Does Voter Fraud Exist? A Scientific Study

You may recall that in 2016, then-candidate Donald Trump had a lot to say about voter fraud, arguing that illegal voting was rampant, that it was exclusively advantaging the Democratic party, and that it numbered in the millions. Since becoming president, Trump has convened a commission on election integrity to investigate issues related to voter fraud, but the commander-in-chief has never cited any evidence that such fraud exists.

Most media reports have contradicted Trump’s claims, stating that there is no evidence of widespread voter fraud, implying that Trump’s efforts to chase down this alleged problem are either misguided, disingenuous, or perhaps even a cover for something more sinister, like voter suppression, an election strategy that has been gaining currency amongst conservative candidates and disproportionately affects people from low-income households and minority groups.

It’s certainly not very scientific to suggest that voter fraud exists based on, apparently, no evidence. On the other hand, it’s not very scientific to state that there unequivocally is no voter fraud without evidence, either. Has anyone done an actual study to actually sort the question out, one way or another? Actually, someone has, and it was released just last week.

If you don’t read the full 70-page study, no worries, I’ll give you the highlights. You may also enjoy this 15-minute story about the study run by This American Life last week, in which study author Sharad Goel of Stanford University describes the work he and his four co-authors did to objectively answer the question of just how large a problem voter fraud actually is.

It’s important to note that this study was designed to look at just one type of voter fraud: double-voting, wherein a registered voter goes to the polls twice and casts two ballots. Other types of voter fraud that are alleged by Trump and others to occur include people voting using the names of people who are dead but still on voting registries, or illegal immigrants voting (again, presumably pretending to be someone else who is on the voter list).

Different studies would have to be done to conclusively discover whether these kinds of fraud exist in any significant numbers, but there’s no reason to think that one kind of voter fraud should be highly more or less prevalent than another, meaning Goel’s study provides strong evidence as to how much voter fraud, of all kinds, may actually exist. Note also that this study hasn’t been through peer-review yet, and wasn’t scientifically vetted outside the five highly competent mathematicians who conducted it. But that’s coming.

So how to check for double-voting? It’s actually pretty simple in concept, though time-consuming in practice. Goel’s team had to look at physical voting records and actually find out how often a person with the same name and birthdate were listed as having voted more than once. And the results were surprising. Goel found about three million matches, a number that seems to vindicate Trump’s claims. The number is so similar, you have to wonder whether Trump actually heard something about this or a similar study that counted identical names on voter lists.

However, this is a mathematical study, and you never stop with the first number. Now you have to consider sources of error and confounding factors, you have to consider whether some of these people with the same name and birthdate might actually be different people, and once you’ve accounted for all of that, then you have the real number. Goel looked at error rates in reporting and found that a certain percentage of the time the counts on who had voted and not voted were incorrect when compared against the actual lists where voters signed next to their name when voting. The error rate is small, but three million people is also a relatively small percentage of the total number of voters, so a small error rate adds up.

When accounting for this and other factors, the possible number of illegal voters drops down to less than a million. Then you have to consider how many people have the same name and birthdate, e.g., Jane Doe, January 1, 1971. You probably figure, not many, right? But then there’s the really interesting mathematical result called “the birthday problem.” Though counter-intuitive, it turns out you only need 23 people in a room before you get even odds that a pair of them have the same birthday (month and day), because there are so many different people who could be paired up.

Extend this to a larger group of people, more than 60 million voters, and bring the birthyear into it, it turns out there are likely to be many groups of people with the same name, and with each group, the several hundred Jane Does, the several hundred John Smiths, the several hundred Jose Villanuevas, a lot of these groups will, in fact, include pairs of people with the same full birthdates. Enough to account for most of the remaining several hundred thousand name matches. The remainder are accounted for by confounding factors to do with non-random name and birthday relationships (e.g., most women named June being born in June) and the fact that voting records default to January 1 when a voter’s birthday is not known, creating a lot of extra matches among same names.

The end result of the study? Once every possible factor is accounted for, the rate of double-voting ends up being zero. This is a statistical zero, mind you. The real number may not be literally zero. It might be five, or 30, or even 100. But it’s not in the thousands, let alone the millions. The data simply doesn’t bear that claim out.

My biggest surprise in skimming this study? That there is even the small possibility that Trump’s claim of millions committing voter fraud might actually be based on an actual measured number, even a misleading and improperly calculated one. I assumed he 100% made it up out of thin air.

Photo credit: Thinkstock

62 comments

Dan B
Dan Blossfeld3 months ago

Joan E.,
Most people will acknowledge that voter fraud is not insignificant, but is only significant in certain less reputable areas (like Florida, Illinois, and Kentucky). The Supreme has even acknowledged the existence of wide-spread voter fraud in its recent decisions. Perhaps the only people denying its existence are those benefitting from it.

SEND
ERIKA S
ERIKA S3 months ago

noted

SEND
Joan E
Joan E3 months ago

David F, I think the self-serving, money-grubbing crooks in both parties, not just one, need to be locked up.

SEND
Brian F
Brian F3 months ago

Mike K Jjll Stein sat with a group of people, and had a meeting with Putin. At least Jill Stein didn't make 140 million dollars by selling 20% of our uranium to Russia, like your Wall Street crook Hillary did. At least Jill Stein didn't take $500,000 for a speech to a Russian Bank like your crook Bill Clinton did. At least Jill Stein didn't support natural gas fracking, Monsanto, the TPP 45 times, the Patriot Act, the Iraq war, refuse to support Medicare for All, and take 21 million for speeches to Wall Street, and refuse to release the transcripts like your crook Hillary did. At least Jill Stein denounced the horrible Dakota Access pipeline and went to Standing Rock, to support the Indian Tribes who were getting brutalized by over militarized heavily armed thug police, unlike your Crook Hillary who refused to denounce the DAPL pipeline, because she profited off it. It's time to take your meds.

SEND
David F
David F3 months ago

Joan, you are correct, Lock Her Up.

SEND
Joan E
Joan E3 months ago

And let's not forget the crime of colluding with Russia to fix an election. But oh, we're supposed to not notice that, we're supposed to worry about two people named Juan Martinez who voted in two different states, so that proves they are the same person who voted twice.

SEND
Joan E
Joan E3 months ago

Individual voter fraud is insignificant, but election fraud -- taking people off the rolls, purposely confusing voters about where and when they can vote, making it harder for people to register and vote -- is a widespread crime conducted by the political parties who cheat to win but deserve to be locked up .

SEND
Ellie M
Ellie M3 months ago

ty

SEND
Winn A
Winn A3 months ago

Petition Signed

SEND
Aaron F
Past Member 3 months ago

@Dan Blossfield. Point taken...and it was an excellent pick-up...

SEND