By Andy May
In my last post I discussed a peer-reviewed study of the 2020 election by Dr. John L. Lott. It examined the vote totals in six critical states that were accused of voter fraud due primarily to illegally changing voter procedures and inadequately verifying the authenticity of the votes cast. Texas filed a lawsuit with the Supreme Court to challenge the vote tallies in critical states that violated their own laws and the Constitution (Article 1, Section 4), but the Supreme Court would not hear the case, claiming that Texas did not have standing to sue. This was correct, Texas cannot control the vote tallies in another state, but the Supreme Court ruling does not address the clear violations of law by judges, bureaucrats and Secretaries of State in the critical states.
Elections set up a legal conundrum. The Constitution leaves the selection of presidential electors to the states, and they select them by public vote or a vote of the state legislature. The only restrictions are that the state legislature must decide on the details of the voting procedures. As a result, the Supreme Court is not part of the process, only the U.S. Congress can intervene, and they are unlikely to do so. The only time in U.S. history, that I know of, when elections were as distrusted by the public as much as now, was in the late 19th century from just before the Civil War until around 1880 when voting booths became common, so people could vote in private. Later reforms that helped were voter registration, which reached most states by World War I. By 2013 most states also required that the voter have identification to vote as well. These reforms created the public trust in the elections required for a republic to function.
Lott did his analysis by comparing the recorded votes in the suspect precincts to neighboring precincts to see if the differences were consistent with previous elections. The results were not definitive, but he did find statistically significant differences between the 2016 election results and the 2020 election, that were consistent with election fraud. The illegal procedures used in 2020 were due to the covid scare. Ballots were generally verified properly and legally in 2016.
There is another recent peer-reviewed study of the 2020 election that was referenced in Lott’s paper, but I missed it when reading his paper and bibliography. It was published in PNAS and written by Andrew Eggers, Haritz Garro, and Justin Grimmer (Eggers, Garro, & Grimmer, 2021). Herein referred to as “Eggers.” Eggers conclude that none of the statistical arguments made by Trump supporters, including Lott’s, to support their fraud allegations are “even remotely convincing.” In fact, the title of their article contains the phrase: “No evidence for systematic voter fraud.” Both phrases are clearly and obviously wrong. Certainly evidence exists of voter fraud, in fact numerous people have been convicted of election fraud due to their actions during the 2020 election, many more are charged with crimes including the Arizona Secretary of State. Voter fraud exists in every election, in 2020 it was made much easier, and thus worse, when states often did not even check absentee ballot signatures, as required by law.
While evidence of fraud, and systematic fraud, exists, what is missing is evidence that the fraud could change the election outcome. This is a very difficult hill to climb. Jimmy Carter and James Baker warned us of this possibility, with respect to mail-in and absentee voting, in their 2005 report on election integrity (Carter & Baker, 2005).
Eggers do effectively challenge several statistical studies unrelated to Lott’s analysis, but only tries to challenge two of the three statistical techniques used by Lott. In particular, Eggers contend that the precinct level absentee ballot statistical study performed by Lott was dependent upon the order in which the data was entered into the analysis, leading to invalid statistical results. However, they were addressing a very early preprinted version of Lott’s paper that was never published. In the final version, Lott corrects this problem, and explains how. Eggers were sent the final version of the paper in plenty of time for them to correct their own paper but ignored it. This suggests that the Eggers paper was disingenuous, and not serious or scientific.
Further evidence that Eggers was not a serious academic article is they ignored and attempted to sweep away, using obscure arguments, the strongest part of Lott’s study. Even with many other possible determinants held constant, areas with alleged fraud had statistically significant correlations with voter turnout. Eggers then cherry picked other data to compare to the suspect counties and make a case that with their new dataset the correlation goes away. This “strawman” test is very different than Lotts, neither proves anything, but both can be used as evidence. They do not refute each other; they are too different. It is up to the reader to decide which test framework they think is valid and which isn’t, it reduces to a matter of opinion.
Their discussion of the turnout analysis can be found in their supplementary materials in section H, the link to the supporting information is near the bottom of the page linked. In section H, Eggers’ provides their opinion that a statistically significant change in turnout does not mean anything. They also claim that local elections, or other factors, might explain changes in turnout. These are all possible explanations, but Lott is not claiming to have proven fraud, he is only claiming that the change in turnout is statistically significant and that it is consistent with fraud, which is true. Eggers’ analysis does not change that.
Eggers also, rather unscientifically, attempts to speculate why Trump and his allies chose the counties to accuse of fraud. Eggers do not dispute that the changes in turnout could be a sign of fraud, they just intuitively believe the amount of the change in turnout that Lott discovered was too small, even though it met Lott’s statistical threshold. Then Eggers set up a different statistical test where Lott’s idea fails, but they do not refute Lott’s test. Their strawman argument is not scientific; it is merely different opinion on how to frame the test.
Lott also examines the provisional ballots issued, often illegally, in the suspect counties and finds a statistically significant correlation there as well. Eggers do not address this correlation, which is stronger than the absentee ballot correlation. The share of Biden’s votes from provisional ballots is significantly higher in Pennsylvania’s Allegheny County that in adjacent precincts.
Much of this is discussed in competing Washington Times op-eds, one by John Lott here, and another by Grimmer and Hall, here. Both the articles and the papers are worth reading. The papers both make some valid points and, for the most part, the conclusions of the two papers are compatible. Most of the differences are in the framing of the tests, tone and opinion, although Eggers’ title and “even remotely convincing” are clearly falsehoods. Eggers avoided the best evidence of fraud presented by Lott, presumably because Lott was correct. That is, the voter turnout, absentee, and provisional ballot patterns in the precincts accused of fraud were statistically anomalous.
While Lott’s analysis suggests irregularities in voting shown by his precinct-to-precinct comparisons in Pennsylvania are not significant statistically, when they are combined with the Georgia data they are. This is another point ignored by Eggers.
Lott merely presents statistical evidence that there could have been voter fraud in 2020, not proof that it did. The problem with the 2020 election is that election officials in several critical states broke their own state laws and the provision in the Constitution that only state legislatures can specify “The Times, Places and Manner of holding Elections.” There were far too many votes cast and counted in a way that precluded verifying that the votes and voters were legitimate. Two hundred years of voting reform to ensure votes can be verified were thrown out the window in 2020, using covid as an excuse. Now we have tens of millions of Americans that do not believe the results. It is a dangerous situation and must be rectified in the next election.
The bibliography can be downloaded here.
October, 2021; the final version is dated March 21, 2022 ↑