Benford Law Election Non-conformity

[Anonymous No.12308592]

https://twitter.com/statsguyphd/status/1324352213595181059?s=21

I am a biologist and don't know much statistics. What does it mean that the p-values are so small, neither even close to .05? In retarded terms, how is this proof of corruption?

[Anonymous No.12308789]

>>12308592
The p-value for chi-squared tests indicate the probability of getting a sample as extreme as the one provided from the given probability distribution. In this case the probability distribution is Benford's Law, so small p-values mean they don't agree with Benford's Law. This indicates the numbers aren't naturally distributed, which can be evidence of some unnatural influence on the results (fraud for example).

[Anonymous No.12310801]

>>12308592
Yea, it's obviously a stolen election. You'd have to be some kind of retard not to see it

[Anonymous No.12310817]

Stupid chuds.
Why are you assuming that all votes are equally distributed random variables? We have a situation where we have two clearlt distinct methods of voting, with different statistical characteristics, you fucking retards.

[Anonymous No.12310849]

>>12308592
does benford work with last numbers of a serie as well as first numbers?

[Anonymous No.12310872]

>>12308592
Glancing over the wikipedia page on Benford's law, it turns out that in the natural world for a lot of situations the first digit of a big number is more likely to be 1 than 9.. This applies regardless of units (see the wikipedia example where they look at the numerical value of the heights of the 58 tallest structures in various units).

However, people are not normally aware of this. Therefore when they artificially create numbers, they are more likely to generate them with a uniform distribution (9s in the first digit are as common as 1s) and so you can detect fraud this way.

I don't pay attention to USA politics so I don't know how this necessarily applies, and to what quantities they are applying the law on. Skimming the wikipedia page, it doesn't seem like elections have the properties that are generally necessary for Benford's law to apply (log random walks and scale invariance).

tl;dr when you make shit up you generally don't follow this pattern in Benford's law, but I don't know if we should expect elections to follow this pattern

[Anonymous No.12310875]

>>12310817
So what is the difference in "statistical characteristics"? Can you elaborate?

[Anonymous No.12310878]

>>12310875
Yes, the probability distribution function of inperson votes and mailin votes are different. And since the pdf is all you need to define the rv, all these stupid models that assume thr same pdf for every vote are wrong by default.

[Anonymous No.12310901]

>>12310878
>the probability distribution function of inperson votes and mailin votes are different.
Any evidence of that, or are you pulling this out your ass?

[Anonymous No.12310908]

>>12310901
Yes, there's widely available and pretty obvious evidence. The probability of a mailin vote being democrat is a lot higher than an inperson one. Historically as well, although it's a more extreme difference this time around for the obvious reasons.

[Anonymous No.12310916]

>>12310908
>The probability of a mailin vote being democrat is a lot higher than an inperson one

They’re also enormously easier to fake, especially in a world where ballots can be “”found”” “”on a mislabeled flash drive”.

[Anonymous No.12310917]

>>12310908
>The probability of a mailin vote being democrat is a lot higher than an inperson one.
thatwill affect the distribution of votes not the DIGITS of number of votes.

[Anonymous No.12311219]

>>12310817
brainlet that has nothing to do with benford- the distribution of first digits

[Anonymous No.12311242]

Please see

>>12310605

[Anonymous No.12311366]

>>12310917
Are you clinically retarded?

[Anonymous No.12311411]

>>12310872
You are paid trump's shill, kys.

[Anonymous No.12311633]

>>12311411
>t. xinnie the pooh

[Anonymous No.12312005]

>>12310908
>>12311366
Retards, read about Benford's law, it has nothing to do with number of votes.

[Anonymous No.12312008]

>>12308592
https://larouchepac.com/20201102/memo-vote-fraud-apparatus-roger-stone-identified-shine-light-them

[Anonymous No.12312010]

>>12310849
No.

[Anonymous No.12312072]

This in no way proves or disproves corruption. The only way to prove or disprove corruption is audio or video evidence. Math isn't the way to go. Unless you have exact data on population and the most recent deaths and there are discrepancies. You shouldn't have to ask. You're part of the reason for all the divisiveness.

[Anonymous No.12312095]

>>12312072
tell that to the irs
what are you a psychology major?

