Proof of Election Fraud or Statistical Hocus-Pocus?

There are many good reasons to believe that Biden’s almost-official election to the presidency was the result of electoral misfeasance, malfeasance, fraud, and judicial bias. But the statistical analysis reported at this link isn’t among them. The authors concocted a statistical model that, according to them,

explains 96% of county-level variance in Trump’s two-party vote share with four demographic variables (non-college white, college-educated white, black and hispanic) and one historical variable (the average of county-level GOP two-party presidential vote share, 2004-2016). All five variables are highly significant. This reinforces the conclusion that the model is generally a very strong predictor of vote shares, and so deviations from it should be considered surprising.

And

regression analysis shows Trump ought to have won AZ, GA, NV, PA, WI.

Are you convinced? I am not, because the authors (perhaps unwittingly) provide evidence that undermines their claim.

There is a table at the end of the article that gives Trump’s predicted share of the two-party vote for every State (except Alaska and Hawaii) and the District of Columbia. I compared the authors’ predictions with the State-level results compiled as of today at Dave Leip’s Atlas of U.S. Elections. The absolute average of the prediction errors in 1.9 percentage points. The absolute errors for the six States listed above are as follows (in percentage points): AZ, 5.0; GA, 3.3; NV, 1.5; PA, 0.6; WI, 1.0; and MI, 1.1. So, as it turns out, the only outcomes (of the six) that the authors’ predictions might point to as fraudulent are the ones in George and Nevada.

Further, the authors don’t bother to highlight Trump’s significant underperformance (relative to their regression results) in many other States: CA, 4.3; DE, 3.5; ID, 2.4; IN, 2.0; KY, 3.0; ME, 2.5; MD, 3.3; NE, 2.1; NH, 2.2; NM, 2.1; OR, 4.9; TX, 3.6; UT, 6.3; VT, 4.3; and WA, 4.0. If their regression results for Georgia and Nevada are indicative of fraud, so are the results in California, Delaware, … , Vermont, and Washington. But I am unaware of any claims that the official outcomes in those States are bogus.

On top of that, Trump did significantly better than the authors predicted in DC (8.5 percentage points) and North Dakota (3.4 percentage points). Is anyone seriously suggesting that there was electoral fraud favoring Trump in DC, or that his campaign had to resort to fraud in deep-Red North Dakota?

The bottom line: The authors made some good predictions and a lot of very bad ones (20 of their 49 predictions exceed the average absolute error). But there’s nothing in the predictions to prove that Biden’s putative victories in Arizona, Georgia, Nevada, Pennsylvania, and Wisconsin (or even Michigan) were obtained fraudulently. There is plenty of other evidence of misfeasance, malfeasance, and fraud in those States, but the authors’ statistical “proof” is nothing but a demonstration of the errors that abound in statistical analysis.

In this case, the errors resulted in the overprediction of Trump’s share of the vote in 39 States and D.C. — including, coincidentally, the six States that the authors claim to have shown were were stolen from Trump.

Election 2020 and Occam’s Razor

Occam’s razor

is the problem-solving principle that “entities should not be multiplied without necessity” or, more simply, the simplest explanation is usually the right one…. This philosophical razor advocates that when presented with competing hypotheses about the same [phenomenon], one should select the solution with the fewest assumptions, and that this is not meant to be a way of choosing between hypotheses that make different predictions.

Similarly, in science, Occam’s razor is used as an abductive heuristic in the development of theoretical models rather than as a rigorous arbiter between candidate models. In the scientific method, Occam’s razor is not considered an irrefutable principle of logic or a scientific result; the preference for simplicity in the scientific method is based on the falsifiability criterion. For each accepted explanation of a phenomenon, there may be an extremely large, perhaps even incomprehensible, number of possible and more complex alternatives. Since failing explanations can always be burdened with ad hoc hypotheses to prevent them from being falsified, simpler theories are preferable to more complex ones because they are more testable.

But simplicity isn’t a guarantee of correctness. More complexity may be necessary in order to explain a phenomenon or to make accurate predictions about it. Thus the weasel-words “without necessity”. If a thing is well explained by two independent variables, and a third independent variable adds nothing to the explanation, only two were necessary. But the number of “necessary” variables isn’t known ahead of time. It takes data-gathering, testing, and statistical analysis of the tests to determine how many are “necessary”.

Occam’s razor, in other words, is merely a tautology. The correct number of “necessary” explanatory variables is an empirical matter, not one that can be determined a priori by a vague and meaningless aphorism.

With that in mind, let us apply Occam’s razor to the presidential election of 2020. The tentative outcome of that election is a victory for Joe Biden. There are at least four explanations for the tentative outcome:

1. Every State that Biden won, he won fair and square. There were no fraudulent votes, no fraudulent counting of votes, and no errors in the counting of votes.

2. Biden’s victories in key States, though perhaps tainted by some degree of fraud or error, are legitimate; that is, the victories would have occurred absent fraud or error.

3. Biden’s victories in at least some States are illegitimate; that is, the victories wouldn’t have occurred absent fraud or error. But overturning the fraudulent or erroneous victories in some States wouldn’t change the outcome; Biden would still have enough electoral votes to be elected president.

4. Biden’s victories in at least some States are illegitimate; that is, the victories wouldn’t have occurred absent fraud or error. And overturning the fraudulent or erroneous victories in those States would change the outcome; Biden wouldn’t have enough electoral votes to be elected president. But this explanation, if true, may not be confirmed in time to change the tentative outcome of the election.

The simplest explanation, number 1, is almost certainly false. So much for Occam’s razor. What about explanations 2, 3, and 4?

