Improbable Results: More votes than voters and garbage inputs
"Just doing the math" with random numbers people provide you isn't a great approach
133 million registered voters
One of the most persistent election claims of 2020 was that that election had “more votes than voters”. The above claim (which again, is presented as evidence) circulated in spring of 2023 (garnering 700,000 views), but is a variation of a claim that has circulated in various forms since the weeks following the election.
It’s an example of what I call an “improbable results'' claim, one of the subcategories of election claims. Such claims generally marshall statistics not to show a particular way in which the election was “stolen” but to claim something more general – that the stated results are impossible. It’s also a version of a do the math argument, which asks the reader to add, subtract, or divide some simple numbers to “verify” that the math does not add up.
As is the case with most improbable results claims, the claim uses backing of the statistics subtype (for a list of our backing subtypes see here). A key part of the warrant of these claims is that there is no better explanation of the pattern than fraud.
Here we map out the argument. The backing is an unsourced statistic that is presented by the author as evidence there were more votes than voters. The assumption is that the “excess” votes were fake votes for Biden, and these votes exceeded the margin of the win.1
The problem here is that the wrong numbers are given to the reader. In this case, for example, the number of registered voters is not only off, but way off. The Census Bureau indicates that at the time of the election there were about 168 million registered voters. The AP reports that there were about 158 million votes. It’s unclear where the 133 million registered voters statistic came from, since it is unsourced backing, but often these issues come from people using outdated numbers. The entire presentation is based on a wrong premise.
The census number is an undercount of registered voters. It can be compared to an alternate measure from a recent Election Assistance Commission report, which states that there were 209 million active registered voters on the eve of the 2020 election. For various reasons that figure is a good count of registrations, but certainly an overcount of registered voters.
Neither of those numbers is 133 million. In fact, once you think about it, the idea that there are only 133 million registered voters in a country of 330 million is a bit ridiculous.
Pennsylvania’s “1.8 million ballots”
Here’s another one, a longstanding claim that went viral yet again last week.
As far as I know, the claim originates in a legislative hearing in PA after the election. Like claim above, it seems like a simple math problem — that’s nearly twice the amount of ballots they sent out!
The problem is the first number is wrong. There were 1.8 million primary ballots sent out. For the general election, there were 3.1 million ballots sent out, of which 2.6 million were returned, about an 83% return rate.
As I’ve mentioned, Toulmin’s diagrams weren’t really designed to critique misinformation, so I’ve modified my notation a bit here and there. One convention I’ve adopted is the following marks to indicate backing is either unsubstantiated (§) or false (*). (This is something I’ll be playing around with a bit in the coming months as I refine an expanded system of notation).
In any case, when someone tells you to “just do the math” be aware that doing the math is meaningless if you are given the wrong numbers with which to compute.
It’s actually a bit comical that this argument takes what it claims are the total votes and subtracts the Trump ones, in effect claiming that if there are invalid votes, 100% of the invalid votes are for Biden.