[Anonymous No.12312097]

You lost.

It's over.

You have to go back.

[Anonymous No.12312100]

>>12312097
>reddit spacing

[Anonymous No.12312102]

>>12308592
>>12312097
Dont care, still voting trump

[Anonymous No.12312104]

>>12312100
The reddit spacing meme is from reddit.

You brought it with you all those years ago.

Now its time for you to go back.

[Anonymous No.12312105]

>>12308592
lets take a look:
here is Detroit. The next two posts will be what trump and Biden should have each respectively if the vote was fair given the reported 68.12%biden and 30.65%trump. Look for yourselves.

[Anonymous No.12312107]

>>12312105
Heres what Biden should have been

[Anonymous No.12312109]

>>12312107
And here is Trumps

[Anonymous No.12312111]

>>12312104
lmao you just replied to ">reddit spacing"
you should really lurk for at least 2 yrs before posting

[Anonymous No.12312113]

>>12312095
It even says it needs to investigated. It doesn't PROVE or DISPROVE anything. Why the fuck did psychology or the IRS even cross your mind?

[Anonymous No.12312126]

>>12312113
how stupid are you? who is saying anything about proof? the voice in your head? benford's law is used to detect irregularities. the irs literally uses it. what the fuck does proof have to do with benford's law. imbecile

[Anonymous No.12312139]

>>12312126
Op asked you moron did you read anything?

[Anonymous No.12312149]

>>12312126
And I haven't had any run one with the IRS so thanks I'm learnin

[Anonymous No.12312155]

>>12312149
Run ins

[Anonymous No.12312167]

OP your p-value is very small too

[Anonymous No.12312175]

>>12312139
yeah but i'm not dumb enough to interpret op's use of "proof" literally like you apparently are. they're obviously asking what it proves in terms of what it shows, not whether it's a fricking legal proof

[Anonymous No.12312188]

>>12312149
neither have i but that's like the first thing they teach about benford's law, forensic accounting. if you actually don't know anything about it, then i'm sorry i called you an imbecile

[Anonymous No.12312196]

>>12312175
Sure okay

[Anonymous No.12312209]

>>12312105
>>12312107
>>12312109
If I understand right, what you're saying is when you adjust for the percentage of the vote each candidate got, the graphs (derived from the total) make sense? So Benford's is only applicable to the total number of votes received, and not to individual candidate tallies?

What do you have to make of large districts where all candidates except Biden fit the curve? Do the smaller, 3rd party candidates not matter because of small total vote count?

[Anonymous No.12312252]

>>12312209
Yes. Those need to be accounted for as you can clearly see the distribution shifts based on the underlying distribution of voter turnout and presinct size. given a known underlying value, and average %, we can then create a specific data structure that each should generate both data sets closely. It does with trumps, but not with Bidens. Biden has too many large numbers which would make sense if there was systemic fraud in his favor.

[Anonymous No.12312255]

>>12312209
Not the same guy, but I already posted some analysis on other threads. Reposting here.

Benford doesn't work for the number of votes either in Milwakee

Look at my left graph, "total" line.

In Milwakee, the average number of votes per ward is 960. Biden got about 70% of the votes, that means on average 600-700 votes per ward for Biden so why we see a spike around there.

On the right you have the frequency table of the number of votes per ward.

100 -> 0 to 100 votes
200 -> 100 to 200
300 - > 200 to 300

We can see that still a bunch of them goes over 1000 votes in a ward (average is 960 so this is expected). However very few reaches 2000.

So If I were to win in a ward where the total number of votes is above 1000, I would either get a 1 as a leading digits or a number >= 5.

So a candidate very popular in Milwakee would much likely not get votes in such a way would lead to have 2 or 3. Since this would mean that they got 200 to 399 votes (this would be a loss on average). Of if they got 2000 to 3999 votes. Almost impossible except in very few wards.

[Anonymous No.12312264]

>>12312255
Another anon in another deleted thread asked for the distribution of the leading digits for Biden and Trump in only the wards they won.

None of them follows Benford law.

[Anonymous No.12312279]

>>12308592
Stuff with 0.05 p happens in one in twenty elections. Retards.

