• Quick note - the problem with Youtube videos not embedding on the forum appears to have been fixed, thanks to ZiprHead. If you do still see problems let me know.

Allegations of Fraud in 2020 US Election

Status
Not open for further replies.
Childlike Empress said:
Elaborate. For me it sounds like the person who analysed that data found that a lot of BIDEN voters in Pennsylvania didn't vote for any local politicians (98,000-57,795). Is that what is meant, and backed by data? Or am I misreading something?
Click to expand...


That is what the person concluded. But in the following Tweet, the same person, claiming to use the same statistical methods, concluded that negative numbers of TRUMP voters (in the tens of thousands) in three states didn't vote for any local politicians.

That means that whatever statistical methods he used are invalid, because it is not possible that negative numbers of voters voted for Trump only. Whatever calculation he performed on whatever data cannot be trusted if it yields impossible conclusions.
 
Myriad said:
That is what the person concluded. But in the following Tweet, the same person, claiming to use the same statistical methods, concluded that negative numbers of TRUMP voters (in the tens of thousands) in three states didn't vote for any local politicians.

That means that whatever statistical methods he used are invalid, because it is not possible that negative numbers of voters voted for Trump only. Whatever calculation he performed on whatever data cannot be trusted if it yields impossible conclusions.
Click to expand...

Exactly. Either the source is reliable, and there were a negative number of Trump voters who didn't vote for anyone else, or there were a non-negative number of Trump voters who didn't vote for anyone else which means the source is unreliable.
 
Olmstead said:
The fun thing about this sort of investigation is that it makes just as much sense for the other side until actual evidence is found. It's Schrödinger's evidence. Until you find it, you won't know whether it was Trump or Biden who cheated.
Click to expand...
So we do the same thing as in 2016 and recount and look into it.
 
Myriad said:
That is what the person concluded. But in the following Tweet, the same person, claiming to use the same statistical methods, concluded that negative numbers of TRUMP voters (in the tens of thousands) in three states didn't vote for any local politicians.

That means that [...]
Click to expand...


That means that according to you I understood the data correctly - Base data minus people who did not vote for any local politician. Almost two thirds in Pennsylvania that voted for BIDEN didn't vote for any local politician, according to your data. Source still not disclosed, meaning open.
 
Last edited:
shuttlt said:
Sure, but elections are the type of process that generates results that fit Benford's law... that's why it's used as a test for election fraud.
Click to expand...

Depends on which data set you use. As has been mentioned, precincts tend to be pretty similar in size. That data won't be expected to conform to Benford's law.

I think an important point is that Benford's law deviations tend to suggest human manipulation. Made up numbers in ledgers is the classic example. People aren't good at making up strings of random numbers. If indeed vote tallies violate Benford's law, then what that suggests is that the vote totals were made up by a person. But no one is actually alleging that. What is being alleged is that a certain batch of ballots is illegal or that ballots were being changed. This is a very different thing and I don't think the resultant tallies would be much different from a random distribution. IOW, the numbers wouldn't be made up by a human.
 
shuttlt said:
You still haven't answered the question about what sample size would be needed. This source:
https://www.isaca.org/resources/isa.../2011/understanding-and-applying-benfords-law
claims sample sizes below 500 are where you should maybe not use it. They are an international association of auditors, so they presumably should know what they are talking about. I choose them only because they were the first number I found in a search. If we are talking about samples of 1000+ it seems like the law should apply pretty well.
Click to expand...

Yes, I did.

Sample size is completely and utterly irrelevant. It's not part of the considerations.


To elaborate, it is true that you need enough data to make a distribution meaningful, but that is always the case with every analysis that relies on sampled data. I don't know how much is "enough", but I would guess that the number of precincts in Allegheny County would be enough.


But I wouldn't expect the results from those precincts to follow Benford's Law. There's no reason they should, and sample size is not part of the discussion.
 
Last edited:
Meadmaker said:
2) It's counter-evidence for fraud. A fraudster is highly likely to be partisan, and vote in down-ballot races as well.
Click to expand...
Wouldn't this greatly depend on the type of fraud being committed and the organization/resource/time constraints around committing it?
 
Childlike Empress said:
That means that according to you I understood the data correctly - Base data minus people who did not vote for any local politician. Almost two thirds in Pennsylvania that voted for BIDEN didn't vote for any local politician. Source still not disclosed, meaning open.
Click to expand...


The highlighted calculation performed with correct figures for people who voted for Trump and Biden and did not vote for any local politician, could not yield a negative result for either Presidential candidate. So either that was not the calculation actually performed, or the data was inaccurate, or both.
 
