before accepting calculations from anonymous (or even non-anymous) sources, insist on a citation to original data so that it can be checked. I'm told (but haven't seen reference by Werise) that this comes from NYT json data static01.nyt.com/elections-asse…. I'll comment on this data.
2/ first, I extracted the data into a csv format using the R package jsonlite. I hadn't used this program before but it worked easily first try.
3/ A first and important problem with this dataset: it does NOT give actual vote counts in integers. It gives total votes and Trump and Biden share only to 3 digits. As a result, candidate counts are accurate only to nearest ~3000. Calculated increments can be off up to ~6000.
4/the dataset contains an entry at 2:14 am (a few minutes prior to supposedly smoking gun entry) which clearly reverses an entry at 1:51 am. This looks to me like garden variety accounting error correction reversing a double count. Further, it corrects more Biden votes than Trump
5/ note that (because of the imprecision of the Trump/Biden share data - only 3 digits), the calculated votes apparently reversed don't exactly match. Keep in mind that this NYT dataset is a secondary report and, while related to official count, is NOT the official count.
6/ now watch what happens at 2:18 am when you step back a little. I've highlighted the large vote increase at 2:18, but four minutes later, this entry appears to have been reversed (together with some other entries). The reversal is heavier in Biden votes than entry reversed.
7/ this analysis seems to me to be a nothingburger. I understand that there's lots to be suspicious of, but I'm not persuaded so far that one can discern fraud merely by inspecting NYT data. And definitely not in this case as presented so far.
8/ @We_Have_Risen sent a cordial reply challenging my rejection of his analysis. I've considered his reply and stand by my criticism. He did use NYT JSON data - which is what I had been told that he used and which had been used in my analysis. He provided diagram shown below.
9/ we are using the same data. I cross-checked by re-plotting data in my collation of NYT JSON data to the plot detail in his diagram. The issue is the interpretation. He focused entirely on the uptick. I observed that uptick that concerned him had accompanying downtick.
10/ in the case of the uptick just prior to the one in Werise's ellipse, I observed previously that this uptick was reversed by a matching downtick. (Excerpt below repeats earlier tweet.)
11/ similarly, the uptick that bothers Werise was almost immediately reversed in a corresponding downtick a few minutes later. These two incidents appear to me to be garden variety error corrections.
12/ since the entry in question was almost immediately reversed, it didn't have any impact on the final count. The Edison feed to NYT is not an official count in any event.
13/ it would be nice to get an explanation of the precise reason for the entry and reversal. My guess is that it will be something like the Wisconsin county where there was an erroneous data entry by collater that was quickly corrected.
14/ Bottom line: I don't see anything sinister in this sequence of data.

I think that conduct of election was unsavory and that there seem to be legitimate grievances, but this (IMO) is not one of them.

• • •

Missing some Tweet in this thread? You can try to force a refresh
 

Keep Current with Stephen McIntyre

Stephen McIntyre Profile picture

Stay in touch and get notified when new unrolls are available from this author!

Read all threads

This Thread may be Removed Anytime!

PDF

Twitter may remove this content at anytime! Save it as PDF for later use!

Try unrolling a thread yourself!

how to unroll video
  1. Follow @ThreadReaderApp to mention us!

  2. From a Twitter thread mention us with a keyword "unroll"
@threadreaderapp unroll

Practice here first or read more on our help page!

More from @ClimateAudit

12 Nov
I'm re-examining MI Oakland County from first principles, since I agree that similarity of 2016 and 2020 doesn't preclude manipulation in both elections, tho I think that it counts against it.
2/ Technically, Shiva's comparison of total margins to straight ticket margins is (at best) "exploratory" - a term of art in statistics. We don't know anything about the properties of this statistic. So Shiva's claim to have PROVED manipulation is unjustified armwaving.
3/ Having said that, merely showing (as I did last night) that there were similar patterns in 2016 Oakland Co doesn't end the discussion, as commenters rightly observed, since we still don't really understand the result.
Read 16 tweets
12 Nov
there's been considerable publicity about Dr Shiva's characterization of the plot shown at right for Oakland County MI 2020 results. However, Oakland MI 2016 plot in same format yields nearly identical results. So, unfortunately, his plot proves nothing.
2/ Dr Shiva said that the downward slope PROVED use of an algorithm to tamper with data. It doesn't. Slope has something to do with straight ticket vs all vote results; and is not due to malicious algorithm.
here's my understanding of phenomenon. Straight Ticket Republicans obviously voted for Trump. So difference arises from balance between "Paul Ryan" non-Trump Reps and "50 Cent" pro-Trump Dems. Presumably pct of Paul Ryans vs 50 Cents increases in strongly R precincts.
Read 5 tweets
7 Nov
I've put my statistical analyst hat back on and have started to parse election data.

I've noticed something very counter-intuitive in Pennsylvania data, which commentators have totally missed thus far.
2/ There is only one PA county where Trump dramatically improved Republican - Democrat margin. Without overthinking it, take a guess which one before looking at the answer.
3/ It's Philadelphia!!

Trump reduced the deficit in Philadelphia by ~44K votes from the 2016 deficit vs Hillary.
Read 26 tweets
5 Nov
Interesting to read US State Dept recent criticism of Belarus elections. "prohibition of local independent observers at polling stations" - uh, tell that to Trump campaign in Philadelphia
"detentions of peaceful protesters". "Peaceful protestors" - isn't that the term used by US media for Antifa rioters destroying cities?
"We regret that OSCE/ODIHR observers did not receive a timely invitation to monitor the vote." Uh, did OSCE/ODIHR observers even get an invitation to monitor vote in Pennsylvania, Michigan and Wisconsin? Why not? Maybe they could have prevented rampant fraud or removed suspicion
Read 6 tweets
5 Nov
PA website shows 763K mail-in ballots still to be counted with Trump lead of 165K. At split of mail in ballots observed thus far, this would indicate fairly comfortable Biden victory. But there's reason to believe that rate of Biden gain/100K count is going to slow down.
2/ PA shows outstanding mail ballots by county and elsewhere shows R/D proportion in each county.
votespa.com/About-Election…
The inventory of Philadelphia+Pittsburgh uncounted mailin is now 165K, which would prob reduce margin by ~103K. Not enough on its own.
3/ of the remaining ~600K, there is a slight majority in counties with Republican majority. For example, York, a Republican county, has 77K uncounted.
Read 7 tweets
3 Nov
@Alexey__Kovalev 1/ good that you're covering this story, but you're still not up to speed on the entire story. In addition to his purported Russian network, Steele instructed Danchenko to locate sources within the Trump campaign in the US. This instruction resulted in the most explosive and
@Alexey__Kovalev 2/ and damaging memos in the dossier - the reports (especially Report 95 right after Wikileaks drop) alleging collusion between Trump campaign and Russian intel services, including the supposed devil's bargain to hack the DNC and publish DNC hack at Wikileaks to elect Trump.
@Alexey__Kovalev 3/ the Steele dossier attributed the information in these memos to Sergei Millian. However, it can be proven that Millian never met Danchenko (or Steele). The narrative in these memos was entirely fabricated by one or both of Steele and Danchenko. None of Galkina, Podobedova etc
Read 7 tweets

Did Thread Reader help you today?

Support us! We are indie developers!


This site is made by just two indie developers on a laptop doing marketing, support and development! Read more about the story.

Become a Premium Member ($3/month or $30/year) and get exclusive features!

Become Premium

Too expensive? Make a small donation by buying us coffee ($5) or help with server cost ($10)

Donate via Paypal Become our Patreon

Thank you for your support!

Follow Us on Twitter!