unless the voter uses the same name and his real social security number in both States.
So the real number of fraudulent votes in NC was obviously higher than 750.
Who is the culprit, and who is the victim?
Unfortunately, it is unknown how many of those 750+ cases of the voter fraud cases were instances of a single person fraudulently voting in two States, and how many were people impersonating other voters (presumably after determining that the actual registered voters had moved out-of-State). Both are examples of voter fraud, but with different culprits.
Those two kinds of fraud cannot be distinguished, because in 2012 NC had no voter ID requirement. Unfortunately, if the culprits cannot be conclusively identified, the crimes cannot be prosecuted.
If criminals cannot be identified and prosecuted, how can crime be deterred?
Without voter ID, we cannot identify the criminals who commit voter fraud, because we cannot distinguish between cases of the same person voting in two different States and cases of voter impersonation.
They are both examples of fraud, but in the former case the named voter is the culprit, and in the later case he is a victim of the crime.
4,505,372 North Carolinians voted in the 2012 Presidential election. A few thousand fraud cases is not a large percentage of that number, but it certainly is not “quite rare.” 750 provable fraud cases is about 0.0166% of the total vote.
That's not a large percentage, but it is large enough to matter. Sometimes even a tiny percentage can change the outcome of an election, with momentous consequences.
In the 2000 Presidential election, the outcome depended on the result in Florida, where Bush beat Gore by a margin of just 0.00922% of the vote, which is much smaller than the percentage of provably fraudulent NC votes in 2012. Likewise, Sen. Al Franken’s 312-vote (0.0109%) MN victory margin in 2008 ultimately provided the deciding vote to enact ObamaCare, and that margin was also much smaller than the percentage of provably fraudulent NC votes in 2012.
Objections & answers
Professor Minnite asked me, “I wonder how you account for predictable (if unmeasured) typos and human error that we know exist in real, actual voter files. For example, numbers are transposed, names are misspelled, and mistakes are made in the recording of votes.”
Of those three sorts of human errors, the first two, name misspellings and transposed SSN digits in voter files, would cause the Interstate Crosscheck to overlook some fraud cases, resulting in an underestimate of the scale of the fraud problem.
The third sort of error, mistakes in the recording of votes, has the opposite effect. If the wrong voters are recorded as having voted, those would be cases of accidental voter impersonation. It is still important to identify such errors, for the sake of the integrity of the election process, but such errors are not intentional fraud.
However, here in North Carolina, procedures are in place and poll workers are trained to ensure that mistakes made in the recording of votes are very rare. Before a voter receives his or her ballot, he must confirm both his identity and his address, via a verbal exchange with a poll worker, and he must sign his name in the pollbook, adjacent to his printed name and address.
That system seems to be working well. If it weren't, we would surely know it! We cannot say that mistakes in the recording of votes have never happened, but it is clear that they are uncommon.
In a high-turnout year like 2012, when 68.3% of registered NC voters actually voted, if the wrong voter were recorded as having voted, that would usually (68.3% of the time) result in someone else being unjustly denied the right to vote, because the pollbook would erroneously show that that person had already voted.
A much smaller number of recording errors will result in apparent dual-State voters, because only a small percentage of still-registered NC voters have actually moved out-of-State.
Unjustly denying someone their right to vote is as great an injustice as allowing a case of intentional fraud to go undetected and unprosecuted. But, fortunately, it doesn't very often happen that voters are denied the opportunity to vote a regular ballot because they were recorded as having “already voted.”
Most still-registered NC voters have not moved out of State, so if there were 750 cases in which legitimate voters were accidentally recorded as someone else who had moved out of State, it would necessarily mean that there were thousands of cases in which voters were accidentally recorded as someone else who had not moved.
So if most of the 765 apparent dual-State voters had been due to accidental recording errors, thousands of other voters would have been denied the opportunity to vote a regular ballot because they were recorded as having “already voted.”
In North Carolina, the usual procedure in such a case would be for polling place officials to require the voter to vote a Provisional Ballot. In the 2012 NC general election there were only 165 provisional ballots issued to voters who were reported as having already voted.
For 76 of those 165 cases, comments were recorded about the circumstances. From those comments, it is clear that most of them could not have been recording errors. In fact, of the 76 cases in which comments were recorded, only one of them sounds like a possible recording error: a Mr. James Arthur Lucas II, of Clemmons, NC (“Father James Arthur Lucas may have voted for him per voter”).
If accidental recording errors were the explanation for the 765 cases of apparent dual-State voters, there would necessarily have been thousands of cases in which voters were forced to vote a provisional ballot because of recording errors, rather than just one such case. So it is clear that accidental recording errors cannot explain most of the 765 apparent dual-State voters in NC's 2012 general election.
Plus, fortunately, the pollbook signatures can help correct most such errors. E.g., if James Lucas Sr. were to accidentally vote as his son, James Lucas Jr, the fact that voters must sign their names in the pollbooks would probably enable election officials to identify what happened and correct the mistake. (James Lucas Jr would have to vote a provisional ballot, of course.)
Prof. Justin Levitt’s 2007 Brennan Center Report, which cited some of Prof. Minnite’s work, claimed that the sort of voter fraud which could be detected or deterred by voter ID is “more rare than death by lightning.” (p.3) It has been cited in the press to support the premise that voter fraud in the United States is “virtually nonexistent.”
Fortunately, in a typical year only 2 to 4 North Carolinians are killed by lightning, not 750.
“When my information changes, I
alter my conclusions. What do you do, sir?”
- John Maynard Keynes
Burton Systems Software
Cary, NC 27513
Online copy, with hyperlinks: