He claims that because we are not dealing with several orders of magnitudes, Benford's law does not apply.
I see the point. Most results are in the hundreds. A few in the 10 and a few in the thousands. That is the precincts were created to all be of similar size.
But what about the second numbers and following?
But Benfords law applies to the second numbers and following digits as well. He mentions that himself.
And then at 14:11 he totally torches himself. Trumps last two digits are in line with Benford's Laws (I like how he acts like this is evidence of Trump's fraud lol) where are Biden's are similar to the random generated Pi digit pairs.
BUT AGAIN. I REPEAT. HE IS ACCUSING TRUMP OF FRAUD!
So why does your claimed source accuse Trump of election fraud? Do you still stand by this turkey? Or is this just part of some disinformation campaign you are running? I'd really like some answers. Do you think Trump committed election fraud? If so, why not just come out and say it. If not, then why are you linking to such bullshit artists?
Benford's law is a statement about the first digit, not the last two digits. At that point he's talking about a different "law" that has an even distribution given certain conditions.
He doesn't actually say Trump committed fraud, it's an act to showcase that an "anomalous" result given a certain "law" is NOT proof of fraud once you actually take a closer look and figure out the "law" is not applicable. I guess you closed the video before he explained why Trump's result is actually completely expected and not proof of any anomaly... which he did like a minute later.
I knew absolutely nothing of Benford's law before this post. I am just telling you he said subsequent digits followed the law as well. If he is wrong about that, then I retract my claim.
I clearly see the point that Benford's does not apply if you have all numbers only between 100-999. As it would make lotteries impossible. Or give premium to betting on numbers between 100-199.
Subsequent digits from the left may follow the law, but that doesn't mean it applies to the last two digits. I can see why you'd intuit that it should be applicable since if you keep going to through the subsequent digits, you eventually cover them all including the last two.
But the key insight you missed is that as the data crosses multiple orders of magnitude, the reading order (left to right) itself starts to matter. The second digit in 2319 is in the hundreds, while the second digit in 12319 is in the thousands. Meanwhile, if you were to look at the last digit instead, it doesn't change as you increase the orders of magnitude. It's always in the ones.
Indeed, the last digit is nine in both cases when looking at it from the right, but if you were to look at it from the left, it's the 4th digit in the first case and the 5th digit in the second case. That's why Benford's law would not be applicable to something that specifies "last digits" rather than "first digits".
I can explain it. It is not complicated.
Benfords law just looks at the first digit of the numbers and then we expect to see a certain distribution of first digit numbers. This can be used in almost all situations of naturally occuring numbers.
The problem is that these are not naturally occuring numbers. These are precincts with around 500-1000 votes each.
Biden has an average of 500 votes in the precincts. This means that most votes will have 4, 5 and 6 as the first digit. This is expected when we know how the data looks. But Benfords law predicts that this shouldn't happen. This is because you can't use Benfords law on this dataset.
The reason Trump looks to be following BL is because his votes in these precincts are around 100, so he gets a lot of 1's, etc. as predicted by BL.
We have 10000s instances of fraud but this BL analysis is just not it. Seriously,
I see you've written words making a claim, but do you have a link which shows this data for every state verifying what you just said and weren't making it up?
Because I've only seen it applied to the swing states so far, so if you have a source which did it for every state, enabling you to make this claim, I'd love to see it.
Benford's law is not a viable tool to review the election.
This dude is a regular on Numberphile, same channel shown in the image.
He claims that because we are not dealing with several orders of magnitudes, Benford's law does not apply.
I see the point. Most results are in the hundreds. A few in the 10 and a few in the thousands. That is the precincts were created to all be of similar size.
But what about the second numbers and following?
But Benfords law applies to the second numbers and following digits as well. He mentions that himself.
And then at 14:11 he totally torches himself. Trumps last two digits are in line with Benford's Laws (I like how he acts like this is evidence of Trump's fraud lol) where are Biden's are similar to the random generated Pi digit pairs.
BUT AGAIN. I REPEAT. HE IS ACCUSING TRUMP OF FRAUD!
So why does your claimed source accuse Trump of election fraud? Do you still stand by this turkey? Or is this just part of some disinformation campaign you are running? I'd really like some answers. Do you think Trump committed election fraud? If so, why not just come out and say it. If not, then why are you linking to such bullshit artists?
Benford's law is a statement about the first digit, not the last two digits. At that point he's talking about a different "law" that has an even distribution given certain conditions.
He doesn't actually say Trump committed fraud, it's an act to showcase that an "anomalous" result given a certain "law" is NOT proof of fraud once you actually take a closer look and figure out the "law" is not applicable. I guess you closed the video before he explained why Trump's result is actually completely expected and not proof of any anomaly... which he did like a minute later.
I knew absolutely nothing of Benford's law before this post. I am just telling you he said subsequent digits followed the law as well. If he is wrong about that, then I retract my claim.
I clearly see the point that Benford's does not apply if you have all numbers only between 100-999. As it would make lotteries impossible. Or give premium to betting on numbers between 100-199.
Subsequent digits from the left may follow the law, but that doesn't mean it applies to the last two digits. I can see why you'd intuit that it should be applicable since if you keep going to through the subsequent digits, you eventually cover them all including the last two.
But the key insight you missed is that as the data crosses multiple orders of magnitude, the reading order (left to right) itself starts to matter. The second digit in 2319 is in the hundreds, while the second digit in 12319 is in the thousands. Meanwhile, if you were to look at the last digit instead, it doesn't change as you increase the orders of magnitude. It's always in the ones.
Indeed, the last digit is nine in both cases when looking at it from the right, but if you were to look at it from the left, it's the 4th digit in the first case and the 5th digit in the second case. That's why Benford's law would not be applicable to something that specifies "last digits" rather than "first digits".
I can explain it. It is not complicated.
Benfords law just looks at the first digit of the numbers and then we expect to see a certain distribution of first digit numbers. This can be used in almost all situations of naturally occuring numbers.
The problem is that these are not naturally occuring numbers. These are precincts with around 500-1000 votes each.
Biden has an average of 500 votes in the precincts. This means that most votes will have 4, 5 and 6 as the first digit. This is expected when we know how the data looks. But Benfords law predicts that this shouldn't happen. This is because you can't use Benfords law on this dataset.
The reason Trump looks to be following BL is because his votes in these precincts are around 100, so he gets a lot of 1's, etc. as predicted by BL.
We have 10000s instances of fraud but this BL analysis is just not it. Seriously,
lol
what's funny about that?
500 is a number. 5 is the first digit of said number.
I see you've written words making a claim, but do you have a link which shows this data for every state verifying what you just said and weren't making it up?
Because I've only seen it applied to the swing states so far, so if you have a source which did it for every state, enabling you to make this claim, I'd love to see it.
You're in luck! This video explains it.