Based upon Plato’s attribution,[1] philosophers credit pre-Socratic philosopher Heraclitus, who was in his prime about 500 B.C., for the oracular observation that πάντα χωρεῖ και οὐδε ν μένει, or in more elaborative English:
all things pass and nothing stays, and comparing existing things to the flow of a river, he says you could not step twice into the same river.
Time changes us all. Certainly 2016 is not 2020, and the general elections held in November of those two years were not the same elections, and certainly not the same electorate. No one would need a statistician to know that the population of voters in 2016 was different from that in 2020. Inevitably, some voters from 2016 died in the course of the Trump presidency; some no doubt died as a result of Trump’s malfeasance in handling the pandemic. Inevitably, some new voters came of age or became citizens and were thus eligible to vote in 2020, when they could not vote in 2016. Some potential voters who were unregistered in 2016 became new registrants. Non-voters in 2016 chose to vote in 2020, and some voters in 2016 chose not to vote in 2020. Overall, many more people turned out to vote in 2020 than turned out in 2016.
The candidates in 2016 and 2020 were different as well. On the Republican side, we had ostensibly the same candidate, but in 2020, Trump was the incumbent and had a record of dismal moral and political failures, four years in duration. Many Republicans who fooled themselves into believing that the Office of the Presidency would transform Trump into an honest political actor, came to realize that he was, and always has been, and always will be, a moral leper. These “apostate” Republicans effectively organized across the country, through groups like the Lincoln Project and the Bulwark, against Trump, and for the Democratic candidate, Joseph Biden.
In the 2016 election, Hilary Clinton outspent Donald Trump, but Trump used social media more effectively, with a big help from Vladimir Putin. In the 2020 election, Russian hackers did not have to develop a disinformation campaign; the incumbent president had been doing so for four years.
On the Democratic side of the 2016 and 2020 elections, there was a dramatic change in the line-up. In 2016, candidate Hilary Clinton inspired many feminists because of her XX 23rd chromosomes. She also suffered significant damage in primary battles with social democrat Bernie Sanders, whose supporters were alienated by the ham-fisted prejudices of the Clinton-supporters on the Democratic National Committee. Many of Sanders’ supporters stayed home on election day, 2016. In 2020, Sanders and the left-wing of the Democratic party made peace with the centrist candidate Joseph Biden, in recognition that the alternative – Trump – involved existential risks to our republican democracy.
In 2016, third party candidates, from the Green Party and the Libertarian Party, attracted more votes than they did in 2020. The 2016 election saw more votes siphoned from the two major party candidates by third parties because of the unacceptable choice between Trump and Clinton for several percent of the voting public. In 2020, with Trump’s authoritarian kleptocracy fully disclosed to Americans, a symbolic vote for a third-party candidate was tantamount to the unacceptable decision to not vote at all.
In 2016, after eight years of Obama’s presidency, the economy and the health of the nation were good. In 2020, the general election occurred in the midst of a pandemic and great economic suffering. Many more people voted by absentee or mail-in ballot than voted in that manner in 2016. State legislatures anticipated the deluge of mail-in ballots; some by facilitating early counting, and some by prohibiting early counting. The Trump administration anticipated the large uptick in mail-in ballots by manipulating the Post Office’s funding, by anticipatory charges of fraud in mail-in procedures, and by spreading lies and disinformation about COVID-19, along with spreading the infection itself.
On December 8, 2020, without apparently tiring of losing so much, the Trump Campaign orchestrated the filing of the big one, the “kraken lawsuit.” The State of Texas filed a complaint in the United States Supreme Court, in an attempt to invoke that court’s original jurisdiction to adjudicate Texas’ complaint that it was harmed by voting procedures in four states in which Trump lost the popular vote. All four states had certified their results before Texas filed its audacious lawsuit. Legal commentators were skeptical and derisive of the kraken’s legal theories.[2] Even the stalwart National Review saw the frivolity.[3]
Charles J. Cicchetti[4] is an economist, who is a director at the Berkeley Research Group. Previously, Cicchetti held academic positions at the University of Southern California, and the Energy and Environmental Policy Center at Harvard University’s John F. Kennedy School of Government. At the heart of the kraken is a declaration from Cicchetti, who tells us under penalty of perjury, that he was “formally trained statistics and econometrics [sic][5] and accepted as an expert witness in civil proceedings.”[6] Declaration of Charles J. Cicchetti, Ph.D., Dec. 6, 2020, filed in support of Texas’ motion at ¶ 2.
Cicchetti’s declaration is not a model of clarity, but it is clear that he conducted several statistical analyses. He was quite transparent in stating his basic assumption for all his analyses; namely, the outcomes for the two Democratic candidates, Clinton and Biden, for the two major party candidates, Clinton versus Trump and Biden versus Trump, and for in-person and for mail-in voters were all randomly drawn from the same population. Id. at ¶ 7. Using a binomial model, Cicchetti calculated Z-scores for the observed disparities in rates, which was very good evidence to reject the “same population” assumptions.
Based upon very large Z-scores, Cicchetti rejected the null hypothesis of “same population” and of Biden = Clinton. Id. at ¶ 20. But nothing of importance follows from this. We knew before the analysis that Biden ≠ Clinton, and the various populations compare were definitely not the same. Cicchetti might have stopped there and preserved his integrity and reputation, but he went further.