[Anonymous No.12312319]

>>12312107
>>12312109
You need to label your x and y axis dumbass or your graph is meaningless

[Anonymous No.12312331]

>>12312107
>>12312109
What do you mean "should have been"?
I have no idea what you're doing here.
Benford has nothing to do with total vote tally...

[Anonymous No.12312384]

>>12312255
True. But wouldn't fraud in terms of adding Biden violating Benford's law also make it so it isn't followed in overall vote count?

Adding numbers of different orders of magnitude would still obey it, but assuming each ward was Biden votes on the same order of magnitude as Trump votes, fraud for Biden votes would shift overall vote counts by regular amounts to violate the curve.

[Anonymous No.12312427]

Who votes what isn't random. How the votes are bundled together isn't random. How the votes are read and registered ISN'T RANDOM. Label your graphs, consider what it is you're trying to say before you say it.

[Anonymous No.12312428]

>https://core.ac.uk/download/pdf/206427437.pdf
>in combination with what we know aboutthe stochastic processes sufficient to yield digits inconformity with Benford’s Law, is that the Law is not universally applicable magic box into which we plug election statistics and out of which comes an assessment of an election’s legitimacy. This is not to say it is fruitless to search for special electoral contexts in which 2BL has some relevance, but our analysis suggests that the data required to validate that relevance must be richer than simple election returns.

[Anonymous No.12312432]

>>12312384
Wards are manmade, I'm guessing they are made to "somewhat" split a population or territory equally. In milwaukee it's an average of 960 votes.

This is also why Benford doesn't apply. It has nothing natural. If you had all the wards in the US and the number of votes range from 1 to 99999 maybe you would see that pattern.

Now, I'm guessing there is some sort of rule of thumbs in the US that wards are defined as being about size X to Y.

Looking at the frequency of total votes, do you think it would make sense that 30% of wards have either 100-199 to 1000-1999 votes.
Currently it's 32% so very close.

But then for leading digits of 2, Benford says 17%, we got 7%. How would you add 10% of wards were the total number of votes is between 200-299 (extremely small wards) and 2000-2999 (very big wards)

Digit 3, Benford says 12%, we got 8%.

Digit 4, Benford says 10%, we got 7%.

For these 2, 3, 4 digits. How would you add leading digits in total votes by remove votes? Are you thinking that 70% of the votes are fraudulent?

[Anonymous No.12312437]

>>12312432
And just to add to the end, if you think that most of the votes are fraud, then you lose a bucnh of leading 1's in the total number of votes from wards with votes >1000. Making them become leading 4, 5 ,6 ,7...

[Anonymous No.12312554]

>>12312255
Your pic actually shows Biden has a very dismilar reporting graphic. His peak is a leading number of barely 1 followed closely by 5, but given what is shown to the right with ward size: His leading number should be one, by more than a factor of 2. Yet again the data shows that further, 4 should have almost twice the number of 6's, yet again that isn't the case. Biden has more larger numbers than the ward size predicts. Oh, and did you take into account turnout rates?

>>12312319
lol, graphs carry their meaning even without labels. You are a tard if you cannot follow that all graphs have the same axis.

[Anonymous No.12312567]

>>12312554
10 This has nothing to do with turnaround rates
2) You're assuming that Biden has 100% of the votes?

Most wards have in between 1000 to 1500 votes, which makes sense that he got a portion of it meaning yes lots of 1's but also 7,8 or 9s.

[Anonymous No.12312579]

>>12312554
I really don't understand how it "should"?

[Anonymous No.12312592]

It's also funny how we only look at a place where Biden won.

What about data from other places? Where Trump had 70% of the votes for example?

Anyone got any data?