I honestly do not know which of them to believe because I am withholding judgement until all of the legal votes have been counted correctly. That probably won’t happen before January 20, 2021, and so it will never happen. And so I will forever suspend judgement — but I will also suspend belief that Biden was elected honestly.

Why won’t the facts emerge before January 20, 2021? Dov Fischer explains:

[F]or those who have actual real-life professional high-stakes litigation experience, people like Rudy Giuliani and those of us who know what’s what, the reality is that no one can just walk into a courtroom a week or two after a massive fraud has taken place and just lay all the fraud on the table. It takes weeks, months, and years to unpack this stuff. No experienced attorney can just show up with all the evidence in a week or two. For example, who among us, even a week ago, had ever heard of “Dominion Voting Systems”? In only a matter of days, we now know not only of them but of their software and that they donated to the Clinton Foundation. And, oh by the way, their equipment was used in the election by North Carolina, Nevada, Georgia, Michigan, Arizona, and Pennsylvania — comprising 84 electoral college votes in six of the tightest battleground states. On the other hand, Texas rejected using them

Or take the dead voters. (Please.) Or the harvested and dumped ballots. Was it legal in the respective state to harvest the votes? If so, were the votes harvested by legally authorized harvesters — or by unauthorized out-of-state college kids who had nothing to do once it got too cold to march with Black Lives Matters thugs and threaten octogenarians at restaurants? It takes times — weeks, months — to unpeel that onion. And, again, what about the dead voters? It takes time to go through the voters’ rolls and to compare them with the rolls of the living.

And signatures. It is commonplace in litigations that, when disputes arise over signatures, handwriting experts are called in. One place to find handwriting experts is at a website called — take a seat for this one — www.handwritingexperts.com. But that is the point. When the stakes are high, you can’t just have volunteer housewives and househusbands comparing signatures. Not only valleys forge but people do, too. Could there be stakes greater than whether we have four more years of a Trump presidency or an alternative quadrennium of a Harris and Biden White House? Who is comparing the signatures on the mail-in envelopes with the actual signatures on registration rolls? How is it done? How carefully? How expertly?

It is wrong, unfair, and preposterous for media, including Fox News, regularly to parrot the Democrats and say that the Republicans so far do not have buckets and suitcases full of vote-fraud evidence. This kind of evidence — fraud — is the hardest to uncover and the hardest to gather. Mueller took two years. Durham, assuming there is such a person, already has been at it for a year and a half. States allow three years for claims of fraud. Usually, it takes document demands, demands for computer discs and drives, interrogatories, and depositions to root out the fraud and corruption. That is how long it takes. I know: I personally did this stuff for 10 years in matters entailing multi-million-dollar complex business disputes. Those of us who actually know the practice of law, not from Ally McBeal and from the 45 cable stations that simultaneously televise Law and Order reruns but from real life, know that the Trump team cannot possibly have all its evidence at hand yet.

Nevertheless,

they indeed are compiling anecdotal evidence, testimony of poll watchers who saw abuses and were kept away from monitoring ballot counting. The Trump team is gathering sworn affidavits, a recognized form of admissible evidence, and they are going as fast as they can. They report that they already have 234 sworn affidavits. Steve Cortes has published a wonderful piece raising four examples of circumstantial evidence arising from logical improbabilities:

1. Incomprehensibly high turnout in Wisconsin. For example, Milwaukee ended up with an 84 percent turnout, while a nearby Midwest city with a comparable demographic, Cleveland, had a 51 percent turnout. In all, Wisconsin reported voting by 90 percent of their registered voters. Numbers like that are off the charts. Biden inched ahead of Trump in Wisconsin by under 1 percent. By contrast, Trump’s lead in Ohio was too large to overcome with shenanigans.

2. The improbability of a lethargic Biden scoring significantly stronger voter turnouts than did an energetic Obama in certain battleground Obama districts.

3. The quirk of over 450,000 Biden-only ballots, on which the submitted ballots showed a vote for Biden but no one else at all, even in states where there were tight congressional and Senate contests down-ticket. That of course is technically possible, and certainly some such ballots could be expected. Curiously, Biden-only ballots were predominantly prevalent in battleground states like Georgia. By contrast, there were only 725 such ballots in Wyoming, which was a Trump–Republican blowout. For comparison, there was only a fraction of Trump-only ballots in Georgia.

4. The virtual absence of mail-in vetting. In New York, which tried large-scale mail-in balloting for the first time last June, the natural process of vetting saw 21 percent of ballots disqualified. Likewise, it is common that, among people mailing in ballots for the first time in their lives, usually some 3 percent get disqualified. That simply is the human nature of some who forget to sign, forget to date the ballot, fill it in wrong, and otherwise mess up. That’s people. Yet, in Pennsylvania only 0.03 percent of such ballots were rejected, 10 times fewer than all experience would have anticipated.

Circumstantial evidence matters and carries serious evidentiary weight. Murderers have been sentenced to death based solely on circumstantial evidence. Honest, reasonable minds cannot expect all evidence of fraud to be at hand only 10 days or even a month or two after the election has ended. Democrats had half a year and more to plan strategies for aspects of their fraud and ways to cover it up. If given enough time, enough production demands for computer drives and discs, enough time to read secret and deleted emails, the Trump team would have an opportunity to say “We have the evidence” or to present as Mueller did after his two-year investigation. Any shorter time frame is unrealistic.

Conclusion: The correct explanation of the tentative outcome of the presidential election of 2020 will never be known. Biden will be inaugurated on January 20, 2021 (if he lives that long), and the memory hole will swallow almost all doubts about how Biden won. Remaining doubts, and even hard evidence, will be dismissed as delusional and fabricated.

And so we will be frog-marched into a brave new world.