Voter Fraud

The fear of voter fraud has been a regular topic in our recent national narrative. It dominated political discussion following the last presidential election. While one side claims that it swung the election, the national security and intelligence apparatus, along with local election officials and multiple external investigations, have concluded not only that voter fraud was not a factor in the election, but this was the most secure election in our recent history. Despite this, discussions of voting rights and voter turnout inevitably include the topic of voter fraud.

The argument seems to be that the easier it is to vote, the easier it will be for fraud to occur. The problem with that argument is that all the evidence suggests this is not at all the case. Luckily, our election system allows us to test these claims with real data.  Local election law builds in all kinds of redundancies to make sure that fraud doesn’t happen and to ensure that when it does happen, it gets caught. And when it gets caught, it gets reported and prosecuted. Unfortunately, because election administration is so decentralized, it can appear confusing and opaque from the outside, a combination that is perfect for breeding distrust and suspicion.

The Heritage Foundation, an organization self-described as committed to promoting conservative public policy, maintains a voter fraud database where they document cases of proven voter fraud.

They stipulate that the database does not represent a comprehensive listing of all voter fraud since it cannot document fraud that is not detected nor fraud that is not proven. Although, if voter fraud is suspected but can’t be proven, perhaps it was not fraud after all. In fact, given the many redundancies that help prevent and uncover voter fraud, we think that proven fraud cases are the best, most accurate measure of actual fraud. The proven cases of fraud in the Heritage Foundation database are relatively few.

Earlier, we reported on our study looking at the association between election administration laws and voter turnout. As part of that study, we also investigated whether there was any relationship between election law, voter turnout, and voter fraud in even election years from 2008 through 2018. To do this, we used the Heritage Foundation data set, in addition to data on state level demographics, voter turnout by state, state level election spending, and state level election law.

Based on the Heritage Foundation data, the highest rate of fraud between 2008 and 2018 was in Minnesota in 2008. That year, Minnesota had 115 incidents of proven voter fraud. 

We decided to look at voter fraud two ways in order to get a handle on what 115 cases really meant.

First we looked at voter fraud in relation to the population of the state.

In 2008, Minnesota had a population of around 5.3 million — 5,241,914 according to the estimates from the American Community Survey. 115 incidents of proven voter fraud out of a population that size comes to a fraud rate of 0.000022. In other words, there were .022 incidents of voter fraud for every 1,000 people in the population. That assumes the population of Minnesota are the only possible pool of fraudulent voters. If we were to add people from neighboring states or other countries to our denominator, the fraud rate would be even lower.

Second, we looked at voter fraud in relation to the number of ballots cast in the election.

In Minnesota in 2008, there were 2,910,369 ballots cast for the highest office in the election. With 115 proven cases of voter fraud, that means the 0.004% of all ballots cast in Minnesota were fraudulent. Voter fraud, even according to the Heritage Foundation data, is exceedingly rare, not exceeding 0.004% of total ballots cast in any state in the US between 2008 and 2018.

With further analysis, we found that there was no relationship between voter fraud and the overall ease of voting in any state across the years that we studied. We also found that there was no relationship between voter fraud and any individual election laws. Finally, we found that there was no relationship between voter fraud and the voter turnout rate.

What does it mean that there was no relationship? Here we are talking about statistical correlations controlling for a host of state level factors. The fact that there is no relationship between voter fraud and voter turnout means that we can’t make any accurate predictions about the value of one based on the value of the other because there is no discernable connection in their patterns. If they were connected, we would expect to see more fraud in places where it is easier to vote. Instead, we see that there is very little proven fraud anywhere and no pattern of fraud that can be connected to how easy it is to vote. The evidence makes clear there is no systematic voter fraud: voter fraud is not related to laws that increase access to voting nor to laws that restrict access to voting.