[Anonymous No.12312613]

Benford's law just doesn't seem to work at detecting election fraud.

https://www.cambridge.org/core/journals/political-analysis/article/benfords-law-and-the-detection-of-election-fraud/3B1D64E822371C461AF3C61CE91AAF6D

[Anonymous No.12312646]

>>12308592
>Benford's Law is problematic at best as a forensic tool when applied to elections. Looking at simulations designed to model both fair and fraudulent contests as well as data from from elections we know, on the basis of other investigations, were either permeated by fraud or unlikely to have experiences any measure of malfeasance, we find that comformity with and deviations from Benford's Law follow no pattern. It is not simply that the Law occasionally judges a fraudulent election fair or a fair election fraudulent. It's "success rate" either way is essentially equivalent to a toss of the coin, thereby rendering it problematic at best as a forensic tool and wholly misleading at worst
Deckert, J., Myagkov, M., & Ordeshook, P. (2011). Benford's Law and the Detection of Election Fraud. Political Analysis, 19(3), 245-268. doi:10.1093/pan/mpr014

[Anonymous No.12312759]

>>12310916
>They’re also enormously easier to fake
In-person votes aren't difficult to fake. You should look into the piece of shit voting machines the USA uses. They're ridiculously easy to tamper with. The USA needs to overhaul their election rules and upgrade their machines.

[Anonymous No.12312768]

>>12312264
Kek. This should be posted elsewhere. Either Trump cheated as well, or this is a poor way to judge cheating.

[Anonymous No.12312779]

>>12312592
See
>>12312264

I didn't see the data though, I'm gonna verify myself this weekend maybe

[Anonymous No.12312786]

>>12312613
That's not a good paper. Read page 260.
They assume it's bad at detecting fraud because it appears to indicate fraud in elections they otherwise assume to be fair.

[Anonymous No.12312994]

>>12308592
Biden is a fucking fraud

[Anonymous No.12313081]

>>12312255
>ward
Are you trying to say bin?

[Anonymous No.12313203]

>>12312613
This is some random undergrad's essay and it was so bad that a professor had to publish another article debunking it.

[Anonymous No.12313205]

Too many stochastic determining factors in elections to say Benford's Law can detect fraud.
HOWEVER
If you check the rate of DOUBLES.

[Anonymous No.12313208]

>>12308592
You can't "prove" corruption that way

[Anonymous No.12313220]

People ITT are trying to gaslight everyone else into forgetting that the second digit also fails the test, and the second digit is not affected by external factors like district size, turnout, etc.

[Anonymous No.12313228]

>>12313208
What about all those other times it was used by Official Sources™ to prove election fraud?

>http://arxiv.org/abs/0906.2789
>http://fivethirtyeight.com/features/karroubis-unlucky-7s/
>http://thomaslotze.com/iran/index.php
>http://blog.jgc.org/2009/06/benfords-law-and-iranian-election.html
>http://www.theguardian.com/news/datablog/2012/mar/05/russia-putin-voter-fraud-statistics
>https://www.washingtonpost.com/news/monkey-cage/wp/2017/01/11/when-the-russians-fake-their-election-results-they-may-be-giving-us-the-statistical-finger/

[Anonymous No.12313263]

biden voter here, this was just sent to me by a crazed neo nazi. can one of my comrades post the deboonk?

[Anonymous No.12313289]

This thread is filled with thinly veiled /pol/posters.
So what exactly is being examined here? Normally with economic fraud you just look at the first digit of each transaction.

[Anonymous No.12313292]

>>12313228
What I'm saying is, you don't understand the meaning of the word "prove"

[Anonymous No.12313299]

>>12312759
Pen and paper is hardest to fake.

[Anonymous No.12313325]

>>12313228
It's right in your sources
>one possible reading of
A "possible reading" of something =/= proof
"Proof" is a slippery word. It can mean almost anything

[Anonymous No.12313346]

>>12313299
You can even do things like give the ballot a serial number and watermarks, so you can detect fakes etc, a bit like paper/plastic currency.

[Anonymous No.12313354]

>>12313299
Ballots filled in with sharpies are legal

[Anonymous No.12313358]

>>12313354
Then use a pencil, biro or fountain pen if you feel like being a faggot.

[Anonymous No.12313362]

>>12310849
Its uniform. Use your brain.

[Anonymous No.12313376]

>>12313263
I gotcha coomrade

[Anonymous No.12313418]

>>12308592
Hm, convert all numbers to binary and rerun analysis. Then do base 64

[Anonymous No.12313446]

>>12313418
From a theoretical perspective, would this still work under a different number base?

