Lies, Damned Lies, and …
How a truly ridiculous statistical notion played its part in the January 6 uprising
I was struck, during Tuesday's Jan. 6 Committee hearings, that my candidate for "Most Idiotic Statistical Claim Ever Put Forward in Public" award—the coveted MISCEPFiP Award—seems to have played a small but important role in the now-famous "unhinged" meeting in the Oval Office.
You may recall the claim to which I am referring from my earlier posts from back in December 2020: the "proof" that it was "statistically impossible"—"less than one in a quadrillion chance"**—that Biden could have won the election without engaging in massive fraud.
**A quadrillon is an actual number: 1 followed by 15 zeroes (10^15). An event with a probability of occurrence of one in a quadrillion can indeed be considered "statistically impossible" for all intents and purposes—roughly equal to the probability of flipping a fair coin 50 times and getting "heads" on each one. Not exactly impossible, but if it happens to you, you can pretty safely conclude that the coin is biased. [The original post had an incorrect assessment of this probability. H/t Jordan Brown and other commenters for correcting this error]
The claim appears to have had two independent sources. One was Texas A.G. Paxton's motion and supporting brief, submitted to the the Supreme Court as part of his unsuccessful attempt to get the Court to overturn the results of the 2020 election. Paxton, citing an expert report by economist Charles Cicchetti, actually asserted that the "probability of former Vice President Biden winning the popular vote in Georgia, Michigan, Pennsylvania, and Wisconsin independently" is "less than one in a quadrillion," and therefore that "the odds of Biden winning these four States collectively" was "less than one in a quadrillion to the fourth power" (i.e. 1/1,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000).
The second source for this preposterous claim** appears to have been the "Special Report" prepared by then-US Trade Representative Peter Navarro (at then-President Trump's urging), colorfully titled "The Immaculate Deception," purporting to document the "Statistical Irregularities in the Battleground States" that demonstrate the existence of "a coordinated strategy to effectively stack the election deck against the Trump-Pence ticket."
**For reasons I and others have set forth elsewhere, these claims are totally nonsensical and ridiculous. In brief: What the statistical analyses actually show is that (A) the odds of obtaining the reported results in any of the four states if the 2020 voting population had precisely the same Dem. v. Rep. party preferences as the 2016 voting population are infinitesimally small, and (B) the odds of obtaining the reported results in any of the four states if mail-in and in-person voters had precisely the same Dem. v. Rep party preferences are also infinitesimally small. There is, of course, no reason whatsoever to believe that either of those two conditions held true, so these analyses have no relevance whatsoever to any of the actual election results. Indeed, what the analyses actually "prove" is that the 2020 voting population did not have precisely the same Dem. v. Rep. party preferences as the 2016 voting population (duh), and mail-in and in-person voters did not have precisely the same Dem. v. Rep. party preferences (ditto).
But back to the Meeting of the Unhinged, at which, apparently, there was considerable discussion of these various "proofs" that Biden could not possibly have won the election. Our then-President bought them hook, line, and sinker—just what he was looking for! Proof! He reportedly turned to his White House advisors ("Team Normal") and said: "Sidney and Rudy are giving me something—what are you giving me?" I guess he didn't have time to look carefully at the underlying statistical models, what with all the screaming and f-bombing going on in the Oval Office late that night (though I wonder what Texas A.G. Paxton's excuse might have been).
And thus do tiny acorns become great oaks. Rep. Jamie Raskin described what happened next:
Not long after Sidney Powell, General Flynn, and Rudy Giuliani left the White House in the early hours of the morning [of Dec. 19], Donald Trump issued a tweet that would galvanize his followers, unleash a political firestorm, and change the course of our history as a country. Trump's purpose was to mobilize a crowd. And how do you mobilize a crowd in 2020? With millions of followers on Twitter, President Trump knew exactly how to do it. At 1:42 AM on December 19, 2020, shortly after the last participants left the unhinged meeting, Trump sent out the tweet with his explosive invitation.
"Peter Navarro releases 36-page report alleging election fraud 'more than sufficient' to swing victory to Trump https://t.co/D8KrMHnFdK . A great report by Peter. Statistically impossible to have lost the 2020 Election. Big protest in D.C. on January 6th. Be there, will be wild!"
In the history of dangerous nonsense, this surely holds pride of place. This would all be hilarious—it is hilarious, soon to be a major motion picture I'm sure—except for the fact that it was actually part of a determined attempt—the first in our history—to unlawfully overthrow the democratically-elected government of the United States. We should probably be careful about focusing too much—tempting though it is—on the farcical and ridiculous aspects of these events for fear of appearing to diminish their seriousness. The Beer Hall Putsch of 1923 was similarly dismissed by many Germans as little more than a pathetic joke, and while Donald Trump is no Hitler (thank God), we should keep in mind that in 1923, Hitler wasn't yet "Hitler," either; I suspect that twenty years later, a fair number of Germans regretted not having taken the threat he posed in 1923 more seriously.