Allegations of Fraud in 2020 US Election

[pgwenthold]

I suggest that any claims of "statistical anomalies" such as violations of Benford's law need to be applied to the 2016 election, too, to verify that it is a unique issue this year.

Didn't there used to be a poster who was claiming that 2016 was fraud because of something like this?

 

[Meadmaker]

Ok, so this says it tends to be more accurate the more orders of magnitude you have... sure. You are claiming that it is too inaccurate based on a sample size ~1000 to be used. The wikipedia article cites examples that seem to be in the low to mid hundreds. What are you basing your claim that it is too inaccurate to use to evaluate these cases on?

It has nothing to do with sample sizes. It's about the distributions of the numbers themselves. If you choose your numbers to be distributed between 100 and 1000, the observed numbers won't follow Benford's law. Precinct sizes are chosen in a way that won't follow Benford's law. They are chosen to be fairly uniform in size and to not span several orders of magnitude.

 

[Minoosh]

I found the third sentence in the Wiki article on Benford's law amusing:

For example, in sets that obey the law, the number 1 appears as the leading significant digit about 30% of the time, while 9 appears as the leading significant digit less than 5% of the time.

IOW "in sets that obey the law, the values obey the law."

Tautologies are fun.

 

[shuttlt]

It has nothing to do with sample sizes. It's about the distributions of the numbers themselves. If you choose your numbers to be distributed between 100 and 1000, the observed numbers won't follow Benford's law. Precinct sizes are chosen in a way that won't follow Benford's law. They are chosen to be fairly uniform in size and to not span several orders of magnitude.

Wouldn't this mean that Benford's law would be nearly useless for analysing election results?

 

[Mader Levap]

Of course he's addressing the second charts, not the first ones..

I like how you pass over fact that someone tried to misrepresent statistical facts.

You'd know that if you understood Benford's Law.

If you understood Benford's law, you would know it is not applicable everywhere. From Wikipedia:

Benford's law tends to apply most accurately to data that span several orders of magnitude. As a rule of thumb, the more orders of magnitude that the data evenly covers, the more accurately Benford's law applies. For instance, one can expect that Benford's law would apply to a list of numbers representing the populations of UK settlements. But if a "settlement" is defined as a village with population between 300 and 999, then Benford's law will not apply.

Considering amount and size of counties, Benford's law is probably completely worthless as fraud detection tool.

 

[pgwenthold]

It has nothing to do with sample sizes. It's about the distributions of the numbers themselves. If you choose your numbers to be distributed between 100 and 1000, the observed numbers won't follow Benford's law. Precinct sizes are chosen in a way that won't follow Benford's law. They are chosen to be fairly uniform in size and to not span several orders of magnitude.

In the other thread, there was a link to someone on Twitter who addressed there. There are academic studies that have explicitly stated that Benford's law should not be applied to voting.

 

[JoeMorgue]

Ah I see one of our primary Trump defenders is now back and refusing to admit that Trump lost fair and square.

 

[Trebuchet]

Neither is this thread, unless you want to put forth an allegation that the popular vote was stolen.

As Trump did four years ago!

 

[Meadmaker]

The Netflix show I saw about Benford's Law is called "Connected". The specific episode is "Digits".

It gives a good overview, but it's a bit sensationalist, and it doesn't go into the properties of the numbers that gives rise to the phenomenon very well. As a result, it, and other references, especially pop culture references, could lead one to the mistaken view that a set of numbers taht doesn't follow Benford's Law is evidence of a flaw, or fraud, in the numbers.

 

[crescent]

Has Orly entered the fray yet?

No, please no.

Right now, this is all fun. Once Orly comes back, it will be crazy fun.

More seriously, there were more than 200 lawsuits filed challenging Obama's place of birth or his citizenship. They never stopped, logic and evidence were meaningless.

This will never end. Biden will get sworn in, but they'll still find ways to file legal challenges. Lawyers will get disbarred for filing frivolous suits, and others will line up to replace them.

Trump has indicated that he'll concede if the legal challenges end without finding anything. But the legal challenges will never stop - too much crazy, too much political and financial support from sane but amoral strategists.

We'll live with this for the next two years at least. American political conservatism now rests on a foundation of conspiracy theories, and the sane conservatives lack the willingness to deal with it.

 

[Meadmaker]

Wouldn't this mean that Benford's law would be nearly useless for analysing election results?

At the very least, it would mean that great caution must be used when applying Benford's Law to election results. I wouldn't want to say that there are no sets of election related numbers that would be expected to follow Benford's Law, but any time I see such a claim, I would want to know how they picked the numbers to study.

I know that claims of Benford's Law as an analysis of election results, such as famously in Iranian elections, are considered controversial.

 

[Norman Alexander]

Anyone here familiar with Benford's Law? It's a mathematical rule governing many sorts of distributions which exhibit scale invariance. Basically, the takehome of the mathematical rule is that for such distributions, the odds that a number in this distribution will begin with a 1 is higher than the odds it will begin with a 2, the odds it begins with a 2 is higher than the odds that it begins with a 3, and so on.