He treated the four states, Georgia, Michigan, Pennsylvania, and Wisconsin, as independent tests, which of course they are not. All states had different populations from 2016 to 2020; all had no pandemic in 2016, and pandemic in 2020; all had been exposed for four years of Trump’s incompetence, venality, corruption, bigotry, and bullying. Cicchetti gilded the lily with the independence assumption, and came up with even lower, more meaningless probabilities that the populations were the same. And then he stepped into the abyss of the fallacy and non sequitur:
“In my opinion, this difference in the Clinton and Biden performance warrants further investigation of the vote tally particularly in large metropolitan counties within and adjacent to the urban centers in Atlanta, Philadelphia, Pittsburgh, Detroit and Milwaukee.”
Id. at ¶ 30. Cicchetti’s suggestion that there is anything amiss, which warrants investigation, follows only from a maga, mega-transposition fallacy. The high Z-score does not mean that observed result is not accurate or fair; it means only that the starting assumptions were outlandishly false.
Early versus Late Counting
Texas’ claim that there is something “odd” about the reporting before and after 3 a.m., on the morning after Election Day fares no better. Cicchetti tells us that “many Americans went to sleep election night with President Donald Trump (Trump) winning key battleground states, only to learn the next day that Biden surged ahead.” Id. at ¶ 7.
Well, Americans who wanted to learn the final count should not have gone to sleep, for several days. Again, the later counted mail-in votes came from a segment of the population that was obviously different from the in-person voters. Cicchetti’s statistical analysis shows that we should reject any assumption that they were the same, but who would make that assumption? These expected values for the mail-in ballots differed from the expected values for in-person votes; the difference was driven by Republican lies and disinformation about Covid-19, and by laws that prohibited early counting. Not surprisingly, the Trumpist propaganda had an effect, and there was a disparity between the rate at which Trump and Biden supporters voted in person, and who voted by mail-in ballot. The late counting and reporting of mail-in ballots was further ensured by laws in some states that prohibited counting before Election Day. Trump was never winning in the referenced “key battleground” states; he was ahead in some states, at 2:59 a.m., but the count changed after all lawfully cast ballots had been counted.
The Response to Cicchetti’s Analyses
The statistical “argument,” such as it is, has not fooled anyone outside of maga-land.[7] Cicchetti’s analysis has been derided as “ludicrous” and “incompetence, by Professors Kenneth Mayer and David Post. Mayer described the analysis as one that will be “used in undergraduate statistics classes as a canonical example of how not to do statistics.”[8] It might even make its way into a Berenstain Bear book on statistics. Andrew Gelman called the analysis “horrible,” and likened the declaration to the infamous Dreyfus case.[9]
The Texas lawsuit speaks volumes of the insincerity of the Trumpist Republican party. The rantings of Pat Robertson, asking God to intervene in the election to keep Trump in office, are more likely to have an effect.[10] The only issue the kraken fairly raises is whether the plaintiff, and plaintiff intervenor, should be be sanctioned for “multipl[ying] the proceedings in any case unreasonably and vexatiously.”[11]
[1] Plato, Cratylus 402a = A6.
[2] Adam Liptak, “Texas files an audacious suit with the Supreme Court challenging the election results,” N.Y. Times (Dec. 8, 2020); Jeremy W. Peters and Maggie Haberman, “17 Republican Attorneys General Back Trump in Far-Fetched Election Lawsuit,” N.Y. Times (Dec. 9, 2020); Paul J. Weber, “Trump’s election fight puts embattled Texas AG in spotlight,” Wash. Post (Dec. 9, 2020).
[3] Andrew C. McCarthy, “Texas’s Frivolous Lawsuit Seeks to Overturn Election in Four Other States,” Nat’l Rev. (Dec. 9, 2020); Robert VerBruggen, “The Dumb Statistical Argument in Texas’s Election Lawsuit,” Nat’l Rev. (Dec. 9, 2020).
[4] Not to be confused with Chicolini, Sylvania’s master spy.
[5] Apparently not formally trained in English.
[6] See, e.g., K N Energy, Inc. v. Cities of Alliance & Oshkosh, 266 Neb. 882, 670 N.W.2d 319 (2003), Center for Biological Diversity v. Pizarchik, 858 F. Supp. 2d 1221 (D. Colo. 2012), National Paint & Coatings Ass’n, v. City of Chicago, 835 F. Supp. 421 (N.D. Ill. 1993), National Paint & Coatings Ass’n, v. City of Chicago, 835 F. Supp. 414 (N.D. Ill. 1993); Mississippi v. Entergy Mississippi, Inc. (S.D. Miss. 2012); Hiko Energy, LLC v. Pennsylvania Public Utility Comm’n, 209 A.3d 246 (Pa. 2019).
[7] Philip Bump, “Trump’s effort to steal the election comes down to some utterly ridiculous statistical claims,” Wash. Post (Dec. 9, 2020); Jeremy W. Peters, David Montgomery, Linda Qiu & Adam Liptak, “Two reasons the Texas election case is faulty: flawed legal theory and statistical fallacy,” N.Y. Times (Dec. 10, 2020); David Post, “More on Statistical Stupidity at SCOTUS,” Volokh Conspiracy (Dec. 9, 2020).
[8] Eric Litke, “Lawsuit claim that statistics prove fraud in Wisconsin, elsewhere is wildly illogical,” PolitiFact ((Dec. 9, 2020).
[9] Andrew Gelman, “The p-value is 4.76×10^−264 1 in a quadrillion” Statistical Modeling, Causal Inference, and Social Science (Dec. 8, 2020).
[10] Evan Brechtel, “Pat Robertson Calls on God to ‘Intervene’ in the Election to Keep Trump President in Bonkers Rant” (Dec. 10, 2020).
[11] See “Counsel’s liability for excessive costs,” 28 U.S. Code § 1927.
