You can add location information to your Tweets, such as your city or precise location, from the web and via third-party applications. You always have the option to delete your Tweet location history. Learn more
DEBUNKING THREAD: I'm an actuary (which is basically just a pretentious, worse data scientist), and I'm here to tell you why this thread is both deliberate misinformation and the most idiotic data analysis I've ever seenhttps://twitter.com/APhilosophae/status/1325592112428163072 …
Basically, the thread author takes incredibly high-level NYT election data and claims that spikes like this are "evidence of ballot backdating, manufacturing of software tampering," which is laughable because it's based entirely on number ROUNDING in the source datapic.twitter.com/oj4HhkzDVK
So, someone wrote a program to scrape the data behind graphs like this:pic.twitter.com/WOops0zwF5
The problem with this is that the data behind this is really high-level. The only data that is used in the "analysis" are the number of votes and the % share of all CUMULATIVE votes that are Dem or Rep. The columns that represent vote shares are rounded to the 0.0% placepic.twitter.com/1vjNOpmECv
How did the thread author get from these basic columns to those fancy looking graphs? By making dumb mistakes. Because the NYT source doesn't give exact Dem and Rep votes, the thread author backs into them (ex. by taking % Dem times total votes)pic.twitter.com/SbYtzY69wN
This would be fine if the % Dem and % Rep figures were exact, but instead, they are rounded to the 0.0% place (which is way too few decimal places to try for this sake). The author takes these calculated fields and backs into another field - marginal Dem/Rep votes in each batchpic.twitter.com/aZmKviyxUK
This is, erm, a problem. Take this highlighted row for example: this is a batch of 50 new votes in Wisconsin, and it didn't change the CUMULATIVE Dem vote share from 49.4%. How can we figure out what % of THIS BATCH was for Biden?pic.twitter.com/1cK5pZkjko
The (hopefully obvious) answer is that we can't. It could have been 50-0 for Biden, 50-0 for Trump, or 25-25: no matter what, the % Dem would stay the same at 49.4% because it's rounded and a batch of 50 votes is obviously far too small to change that CUMULATIVE figure.
The mistake that the author makes is implicitly assuming that this batch will be 49.4% for Biden. This batch is circled in red here. Again, WE HAVE NO WAY OF KNOWING FROM THE SOURCE DATA HOW MANY VOTES BIDEN OR TRUMP GOT IN THIS BATCH. Yet, the author assumes it to be 49.4%.pic.twitter.com/IGKLlqymLp
So how do we explain this spike the author harps on (red circle, 1st pic)? Well for STARTERS, we could expand the axis to look beyond just 0 to 2. When we do this, note the green circle (pic 2). This represents the famous 4am cache of 170k ballots that broke for Biden by 120k.pic.twitter.com/RVejb6V9RD
This HUGE Biden batch (highlighted below) unsurprisingly pushes the cumulative Biden % way up from 47.3% to 49.3%. The clusters in the above red circle each represent SMALL batches (like the example of the 50 vote batch)...pic.twitter.com/1NvGcX9fwg
Remember the author implicitly assumes each small batch breaks exactly equal to the CUMULATIVE vote share. So, each of the batches in the red circle did not ACTUALLY have D/R ratios around 1 - they could have been anything (and were likely mostly >1, bc they were mostly mail in)pic.twitter.com/BPVDLdn9Nc
In fact, the line of dots that always hovers around 1 - these are each just small batches that were too tiny to move the cumulative vote share. I have a hard time believing the author didn't realize that - they are either an idiot or nefarious (probably both).
This is what explains each of the "anomalies" that the author sees: a big pro-biden batch fell in the middle of the tiny batches and drove up the cumulative Biden vote share. If this sounds profoundly silly, it's because it is. It was asinine to perform a data "analysis" this way
Just to be thorough, here's the same thing for PA. In red is the "anomaly", in green is the rational, obviously-not-fraud explanation. For PA it was a bunch of somewhat-large mail in batches (rather than the one big WI batch).pic.twitter.com/SbvA2BBGuU
TLDR: the author had to back into the data he uses, and the way he does so is stupid and pointless because of how the data is rounded. Here's the source data I used (I didn't run the program myself, I just downloaded the .csv file)https://twitter.com/datafox21/status/1325620144215891968?s=20 …
I haven't seen anything debunking this yet, so please share this to combat the spread of misinformation. I don't care if you copy or co-opt my thread here without crediting me - I just want to help debunk this constant stream of conspiratorial BS
It’s hard to explain via text the effect of the glaring data issues, but hopefully this helpshttps://twitter.com/Corey_Sk8er/status/1325789218489184262 …
Hopefully another example will help clear things up (apologies for not being clear originally at 3am lol)https://twitter.com/cb_miller_/status/1325799039288348672?s=20 …
Maybe this one will help too, but i'm agnostic at this point. It's hard to explain over text. The bottom line is that the author used crap data and tried to basically create his own data with shoddy calculations.https://twitter.com/cb_miller_/status/1325807291518554113?s=20 …
To clear up another point of confusion - there are plenty of data analyses that are technically correct but deeply misleading. This is not one of those - this “analysis” is simply incorrect
And to those trying to dismiss my rebuttal because of my leftist views - there is a difference between facts and opinions/views. I have leftist beliefs, but everything I listed here is factual. As many of you like to say, facts don’t care about your feelings
Sigh, apparently the point is still lost on some. Maybe this will helphttps://twitter.com/cb_miller_/status/1325873103331790848?s=20 …
Get @BretWeinstein to take down his tweets! Or admit he was wronghttps://twitter.com/Corey_Sk8er/status/1326314262659280897 …
Twitter call to arms
ALSO apparently there are other conspiracy theories revolving around votes being suddenly “stolen” from Trump and given to Biden. This is directly related to the Edison source data (apparently this NYT source data is the same data ALL the networks used).https://twitter.com/GWB45/status/1326011402675494912 …
So link the above tweet/this whole thread to wherever you see that pop up about networks “stealing” votes onscreen - it’s because the networks are doing the same thing as the original thread author: backing into/calculating votes on their own.
This is less consequential because the networks were only showing results at a snapshot in time, whereas the thread author was trying to break down by batch. Either way, I’ve seen it going around, so use this thread to debunk it
@BretWeinstein still hasn’t taken his posts down; I’m very disappointed. He has to have seen my debunking by now, he seems to be refusing to admit he was wrong. Fine, whatever, but he’s being a massive megaphone for misinformation, and he needs to stop immediately
Great addition here by @alex_kerchner here. If anyone else has useful stuff to add, just @ me and I’ll probably include it in this threadhttps://twitter.com/alex_kerchner/status/1326329249800785920 …
@brettjrob does a great job of coming to the same conclusion independently, check out this brief thread and commentshttps://twitter.com/brettjrob/status/1326296731609419776 …
Alright this chump tweeted something else out without addressing the misinformation he spread. I’ve always wanted to call for ratio’ing someone, so go ratio this tweet with me 
https://twitter.com/BretWeinstein/status/1326371785709768705 …
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.