Benford's Law gets used in forensic accounting a lot, because accounting figures normally should follow Benford's Law. But when people artificially generate numbers, they don't tend to generate numbers which follow Benford's Law. So if you take the accounts of a company and look at all the spending items, and it doesn't follow Benford's Law, then that's very suggestive statistical evidence that someone is cooking the books, and you had better look closer.

Election results also generally follow Benford's Law. But according to this source, the results for votes for Biden in a number of key cities... doesn't.

The article doesn't say what numbers were examined. Just "results by county". What "results"? Vote counts? Street numbers? Phone numbers? Heights above sea level? Numbers pulled from some guy's ass?

I suspect it was the latter.

ETA: Not just ninja'd but daimyo'd numerous times.

 

[acbytesla]

Right now, this is all fun. Once Orly comes back, it will be crazy fun.

More seriously, there were more than 200 lawsuits filed challenging Obama's place of birth or his citizenship. They never stopped, logic and evidence were meaningless.

This will never end. Biden will get sworn in, but they'll still find ways to file legal challenges. Lawyers will get disbarred for filing frivolous suits, and others will line up to replace them.

Trump has indicated that he'll concede if the legal challenges end without finding anything. But the legal challenges will never stop - too much crazy, too much political and financial support from sane but amoral strategists.

We'll live with this for the next two years at least. American political conservatism now rests on a foundation of conspiracy theories, and the sane conservatives lack the willingness to deal with it.

This reminds me what former US Attorney Chuck Rosenberg said today on MSNBC.

"There's a huge difference between filing a lawsuit and filing a "meritorious lawsuit. Anyone is capable of filing a lawsuit, but to file a meritorious lawsuit, one in which you could succeed, you need this thing called evidence. And so far there doesn't seem like there is much of that."
https://youtu.be/uyA0DhYnNQI

 

[Minoosh]

Other answers on that page address the Chicago data and it's small size.

So, what kind of fraud are you claiming?

I think the trick is to not make a claim. Get everyone talking about a statistical theory instead. Change the conversation. It's a very effective tactic.

 

[smartcooky]

You have to understand that this is a new kind of conspiracy theory, one unique to our new, wonderful Post-Fact world.

Because the goal is not to actually convince anyone that the election was rigged. It's to just get the idea out there because since there's no intellectual or argumentative standards anymore and everything is "just like, you're opinion man" that's all it takes.

In other words, throw as much feces at the wall as possible, and hope some of it sticks.

 

[Ambrosia]

The Netflix show I saw about Benford's Law is called "Connected". The specific episode is "Digits".

I'd recommend giving the series a watch sometime. I thought it was really good overall and the host Latif Nasser was a good fit.

I *think* its a remake of a decades old show on the BBC called Connections by James Burke.

 

[johnny karate]

It seems to me that he was indicating he strongly suspected it, rather than that he was asserting in an ironclad manner that he knew it to be the case.

Either way, I'm not really in the business of believing or not believing his claims. I find it believable, but I'm not like firmly signing off on it or something.

What specifically is it about these claims that makes you unwilling to commit to believing them?

 

[JoeMorgue]

In other words, throw as much feces at the wall as possible, and hope some of it sticks.

And more of it is sticking because for the last 20 years the internet has screamed down anyone who tried to come up with a way of combating it beyond "LOL they are just trolls, it's apparently a religion that you have to ignore them."

We ignored them and they started voting.

I'm sorry to keep going back to this well but it's what the poisoned water is coming from. We let the internet get turned into a hyper-efficient "min/max intellectual standards/crazy ideas" fountain and at the same time created the "The only way to deal with crazy ideas is to ignore them" fetish and... that IS why we are where we are now.

Because none of this argumentative trolling is new. We've been seeing it in the Woo Slinger circles for my entire life. It's just being applied to politics. That is why we are in a post-fact world right now.

 

[Meadmaker]

Here's a story about the Maricopa County over-vote controversy

https://www.azcentral.com/story/new...ign-over-ballot-rejection-process/6207207002/


Basically, some allegations that poll workers (probably volunteers, if elections are run the same way as in Michigan) gave more instructions than they were supposed to in cases where a ballot was rejected by a machine at the voting site.

I have only briefly looked at it. It would not surprise me at all to find that this has a grain of truth in it. I have tried to write instructions for use by people who have never seen those instructions before and never followed procedures. They always get it wrong. Maybe more later.

It isn't evidence of any fraud, but it might be something for people to look at when trying to make the voting easier. It is conceivable it cost one candidate or the other a few votes. If the final election were to come down to a couple of hundred votes in Arizona, it could be this year's hanging chad.

Maybe.

But the effect could easily be picked up on an audit, to determine if the events even happened at all. If someoone wants to allege that this is evidence of fraud, after reading what happened, we could explore it in more detail

 

[Cain]

We know Biden cheated. We just need to be good skeptics and figure out how.