As promised, I have carefully reviewed the intriguing Appendix in the court filings with SCOTUS (pp. 20-29) in the case of Texas v. Georgia, et al. (specifically, the “Declaration of Charles J. Cicchetti, Ph.D.”) as well as David Post’s critique of Dr Cicchetti’s statistical analysis, and below are my tentative thoughts on this matter. (For brevity, I will limit my comments to Cicchetti’s analysis of the election results in Georgia.)
In summary, Cicchetti makes two claims: one regarding the election results in Georgia overall (let’s call this “Claim A”); the other specifically about the mail-in ballots (“Claim B”). (Also, for the reasons I will give below, I agree with David Post’s critique of Claim A, but disagree with Post’s critique of Claim B.) Let’s start with the weaker claim, shall we? Claim A is simply that, holding voter preferences from 2016 constant, it is extremely unlikely that Joe Biden could have won the Georgia contest in 2020 by honest means. Alas, this claim is total bullshit. Why? Because holding voter preferences constant makes absolutely no sense in the context of an election contest. The whole purpose of an election is to measure voter preferences, and voter preferences may change from one year to the next. That is the whole point of holding elections!
Let’s now turn to Claim B, which is the stronger of the two claims. Claim B is that the mail-in ballots are suspect and probably rigged because the final results of the mail-in ballots diverge greatly from the final results of the early-voting ballots and the in-person ballots. Specifically, Cicchetti states in Paragraph 14 of his Declaration (emphasis added by me):
“At 3:10 AM EST on November 4 the Georgia reported tabulations were 51.09% for Trump and 48.91% for Biden (eliminating third-party candidates). On November 18 at 2 PM EST, the reported percentages were Trump 49.86% and Biden at 50.14%…. For this turnaround to occur, the subsequent “late” [i.e. mail-in] ballots totaling 268,204 votes (5.4% of the votes reported on November 18) had to split 71.60% for Biden and 28.40 for Trump.”
Here, Post is wrong to dismiss the substance of this claim. Although Post is correct to conclude that we can’t hold constant voter preferences from one election to the next, we can certainly do so with randomly selected subsamples of voters in the same election. The key question, then, is whether the population of voters who voted by mail is in any relevant respect different than the population of all voters, or early voters, or in-person voters? For my part (unlike Prof Post), I remain agnostic on this question.
Addendum (12/11): I have been informed via Twitter and by some of my followers that the mail-in ballots were expected to lean toward Joe Biden because Trump told his voters to vote in person. But if that is the case (that most mail-in ballots were submitted by Democrats or left-leaning voters), why did Trump apparently hold his own with early voters? Also, either way, how could we “test” or try to prove this assertion? I wish the secretary of states of GA, PI, MI, and WI would tell us the number of “mail-in” ballots that were received and counted as well as as the number of “early ballots” and “in-person/election day” ballots.
(My most sincere apologies to William Shakespeare!) Earlier today, I requested a copy of the Appendix mentioned in Paragraph 10 on pp. 6-7 of the Texas’s Attorney General’s “Motion for Leave to File Bill of Complaint.” (See screenshots below.) As soon as I obtain this Appendix, I will conduct my own analysis of the matter and report back soon!
Update #1 (12/8, 9:32PM): I was able to track down the Appendix–it is available here, see pp. 20-29. Due to other commitments, I hope to report back by the end of this week.
Update #2 (12/9, 8:53PM): My colleague David Post has written up a strongly-worded critique of the statistical analysis in the aforementioned Appendix. I agree with most of Professor Post’s critique but disagree with one part of his analysis regarding the distribution of voter preferences on mail-in ballots. Alas, because I am currently bogged down with…
View original post 19 more words
Pingback: Pennsylvania Presidential Election Reality Check | prior probability