[Anonymous No.12313457]

>>12312613
>just doesn't seem to work
lmao you paid shills are way too easy to detect based on the faggy language you use

[Anonymous No.12313462]

>>12313263
This is what the real image from FiveThirtyEight looks like. Nice try.

[Anonymous No.12313472]

What about last-digit analysis, comparing it to a uniform distribution using the Chi2 goodness of fit test?

I saw that they used something similar in a research paper to detect election fraud in Nigeria: https://pdfs.semanticscholar.org/4389/ab155896e119e59aa231c42b24ab5371b607.pdf

Running the same test on the MI results by county produces low p-values on the Biden votes (p = 0.054), and the Senate Dem votes (p = 0.12 when comparing just the last digit, p = 0.0135 with two). Could this be indicative of anything?

[Anonymous No.12313476]

>>12310817
then why do all candidates' votes conform to it but one's votes do not?

[Anonymous No.12313484]

>>12313476
This is the smoking gun and nobody is actually capable of answering it.

[Anonymous No.12313487]

>>12313446
Yes. Whatever the first number is in your number system, it will occur more often.

[Anonymous No.12313490]

>>12313476
All candidates in every country, in every election around the world in every election that's ever happened (where there hasn't been fraud.

FTFY

[Anonymous No.12313499]

>>12313476
do they consistently? see >>12312264
genuine question cause i haven't seen anyone do the work across states/counties

[Anonymous No.12313510]

>>12313476
Take the Chicago data. There are a little over 2000 precincts, only about 9 of those have at least 1000 votes. Only 2 of those 9 have at least 1000 Biden votes, while none have at least 1000 Trump votes. This matters because the only way for a candidate to have a leading "1" is to have the precinct level votes be in the 10-19 or 100-199 range. Biden's average precinct level total in the data set is about 379. Trump's average precinct level total is 78. Therefore Trump would be more likely to have a leading 1 digit values than what Benford's predicts because his precinct level votes are low. On the opposite end Biden's leading 1 digits would be low because he's more popular in Chicago.

[Anonymous No.12313515]

>>12313510
Should Biden's data look similar for all major cities then?

[Anonymous No.12313516]

>>12313499
I mad this last night based on county data from all the states except Alaska because they're still only at 50%.

[Anonymous No.12313520]

>>12313476
Yeah, but from what I gather that has been the case with US elections since 1992.
Don't hate the player, hate the game

[Anonymous No.12313523]

>>12313520
>from what I gather
lol this has already been debunked by independent fact-checkers

[Anonymous No.12313527]

>>12313515
>>12313523

[Anonymous No.12313529]

>>12313527
Uh-oh, it looks like Biden is in trouble...

[Anonymous No.12313530]

>>12313515
Depends on the size of the precinct, how popular a given candidate is, and how many votes are cast in it. If you have a similar situation as in Chicago, you would expect a similar distribution. Based off of Milwaukee >>12312209
it seems like their district sizes might be bigger.

[Anonymous No.12313538]

>>12313530
Then how would you explain Miami? For some reason, Biden's data adheres to Benford's Law there. See >>12313523

[Anonymous No.12313539]

>>12313523
Meh
Just something I read in the previous /sci/ Benford thread, couldn't be fucked to actually check since I'm EU and really don't care who wins

[Anonymous No.12313563]

What are the assumptions for Benford's Law and do they apply here with the way the data is being analyzed? When it comes to election shit, it's hard to take anything at face value since everyone will try to spin anything they find to fit their agenda.

[Anonymous No.12313565]

>>12313538
The vote totals was split more evenly in Miami-Dade so you would expect a better Benford fit.
Miami-Dade - 53.4% for Biden. 46.1% for Trump
Milwaukee - 69.4% for Biden. 29.4% for Trump
Chicago - 72.9% for Biden. 25.6% for Trump.

[Anonymous No.12313567]

>>12313563
Why is Biden's data anomalous compared to the other candidates in very specific cities?
>>12313523
>>12313515
>>12312209

[Anonymous No.12313583]

>>12313565
Then why don't you see anomalous data in areas where Trump won with a greater majority?

[Anonymous No.12313591]

>>12313583
You mean like in Iowa or Kentucky?
>>12313516

[Anonymous No.12313598]

>>12313591
No, I mean like Oregon, Vermont and New Mexico you fucking retard.

