I've retrieved all of The New York Times' 2020 presidential datasets, which have a time-series of voting data from a company called Edison Research. Here, for instance, is Pennsylvania's.
What I found is pretty fascinating, especially in Pennsylvania.
As just one example, here is a graph depicting the shifts in votes starting on election day. The X-axis is the date/time and the Y-axis represents the change in votes (positive values denote shifts for Trump, negative values represent shifts for Biden, in hundreds).
May I call your attention to the differences between the left and the right sides of the graph?
Please note that after about 23:00 hours on 11/4/20 Zulu, virtually all vote swings went for Biden.
But you need to see the data. Below is a spreadsheet with the time-series.
Trump's running vote count is in column C, and Biden's is in column D. The change in Trump's vote total from the prior count is in column E and Biden's is in column F. Lastly, column G shows the difference in vote swings between columns E and F.Do me a favor and click on the spreadsheet below. Then scroll down until you start seeing highlighted cells. Then continue to scroll and tell me if that looks legit.
You'll notice that after the 11/4/20 23:00Z cutoff time, only swings of Biden votes (in multiples of 6,000 at a time) seem to have been received.
Statistically, these changes make no sense. They're not just improbable, they appear to be impossible.
It looks to me like bundles of around 6,000 votes were used to slowly overcome any Trump lead. The correlation is uncanny.
I'll have more info from other states --- that show similar bizarre trends -- shortly.
In the mean time, your thoughts are appreciated.
Update:- Startling analysis of Georgia, where the same pattern occurs in bursts of 4,800 vote swaps
- Likewise, in Michigan, a "massive vote dump [occurred] at 6AM Wednesday (96% for Biden) used to overcome huge Trump lead"
Could the legislatures of Trump-won red states try to lead by example and pass a generically worded resolution like:
ReplyDelete“Due to the nationwide concert about overwhelming fraud, this legislature is exercising its Constitutional authority to directly select a slate of electors.”
This would not change the electoral votes of a state that Trump already won, but if a cascade of identical resolutions like this happened, it could help motivate the legislatures of critical swing states to do the same
Watched the abuses in DC, and "I can feel it coming in the air tonight", no more will Republicans be wronged, insulted, or laid hands on. They don't do these things to other people, and they require the same from them."
ReplyDeleteThe time for patience is over. We reject the Religion of the Left.
The mere drop in votes at 11/4/2020 2:22 is ample reason to stop any certification until full investigation.
ReplyDeleteAsk, on shredded Trump ballots, how did they register them as counted?
Not getting negative ammounts on both candidates, Trump has total of 408,537 in 31 times plus a minus 1 in 1 box. Biden has 636,974 in 17 times plus 1 minus 1 in 1 box, 613,012 in just 2 boxes.
ReplyDeleteDoug - the missing element in all these drops is the physical origin of the values. Without a specific point of origin, the values are meaningless. Aggregations of counts, which are supposed to represent real ballot sheets must have a traceback, and contain the audit trail to the individuals that performed the voting and their marks upon the ballot sheets. In fact, if you cannot sufficiently demonstrate that traceback to the primary source, then the values themselves are of no merit to the integrity of the election.
ReplyDeleteMath and graphs aside.
ReplyDeleteWould it help that people found 6k votes switching from Trump to Biden in Michigan? 6,000 vote increments may be another trail to follow.
https://www.truthorfiction.com/did-an-antrim-county-michigan-software-glitch-send-6000-trump-votes-to-biden/
https://www.dailymail.co.uk/news/article-8923165/GOP-claims-software-glitch-Michigan-incorrectly-gave-6000-votes-Biden.html
Overall, I am confused by the negative vote counts. Earlier in the day votes were taken away from both Trump and Biden. If you've counted a group of votes for one candidate, why would they be taken away?
ReplyDeleteLater, though, the only negative votes counts were only applied to Trump. Why? Just a quick scan seems to show that giving back the votes that were taken away in the late hours could push Trump over the top.
Report a case: djt45.co/stopfraud Call: (800) 895-4152
ReplyDelete"Overall, I am confused by the negative vote counts."
ReplyDeleteIt's "net change in Trump's lead" - positive means Trump gained more votes than Biden in that batch of counting, negative means Biden gained more votes in that batch than Trump.
No actual losses of votes, just net relative change.
"No actual losses of votes, just net relative change."
ReplyDeleteLook at the Trump and Biden deltas (delta = "change"). Both show votes being removed, which you can verify in the totals columns.
"Both show votes being removed, which you can verify in the totals columns."
ReplyDeleteBloody hell, you're right! Didn't notice that before... wow, that's just ridiculous.
Hi Doug,
ReplyDeleteVery thought-provoking analysis!
May I ask you a question? Examining the JSON file from the NYT I don't see the incoming votes explicitly marked by candidate.
Did you get the incoming votes _explicitly_ from this file, or did you calculate them _implicitly_, using inputs like "Trump has a 0.531 share and Biden has a 0.456 share of 1,765,944 current total"?
