The suit uses statistical arguments that experts called ‘comical.’
Mr. Paxton’s filing repeatedly cites an analysis by an economist in California that statisticians have said is nonsensical. Mr. Biden’s chances of winning the four battleground states in question, the analysis says, were “less than one in a quadrillion.”
The economist, Charles J. Cicchetti, who donated to Mr. Trump’s campaign in 2016, arrived at the minuscule probability by purporting to use the results of the 2016 election as a backstop. His flawed reasoning was this: If Mr. Biden had received the same number of votes as Hillary Clinton did in 2016, he wrote, a victory would have been all but impossible.
But Mr. Biden, of course, did not receive the same number of votes as Mrs. Clinton; he received over 15 million more. Nor would any candidate be expected to receive the same number of votes as a previous candidate.
That one-in-a-quadrillion figure has echoed across social media and was promoted by the White House press secretary. But an array of experts have said that the figure and Mr. Cicchetti’s analysis are easily refutable.
Stephen Ansolabehere, a professor of government at Harvard University who runs its election data archive, called this analysis “comical.”
The analysis omitted a number of obvious, relevant facts, he said: “the context of the elections are different, that a Covid pandemic is going on, that people reach different conclusions about the administration, that Biden and Clinton are different candidates.”
By the same logic and formula, if Mr. Trump had received an equal number of votes in 2020 as he did in 2016, there is also a one in a quadrillion chance that Mr. Trump in 2020 would outperform his totals in 2016, said Stephen C. Preston, a professor of mathematics at Brooklyn College. “But that doesn’t prove Trump cheated, it just shows that the numbers are different,” he said. “It’s like finding a low probability that 2 equals 3.”
Mr. Cicchetti also wrote that votes counted earlier in the process and votes counted later favored different candidates, and that there was “a one in many more quadrillions chance” that votes counted in the two time periods were coming from the same groups of voters.
But that is exactly what was expected to happen: Democrats tended to prefer voting by mail, and those ballots were counted later in the four battleground states, while Republicans tended to prefer voting in person on Election Day, and those ballots were counted earlier.
“The order and tempo of vote counting was unlike previous elections,” said Amel Ahmed, a professor of political science at the University of Massachusetts Amherst.
What Mr. Cicchetti wrote was not especially revelatory, experts agreed.
“The model is silly,” said Philip Stark, a professor of statistics at the University of California at Berkeley. “This is not science or statistics. It’s not even a good cartoon of elections.”
