New Research Shows “No Evidence of Systemic Voter Fraud in 2020 Election"
November 2, 2021
UChicago research team uses statistical reasoning and original data analysis to negate claims of voter fraud
By Sarah Steimer
Skepticism can be important, especially in a democracy. If you’re unsure about something, if it doesn’t quite add up, it doesn’t hurt to challenge norms or seek answers.
“There's a healthy skepticism that makes sense for people to have about something like an election,” Professor Andrew Eggers, Department of Political Science, says. “It's good for people to ask questions about what's going on. It's a big country, it's a huge election, it's hard to do everything perfectly.”
After the 2020 election, there was skepticism aplenty as many Americans suspected that fraudulent practices got Joe Biden into the White House. But what struck Eggers as particularly concerning was just how flimsy the evidence was that many based their skepticism on — and he explored this evidence further in a new paper published in the Proceedings of the National Academy of Sciences. As the paper’s title indicates, he and his co-authors found “No Evidence for Systematic Voter Fraud.”
What originally led Eggers to investigate the claims was a paper published by economist and political commentator John Lott (also a former visiting professor and fellow at UChicago). In the paper, Lott claimed to find evidence of anti-Donald Trump fraud in the absentee ballot counting procedures in two counties. But Eggers and his co-authors used Lott’s own data to disprove his claims.
After publishing their response in January, the researchers chose to broaden their work and look at other claims of fraud around the 2020 election. “It turned into a paper that tried to take on more of these allegations, but in the same serious way,” Eggers says. “We tried to look at the best version of these arguments that you could make. What does the evidence look like? And we tried to see if there's anything there that really does suggest something was wrong with the election results.”
The team chose to look at the most prominent statistical claims of fraud, which all had the common logic that, if the election were fairly conducted, some feature of the result would be unlikely or impossible. The research explored the purported evidence of the claims using statistical reasoning and original data analysis.
In one example, the researchers used statistical reasoning to check claims that the election was fraudulent because the late-reported counties had different results than early reported counties. The team looked at the numbers and computing used to make this claim and found — simply put — that it didn’t make any sense.
The researchers also used original data analysis to dispel other claims of fraud. In one case, they looked at allegations that the Dominion Voting machines shifted votes from Trump to Biden. “We showed that if you look at places that were using Dominion Voting machines, and you control for the most basic things — like the proportion of vote for Clinton in 2016, for example — then you don't see any difference between places that used Dominion and those using other kinds of voting machines.”
Each of the statistical claims the team investigated failed in one of two ways: In some cases, accurate claims were made about the election results, but the facts being present are fully consistent with a free and fair election. In other cases, a supposedly anomalous fact about the election result turns out to be incorrect.
The team hopes the analysis of these claims will contribute to public discussion about the integrity of the 2020 election, along with broader conversations of election security and administration — but the researchers are also realistic about who, exactly, their study will reach.
“There's no way that we're going to convince Trump, for example,” Eggers says. “He's not going to read our paper and say, ‘Oh, I guess I was wrong about this.’”
Instead, the researchers are focused on offering the paper to those who heard the claims of a fraudulent election and were curious to know more; people who found the arguments compelling but are open to digging a little deeper.
“I think it's unreasonable for citizens to be expected to be able to analyze, to respond to a lot of these things,” Eggers says. “If someone tells you there's a one-in-a-quadrillion chance of the election results having been fair on the basis of some data analysis, that's not something people can really assess. So we thought it was important to actually do that work and give our own assessment of it.
“But it's alarming to me that people can just make these kinds of claims about something like this and get so much attention.”