Best regards, Steve
The data from the New York times gives total vote counts and the percentages of those counts that Trump and Biden have, but only to a precision of three digits, or 0.001. For example:
ReplyDelete{"vote_shares":{"bidenj":0.497,"trumpd":0.49},"votes":6758279,"eevp":98,"eevp_source":"edison","timestamp":"2020-11-08T19:58:46Z"}
Therefore, we can only calculate the votes for Biden and Trump to a maximum of 3 significant figures. The corresponding line in the spreadsheet:
11/8/2020 19:58 3311556 3358864 -47308 -6083 684 -6767
The total vote in the spreadsheet is 3,311,556 + 3,358,864 = 6,670,420. The smallest vote percentage change in the Edison data is 0.001 * 6,670,420 = 6,670 and that is exactly what is seen. We cannot know the votes more accurately than this from this data stream. These are not multiples of 6,000, they are 0.001 of the total vote count at any given time. Since there are around 6 million votes at this time, the incremental change will be around 6,000. No mystery here.
However, lines like the following where large numbers of votes were switched from Trump to Biden cannot be explained by the above:
11/4/2020 4:08 1671332 1271406 399926 -17876 17930 -35806
That is not a rounding error.
R.W.
Thank you, R.W., you're thinking exactly what I'm thinking.
ReplyDeleteI believe that the Dems cheated. I'm willing to accept that our host has found proof.
But I no longer think the same-size "bundles" are real. I did at first, and found it extremely suss, but examining the data more closely I came to the exact same conclusion as RW, that the bundles are an artifact of the analytical process. In PA and in GA, where the "bundles" are 4,800 = 1/1000 of the 4.8M total.
The "bundles" part of the argument does not hold up. To nail the Dems as the frauds they are, we need to focus elsewhere.
Best regards, Steve
Hi Doug, I completely know what you are seeing. Here is the document I produced in 2016 when following the Democrat Primary with Sanders vs Clinton. No one knew what I had found except that I knew votes had been deleted and flipped. https://www.dropbox.com/h?preview=Summary-of-Primaries-NY-to-CA.pdf - I have the data as well, some 2.5 GBs. Good on you for finding this and sharing with the world. Hope to connect with you sometime. JPR
ReplyDeleteI agree with Steve and R.W. The "bundles" are a result of the limited precision of the vote totals when calculated based on a percentage that is only specified to three decimals.
ReplyDeleteIs anyone aware of a data feed that has the vote counts for each candidate EXPLICITLY specified at each timepoint?
Damning evidence but I want to verify the source. I clicked on the NY Times link that they provided and it gives me a bunch of code. Where can we confirm where all of this real-time election feed results were coming from?
ReplyDeleteAnonymous 10:20 PM:
ReplyDeleteI agree, explicit vote counts are the gold standard. This data file does NOT contain that, not that I have seen.
I respect Doug's effort to reconstruct the actual vote numbers from votes x vote_shares, but those reconstructions are estimates, good to about one part per thousand.
That introduced a systemic error and a degree of uncertainty into the data. Any reasoning that we attempt about effects smaller than 1ppt must be done with care.
Best regards, Steve
I have looked at this data for PA also.
ReplyDeleteUnfortunately, I don't think it's adequate to prove anything. As others have pointed out, the vote numbers are only accurate to three places, so whenever Trump's share ticks down, for example, from .491 to .490, you see an apparent drop of about 6000, but it's not real. The data also doesn't tell us what county the data is coming from, and counts from different precincts are all bundled up together.
It DOES tell us some things though. It tells us that 7.23 hours into the counting it stopped, and didn't resume for 6.05 hours. At that point, Trump was ahead about 680,000 votes.
About halfway though the 6.05 hour gap, there were two isolated additions. The first was small. The second was about 65,000 votes, 92.5% of which went to Biden. So that's suspicious.
After the gap, counting resumes for something like 12 hours, and 73.6% of about 950,000 votes went to Biden. The counting is markedly slower.
Then it stops AGAIN, this time for 9.75 hours. Counting continues for about 18 more hours. This time, 73.5% of the votes go to Biden (almost the exact same fraction). This time it's about 290,000 votes, so they took 50% longer, to count 1/3 as many votes. At this point, Trump's lead had dropped to about 20,000.
The next stop is 5.88 hours. About 94,000 votes are counted, 78.5% for Biden, and Biden takes the lead.
There are more gaps, but only about 7,000 more votes.
So are these the mail-in ballots, that were expected to break for Biden? Were they the urban counties? More information is needed.
But it DOES seem to follow the pattern of the vote-fixing Dominion machine, stopping and resetting to get a certain fraction of the votes awarded to your candidate.
About 1.3 hours into it, a significant number of votes suddenly disappear, at three times that are close together, but add all three together and Trump loses 315k but Biden loses even more, 600k.
So we can't really tell much. It seems awfully suspicious that suddenly over 70% of the vote went to Biden after the first pause, and it seems suspicious that that continued JUST as long as it needed to. We'd need more information.
FJS
So, while watching the CNN live results on the 4th, I screen captured one of these PA "switches" but I dont see either the before or after counts in the PA data provided here. These are the numbers I saw "switch":
ReplyDeleteBiden before: 1,252,537
Trump before: 1,690,589
Biden after: 1,272,495
Trump after: 1,670,631
Or, for a better visualization...
Trump before: 1,690,589
Trump after: 1,670,631
Difference: -19,958
Biden before: 1,252,537
Biden after: 1,272,495
Difference: +19,958
As you can see, Trump's count decreased by 19,958
While Biden's count increased by 19,958
But I dont see these numbers on the PA spreadsheet
How could I see this kind of data from all states?
ReplyDeleteWould you present an updated version for Georgia and Pennsylvania?
Thanks for this work!!! It is very eye opening!