How 2020 Presidential Election Data Indicate Fraud

There have been dozens of claims of election fraud from across the United States during the 2020 presidential election and while eye witness accounts may be enough to initiate an investigation, it may not be enough to prove fraud.

Statistics are much more convincing and we’ve seen how Benford’s law indicates there was widespread fraud throughout the country. One statistician who goes by the Twitter handle @APhilosophae has looked at the election numbers from another angle to come to the same conclusion.

An anonymous data scientist and another unidentified person used a script to scrape election data from the New York Times website as it came in throughout the election. These data are reportedly from the Edison vote tabulation software, which ABC, CBS, CNN, and NBC also use, but which is not publicly available.

The dataset, which can be found here, includes the incremental vote totals that were reported through Edison to the news media. @APhilosophae took these data and create graphs for each state. He noticed that, for most states, the data revealed a consistent ratio of Biden to Trump votes, which produces a line of data points with dozens of additional outliers.

Here is the ratio map for Texas:

The line may seem anomalous but it is due to the fact that groups of mail-in ballots were randomized in the mail sorting process. It’s as if you gathered all the eventual ballots and shuffled them together, then reported on groups of them at a time. They would all have roughly the same ratio of Biden to Trump votes.

This in itself is not an indication of fraud; it’s expected and is consistent across all states.

When you plot these data points versus time, another pattern emerges as in the case for Florida:

The initial reporting (on the left of the above graph), which is widely scattered, represents in-person voting from heavily partisan precincts. As the election counting progresses, the data points get closer to the randomized ratio, which should reflect the eventual ratio for the entire state.

Here’s Minnesota:

This trend is what we see in every state except for several anomalies: Wisconsin, Michigan, Pennsylvania, Georgia, and Virginia.

In Wisconsin, for instance, we see the ratio line forming around 10pm and becoming clear by midnight. Then, abruptly at 4am November 4, the ratio shifts to favor Biden:

If the mail-in ballots consisted of the same sample as the other state-wide mail-in ballots, we would expect to see the same ratio as previous drops, but that’s not the case. The new group of ballots were delivered at 3am as documented by award-winning Milwaukee journalist Matt Smith:

Milwaukee’s 169,000+ absentee ballots en route to county election officials pic.twitter.com/kXTiAJ8Xsp
— Matt Smith (@mattsmith_news) November 4, 2020

One of two things happened:

This group of mail-in ballots were sent in from more pro-Biden areas and inexplicably were not shuffled in with the rest of the state’s mail in ballots as was the case in nearly every other state or the mail-in ballots previously counted in Wisconsin and they were only able to find and deliver these ballots seven hours after voting ended.
Thousands of illegitimate ballots were added to a legitimate group of mail-in ballots to flip the state in favor of Biden.

Georgia, Virginia, and Michigan show similar blips upward toward Biden:

In Pennsylvania, we see the ratio line starting to form around .75 Biden to Trump votes, then counting stopped for about an hour and the next batch shows a steep increase in Biden votes to Trump votes: