You select the topic.

Significance

President Donald Trump claimed that the 2020 US presidential election was stolen; millions of Americans apparently believed him. We assess the most prominent statistical claims offered by Trump and his allies as evidence of election fraud, including claims about Dominion voting machines switching votes from Trump to Biden, suspiciously high turnout in Democratic strongholds, and the supposedly inexplicable failure of Biden to win “bellwether counties.” We use a combination of statistical reasoning and original data analysis to assess these claims. We hope our analysis contributes to public discussion about the integrity of the 2020 election and broader challenges of election security and election administration.

Abstract

After the 2020 US presidential election Donald Trump refused to concede, alleging widespread and unparalleled voter fraud. Trump’s supporters deployed several statistical arguments in an attempt to cast doubt on the result. Reviewing the most prominent of these statistical claims, we conclude that none of them is even remotely convincing. The common logic behind these claims is that, if the election were fairly conducted, some feature of the observed 2020 election result would be unlikely or impossible. In each case, we find that the purportedly anomalous fact is either not a fact or not anomalous.

Why we can be confident that the election wasn’t stolen

The best starting point is to note that there has been no — zero, nada, none — demonstrated, credible example of even a small-scale systematic effort to illegally cast votes. There have been a few dozen isolated arrests, generally of people illegally casting ballots for themselves or family members. In fact, the Associated Press contacted elections administrators in each swing state more than a year after the election, learning that, at most, there were a few hundred questionable ballots cast. In total. Across all of the states. Out of millions cast.

There are few better examples of the proper use of Occam’s razor than to therefore dismiss any idea that rampant fraud occurred. The idea that some systemic, multistate effort to rig the election occurred without detection nearly two years later — in an environment where there’s millions to be made exposing one — is simply noncredible in the face of the alternative: There was no such effort.

Of course, there is no shortage of claims about alleged fraud floating out there. These fall into one of three categories: claims that depend on vague statistical analyses, claims that depend on unseen evidence and claims that have already been debunked or explained.

Before presenting examples of each variety, it’s worth pointing to one of the most robust assessments of fraud claims. In July, a group of Republican officials released a lengthy report documenting and debunking each of the lawsuits filed by Trump and his allies in the wake of the election. It covers a lot of ground. The odds are good that if you’ve heard some claim about fraud or “rigging” (see below), it’s addressed in that document.

Now, instead of debunking each of the common allegations about fraud, I’ll simply list them and link to places where you can read more detailed analyses of why each is inaccurate.

Claims that depend on vague statistical analyses

• There was no secret “key” discovered by Douglas Frank that proves voting machines were used to guide vote totals.
  
• There was nothing odd or unexplained about how votes were counted in Wisconsin or other states on election night. These analyses often depend on willfully ignoring the day-of-vs.-mail-in partisan patterns mentioned above.
  
• The odds were not “1 in 1 quadrillion” that Biden could win as more votes were counted. (This one also ignored the partisan pattern.)
  
• There was no sketchy “drop and roll” process that led to Biden gaining more votes.
  
• 
  

Claims that depend on unseen evidence

• There is no evidence that foreign actors somehow changed votes over the internet, as MyPillow CEO Mike Lindell has repeatedly claimed.
  
• There is no evidence that nonprofits collected illegal ballots that were then distributed to drop boxes by paid staffers, as alleged in the film “2000 Mules.” The purported evidence that was presented in that film is either false, contrived or misleading.
  
• Various audits of electronic voting machines have found no evidence of improprieties. In fact, swing-state counties in which Dominion Voting Systems machines were used mostly voted for Trump.
  
• 
  

Claims that have already been debunked or explained

• Inaccurate vote totals in Antrim County, Mich., were a function of improperly configured voting machines.
  
• There were no rampant irregularities in Maricopa County, Ariz., just a pattern of observers not understanding voting systems and tools.
  
• Vote totals in states such as Pennsylvania or Wisconsin were not dependent on more people voting than were registered.
  
• There was no uncaught double-counting of mail-in ballots in Georgia.
  

One common response to delineations like this is that of course the media/the government/the FBI are going to claim that their analysis showed no fraud. After all, you can’t have a healthy conspiracy without a gaggle of conspirators.

So we back up a step. To assume that I’m in on the con along with all of the other sources linked above is to postulate a system involving thousands or tens of thousands of people, all of whom have agreed to stay silent simply to protect Biden. Or, at least, that hundreds of people in the government have all kept quiet about agreeing to mislead the public, despite the obvious financial and moral rewards for revealing a part in such a scheme.

Occam’s razor. Who has more reason to make dishonest claims about the election, the guy trying to get people to watch his movie claiming fraud or the guy who works for a privately owned newspaper? Who is more credible on the likelihood of fraud, independent researchers or a former president eager to maintain his grip on his base?