In the voter fraud debate, be wary of junk science

Allegations of fraud in the 2020 U.S. election show no signs of slowing down. Former President Donald Trump continues to claim the 2020 election was stolen. The impact of those claims has caused significant harm to Americans’ faith in their elections; 71 percent identify illegally cast ballots as a problem in U.S. elections and 30 percent don’t believe Joe Biden was elected President. The summary dismissal of Trump’s charges by many in the establishment and mainstream media over the last 9 months has done nothing to reduce these numbers.

With Trump and his supporters continuing to level these charges, it is vital to engage with and examine the few specific claims of voter fraud they make — both to ensure the security of our electoral system, and to help the public, policymakers, and the judicial system understand the nature and quality of the evidence.

As academic elections researchers, we have done that with an examination of “The Georgia Report” by Matt Braynard and his organization, “Look Ahead America.” This report is likely the most-cited specific fraud claim from Trump supporters, and Trump referenced initial findings from the report in his post-election telephone call with Georgia Secretary of State Brad Raffensperger.

Our detailed findings are posted online here.

After examining the report and its underlying data, we conclude that the “Georgia Report” is fatally flawed in ways that dismantle its veneer of scientific rigor. For the reasons we describe, it cannot confirm any cases of fraud, and its estimates of the amount of illegal voting in the 2020 Georgia election are completely unreliable. Its flaws are so deep that it casts doubt on the similar “methodology” behind most of the Trumpian claims of ineligible voters around the country.

At the heart of the “Georgia Report” is a claim that more than 10,000 people voted illegally in Georgia in 2020 while residing in other states. The authors arrive at this claim by finding individuals who appear in official Georgia records as having voted in 2020, but who also filed “permanent” out-of-state changes of address prior to the election.

But filing for an out-of-state change of address does not establish that an individual is no longer eligible to vote in Georgia. People often file changes of address for temporary relocations and may often call these moves “permanent” based on the USPS’s guidance online. Obvious examples include students studying out of state, people taking care of sick family members out of state, people on temporary work assignments, and people visiting vacation homes. In addition, people can file a change of address well before the election yet not complete their move until after Election Day.

The report simply has no way to deal with this fundamental problem. No amount of snooping on people’s social media accounts can tell you whether a person was or was not a legal resident of Georgia eligible to vote in the 2020 election.

The “Georgia Report” also suffers from basic statistical errors. The authors decide which voter records to sample in a haphazard way, and then fail to account for this haphazardness in their statistics, leading to incorrect extrapolations of the rate of fraud and the associated uncertainty.

After we sent them a draft of our analysis, the “Georgia Report’s” authors released a revised version on July 30. The revision quibbles with three arbitrarily chosen example cases we used for explanatory purposes in our evaluation while doing nothing to address the lack of scientific basis that invalidates their report’s methodology.

The revised report also claims that if the change-of-address data is good enough for state officials to use to remove individuals from the voter rolls, then it is good enough for identifying fraud. This reflects a profound misunderstanding of how election administration works. A voter removed from voter rolls after an election is notified in advance and given the chance to restore their registration to active status, but a researcher making claims about the ballots of people found in the national change-of-address database provides no opportunity to the impugned individuals to correct the record.

Evaluating the American election system and ensuring it is secure is crucial. Americans should regard all claims from all sides with healthy skepticism, and always ask: “What is the evidence?” In the case of “The Georgia Report” and its claim that thousands of people voted illegally in 2020 in Georgia, the evidence simply is not there.

Justin Grimmer is a Senior Fellow at the Hoover Institution and Professor of Political Science at Stanford University. Andrew B. Hall is a Senior Fellow at the Stanford Institute for Economic Policy Research and Professor of Political Science at Stanford University. Daniel M. Thompson is Assistant Professor of Political Science at UCLA.