© Scott Olson/Getty ImagesElection officials count absentee ballots in Milwaukee, Wis., on Nov. 4, 2020.
Before turning specifically to the 2020 election, it is important to note that election fraud is not unheard of in the United States. This is worth establishing because if we think election fraud almost never happens then we will have a higher bar of evidence for the justification of any particular claim of fraud.
In recent decades, there have been over 1,200 known instances of voter fraud due to which 20 US elections had to be overturned to declare a new winner (Heritage, 2020). Of these, 15 were overturned due to fraud involving mail based ballots (Lucas, 2020). This isn't surprising since mail in ballots have long been noted to be particularly vulnerable to fraud. This is why most European nations have either banned absentee voting or require an ID for someone to obtain such a ballot (Lott, 2020).
This fact has been attested to by American media on both the right and left. For instance, in 2012 the New York Times wrote the following: "While fraud in voting by mail is far less common than innocent errors, it is vastly more prevalent than the in-person voting fraud that has attracted far more attention, election administrators say. In Florida, absentee-ballot scandals seem to arrive like clockwork around election time." - Liptak (2012)
The piece even went as far as to cast doubt on the 2000 US presidential election results on the basis of such concerns:
"Voting by mail is now common enough and problematic enough that election experts say there have been multiple elections in which no one can say with confidence which candidate was the deserved winner. The list includes the 2000 presidential election, in which problems with absentee ballots in Florida were a little-noticed footnote to other issues...So being worried about mail-in ballot fraud, and even suggesting it could alter the results of a US presidential election, is not outside the norm.
Voting by mail also played a crucial role in the 2000 presidential election in Florida, when the margin between George W. Bush and Al Gore was razor thin and hundreds of absentee ballots were counted in apparent violation of state law. The flawed ballots, from Americans living abroad, included some without postmarks, some postmarked after the election, some without witness signatures, some mailed from within the United States and some sent by people who voted twice. All would have been disqualified had the state's election laws been strictly enforced." - Liptak (2012)
Before, I said that there are roughly 1,200 known cases of recent American voter fraud. Of course, this doesn't tell us how common voter fraud really is because it could be that the vast majority of instances are never detected. County level analysis shows that there are at least 1.8 million excess registered voters relative to the eligible population size and this at least opens the door for the possibility of voter fraud on a truly massive scale (Judicial Watch, 2020).
More importantly, non-citizens in America report voting at rates that imply at least nearly a million cases of voter fraud per year. This suggests that only a tiny proportion of the voter fraud that takes place is normally caught and that fraud at a scale that can change elections is actually the norm. Of course, these figures leave out voter fraud not involving immigrants, and so is surely a significant under estimation of the total amount of fraud that takes place.
Citation - N - Self Reported Voting - Implied Illegal Votes
- Richman et al. (2016) - 85 - 11.5% - 5,152,000
- Richman et al. (2016) - 23 - 8.7% - 3,897,600
- Richman et al. (2014) - 328 - 11.3% - 5,062,400
- Richman et al. (2014) - 364 - 3.5% - 1,568,000
- Epigone (2019) - 814 - 2.2% - 985,600
Turning to the 2020 election, there are multiple known instances where election workers were caught entering wrong numbers into the system that inflated Biden's vote county relative to trump. Of course, in each case this is claimed to have been an accident, and, of course, there is absolutely no way of showing this is true.
Citation - State - Description
- Norimine (2020) - VA - In Fairfax County, a "clerical error" caused Biden votes to be inflated by roughly 100,000.
- Lafurgey (2020) - MI - Antrim county was initially reported as going to Biden. This was so implausible that votes were recounted and it actually went to trump. The initial miscount was due to "human error".
- Payne (2020) - MI - Democrat absentee votes in Oakland county were counted twice. Once correct, the result of the local election flipped for the republican.
Comment: Another 'glitch' seems to have occurred in Wisconsin:
On election night the results of [Rock County] were presented during the evening and showed that President Trump won Wisconsin in 2016. At 10:59 the votes came in and the race was close with both candidates at around 29,000 ballots. Then by 11:12 President Trump had taken a nearly 1,000 vote lead on Biden with 31,000 votes to Biden's 30,000:At 11:21 these results had not changed much. Then at 11:43 more votes came in and they showed Trump had taken a commanding lead at 46,649 to Biden's 37,133. This was a 9,516 vote lead for Trump.
But then suddenly at 11:57 these votes had swapped. Biden was reported with 46,649 and Trump was reduced to Biden's former total of 37,133. These votes had swapped from the President Trump to Biden - again a swap from a Republican to a Democrat.
The net impact was 19,032 votes.We checked these numbers again tonight from a different source and the final numbers are still showing Biden ahead of President Trump by the same 9,516 votes.Currently the race in Wisconsin is showing Biden with a lead of 1,630,570 votes to President Trump's 1,610,030 votes. When this adjustment is confirmed Biden will only hold a 1,508 vote lead.This of course is before any of the hundreds of thousands of questionable votes showing up for Biden in Milwaukee early in the morning after the election are validated.
Election officials have said that these errors being caught is evidence that they are really good at catching errors. But there are at least two good reasons to doubt this. First, these errors were all huge and it seems unlikely that people only make huge errors and never small ones, so the lack of stories about small errors suggests that they are actually pretty bad at detecting "errors". Second, the sheer number of voter fraud convictions that have taken place, the rate at which non-citizens seem to vote, and the number of excess voters illegitimately kept on voter rolls, all seem to imply that, at the very least, this is a highly error prone process.