Meadmaker said:
But I wouldn't expect the results from those precincts to follow Bedford's Law. There's no reason they should, and sample size is not part of the discussion.
Click to expand...
Why not. You said something about the precincts being similarly sized being the issue. Why would that impact the last or 2nd to last digits of the votes?
 
Myriad said:
The highlighted calculation performed with correct figures for people who voted for Trump and Biden and did not vote for any local politician, could not yield a negative result for either Presidential candidate. So either that was not the calculation actually performed, or the data was inaccurate, or both.
Click to expand...


LOL. Myriad, this sounds like a bit of "I roll over", just like what Meadmaker just posted. There is no calculation, there are numbers I challenged and you didn't qualify. You gave up before I even was convinced that there was voting fraud. :confused:
 
Last edited:
shuttlt said:
Wouldn't this greatly depend on the type of fraud being committed and the organization/resource/time constraints around committing it?
Click to expand...

Sure, but it's hard to understand what sort of real people in the real world would be the type of people who would be both willing and able to commit fraud that was adequate to elect a president, but not sufficiently interested to care about other races in the same party.


On the other hand, it's pretty easy to imaging a voter who doesn't know anything about elections other than the one about the president. Furthermore, as someone else pointed out earlier, the voter roll-off phenomenon is common and well understood from previous elections. It is not surprising in the least that it would be more prevalent in this election. Donald Trump inspires strong emotions, and there was a massive turnout for this election. I'm fairly confident that massive voter turnout compared to usual elections was inspired by Donald Trump. Some people who usually don't vote came out to vote for him. More people who usually don't vote came out to vote against him.
 
Mader Levap said:
No Other said:
Apply Benford's Law to the vote total and activity by precincts. Since Benford's Law is empirical, it is recognized as valid for detecting fraud in accounting ledgers. Math doesn't care if you're Republican, Democrat or Independent..
Click to expand...
Already debunked.
Click to expand...
When and how?
 
Childlike Empress said:
LOL. Myriad, this sounds like a bit of "I roll over", just like what Meadmaker just posted. There is no calculation, there are numbers I challenged and you didn't qualify. You gave up before I even was convinced that there was voter fraud. :confused:
Click to expand...


It's two Tweets. Five hundred sixty characters max. There are numbers in them. Some of those numbers are negative. The text in the same Tweets attributes meanings to those numbers, which are impossible given that the assigned meanings (quantities of voters) does not logically permit negative quantities. I don't know what else you expect me to clarify or explain.
 
shuttlt said:
Why not. You said something about the precincts being similarly sized being the issue. Why would that impact the last or 2nd to last digits of the votes?
Click to expand...

To understand that, you have to understand why the phenomenon exists at all.

I'm not going to try to explain that, because I don't think I could do a great job. On the other hand, we could just look at the Wikipedia article on Benford's law.

"Distributions that do not span several orders of magnitude will not follow Benford's law."
 
Paul2 said:
Didn't someone say it's not a question of sample size, but of the range of values? If all the values lie between 100 and 1000, that's not big enough. It has to be multiple orders of magnitude.
Click to expand...
They said something like that, but I'm not sure what they are basing the statement on. I don't see in your example why it wouldn't apply to the case you describe, so long as the underlying process was one where the odds decreased as the number increased in the appropriate way. If pages of a log table work here, where we are presumably talking about numbers not wildly greater than the ones you are talking about, I don't see what the objection is.
 
Myriad said:
It's two Tweets. Five hundred sixty characters max. There are numbers in them. Some of those numbers are negative. The text in the same Tweets attributes meanings to those numbers, which are impossible given that the assigned meanings (quantities of voters) does not logically permit negative quantities. I don't know what else you expect me to clarify or explain.
Click to expand...


You do. It's obvious.
 
Dave Rogers said:
They're addressed further down the page. Benford's Law only really works with datasets spanning many orders of magnitude, and doesn't work at all well with the sort of data being looked at.
Click to expand...
Benford's Law is perfect for this type of forensics. All you need to use Benford's Law is that your universe has a range across it of at least an order of magnitude.


Add in the cherry picking effect and the fact that an over-representation of leading 1's is no less a violation of Benford's Law than an under-representation - which the source strongly suggests it doesn't - and it becomes seriously unconvincing. And did you see the bit where the source that supposedly states Benford's Law is useful to detect election fraud actually says it isn't?

Dave
Click to expand...
I find this interesting; what was his basis for making the comment regarding election fraud? Can you provide a link? thx
 
Stacyhs said:
In other words, you'd rather post something you know is likely false than post nothing at all. Fascinating.
Click to expand...
It amazes me that so many people don't mind being wrong. I hate being wrong.
 
Status
Not open for further replies.

Back
Top Bottom