So we have good reasons to think they aren't great at catching errors. By contrast, we don't have any good reasons for thinking these were actually errors. The fact that all three stories about such errors each favored Biden is reasonably improbable to happen by chance (12.5%). But it's highly probable given mass election fraud, and so these stories should increase the probability we assign to such fraud having occurred.
We also have multiple eye witness accounts, mostly from election or postal workers, who claim that fraud was being engaged in and ballots were being wrongly discarded.
Citation - State - Description
- Solomon (2020) - MI - A city worker from Detroit claims he and fellow employees were told to back-date ballots that came in too late.
- Constantino et al. (2020) - MI - Poll watcher and former assistant state AG Zachary Larsen filed an affidavit stating he had seen voter fraud being committed by election workers.
- Dinan (2020) - PA - "A Pennsylvania mail carrier has said in a sworn affidavit that he was ordered by supervisors to collect and submit late ballots, which he said supervisors then backdated so that they appeared to have been mailed in time."
- Larsen (2020) - NV - "I personally witnessed disregard of signature verification as well as other irregularities," the whistleblower said in the affidavit. "While working, I observed a significant number of signatures on mail-in ballots I believe did not match the name and should have been reviewed. When I asked the supervisors, [redacted] and others, about it, instead of taking the ballots to verify the signature in the electronic database, the supervisor told me to push the envelope through without verification."
- Singman (2020) - AZ - Based on sworn testimony, it is alleged that ballots in Maricopa county featuring "over-voting" were invalidated rather than fixed.
Citation - State - Description
- Heller (2020) - PA - Election observers were kept 30 feet from the actual vote counting, meaning they could not see the ballots.
- Payne (2020) - WI - Election officials tracked down witness addresses for absentee ballots, a policy that may be illegal.
- Payne (2020) - WI - Only the state legislature can change election procedures, but election officials allowed potentially over 100,000 voters to by pass ID requirements by claiming that COVID prevented them from leaving home.
- Mills (2020) - PA - Democrat election leaders told county officials to give party operatives information about rejected mail ballots in violation of state law.
First, in PA, where by exit polls Biden won those over 65 by only 3 points (CNN, 2020), 69 people recorded as being born before 1900 requested (and most returned) ballots of which 80% were registered democrats and 16% of which were registered republicans, meaning this isn't well explained by random birth day entry errors. This can all be seen using data published on the PA government's web domain (PA.gov, 2020). Here's a screen shot of what that data looks like:
Speaking of which, this data set also tells the dates on which ballots were requested and returned. As pointed out by some guy on twitter, there are 109,421 instances in which the return date is earlier than the request date.
Given these numbers, for Biden to have a 38 points lead in MI he would have had to have got every democrat vote, every independent vote, and roughly one in four republican votes. But common sense and exit polls make it obvious that that did not happen.
To some degree these numbers will be thrown off by the fact that not everyone who requested a ballot returned a ballot, and by the fact that there are still many mail in ballots to be counted, but a huge partisan skew in this would be required to make these numbers seem remotely plausible and so this data should increase our skepticism about mail in ballots in PA and especially in MI.
A third line of statistical evidence involves Wisconsin. After the nighttime pause in reporting, vote shares shifted in favor of democrats *within* city wards, meaning this was not due to more democratic areas reporting later, and this was more true the more behind democrats were in a given ward, contrary to what an explanation based on late mail in voters would predict (Holmes, 2020).
The fourth line of statistical evidence also involved Wisconsin, and specifically Milwaukee. As noted by the BBC, extremely high turnout in certain areas is a classic sign of voter fraud. In Australia, where voting is legally mandatory, voter turnout rates reach 90-95%. Non compulsory elections should not approach these sorts of numbers (BBC, 2020).
Yet, turn out figures in many city wards in Milwaukee were initially reported to be over 100%.
This was later said to be a mistake and revised, but the revised numbers still include various wards with turnout levels that would normally be considered evidence of fraud.
A fifth and final line of evidence concerns so called bell weather counties. These are counties where the person who wins the county almost always wing the presidential race. As a set, they've correctly predicted every US election of the last 50 years, until this one.
There are other lines of evidence out there, but this article has presented the one's I find most compelling. (See here and here for problems with the Benford analyses going around). And these lines of evidence cast hundreds of thousands of votes in doubt, more than enough to alter the result of the presidential election.
As I've argued, voter fraud is reasonably common in the US and so we would be unjustified in dismissing the case for fraud in this election out of hand.
We should also note that, if fraud were to occur, the sort of evidence we'd expect to have is exactly the sort of evidence we do have. Even if fraud were to occur at a mass level, it would be surprising if we had evidence even better than this, if we had video of fraud being committed clearly in view of the camera, or we had a mass wave of admission to thousands of instances of fraud. Given this, we should not require this sort of utopian evidence to be convinced that fraud likely occurred. Instead, we should require the exact sort of evidence I've presented and so we are justified in thinking significant fraud occurred until we are presented with equally powerful arguments to the contrary.
Importantly, these arguments have not shown that Trump would have certainly won the election absent any voter fraud. But they do show that the official election results do not justify us in thinking that Biden would have won a fair election either. Given this, to ensure fairness we need to either find a way to remove ballots that are fraudulent, or throw out the entire election and have a re-do. Of course, something being in the interest of fairness is no guarantee that it will occur. I am not predicting that these things actually will happen, but hopefully I've been successful in showing that they should happen.
Comment: Statistician William Briggs argues just two anomalies in the Pennsylvania counts were enough to give Biden the edge over Trump. From Epoch Times: Briggs posted another analysis: And here's a concise rundown on some of the statistical absurdities of Biden's "win":