[Anonymous No.12313603]

>>12313516
Where are you getting the county data? I'm running some analysis myself and I've been having to scrape things manually state-by-state, which is a bitch. Is there somewhere I can get it in CSV?

[Anonymous No.12313605]

>>12313598
But you asked for states where he won

[Anonymous No.12313608]

>>12313603
This is some paid shill trying to poison the well of information with misinformation and false data. Benford's Law was what brought down Enron, and it will bring down Joe Biden too.

[Anonymous No.12313612]

magafags reading entrails to cope

[Anonymous No.12313613]

>>12313603
I nabbed it from politico, they've got everything in a nice table instead of just a map but I still had to copy and paste it into a separate excel file.

[Anonymous No.12313622]

>>12313289
The first digit of polling results from wards/districts
There's definitely /pol/ posters in here because this is legit which makes this thread more exciting for /pol/ posters than whatever left-libertarian bloc you think makes up the real 4chan

[Anonymous No.12313638]

>>12313622
>left-libertarian
>4chan left
>american left
>as in today's defintion of it
>as of 2020
>NO

[Anonymous No.12313641]

Bros, does anybody have the raw time series feed that they've been showing on TV? I want to do some analysis on the accumulation of the votes as they came in.

t. data science fag

>>12313638
sew your dick back on, tranny

[Anonymous No.12313643]

>>12313638
where do you think the real non-/pol/ posters lie?

[Anonymous No.12313644]

>>12313641
https://www.bitchute.com/video/MOUqy62fH06b/
This is the only video I can find of the potential Wisconsin voter fraud. It should be noted that all the votes arrived on hard drives.

[Anonymous No.12313648]

>>12313644
What about the actual data itself in tabular format? I want like a CSV or excel sheet or something.

[Anonymous No.12313651]

>>12313648
https://mielections.us/election/results/2020GEN_CENR.html

Best I can do

[Anonymous No.12313653]

>>12313643
Hmofa-libertarian

[Anonymous No.12313654]

>>12313462
lmao

[Anonymous No.12313657]

>>12313651
Hrmm, I would really like to see something with a time axis as the ballots were being counted, because that makes it really, really hard to hide fraud.

[Anonymous No.12313660]

>>12313657
The only thing I can find that has a time axis are the FiveThirtyEight live blogs, but the problem is those are just generalizations. See: >>12313462

[Anonymous No.12313661]

>>12313613
do you still have the link to that table?

[Anonymous No.12313666]

>>12313661
It's on their live results page for each state. As an example:
https://www.politico.com/2020-election/results/alabama/

They don't have it all in one nice table to use.

[Anonymous No.12313667]

>>12313661
I FOUND SOMETHING!

>https://results.elections.virginia.gov/vaelections/2020%20November%20General/Site/Statistics/Absentee.csv

Raw csv for Virginia. This wasn't a sketchy state in terms of fraud, but it will be good to compare Michigan or Wisconsin once that data becomes available.

[Anonymous No.12313677]

>>12313567
The reason I'm asking is because I don't know if Benford's law assumes anything about the underlying distribution. Does this same pattern of going against Benford's law also occur in areas where Trump won big? The reason I'm skeptical is because people are trying to show this in Chicago which an incredibly blue city historically. So trying to call fraud there feels dumb as shit. Has there been similar analysis throughout the years which show every area following Benford's Law?

[Anonymous No.12313692]

>>12313677
>people are trying to show this in Chicago which an incredibly blue city historically. So trying to call fraud there feels dumb as shit
Why? It wasn't for a local election, it was for the presidential election. If anything, being the center of 20th century corrupt Democrat politics, it would be one of the more likely places for fraud.

[Anonymous No.12313703]

>>12313666
Oh you're using county data
Benford's law doesn't apply so well to datasets that small

[Anonymous No.12313709]

For the charts people are posting, what is the underlying data? Is it vote counts per polling location/district?

[Anonymous No.12313727]

>>12313667
63.5% of those supposedly absentee ballots ended up being delivered in person. Now that really activates my almonds.

[Anonymous No.12313783]

>>12313677
> I don't know if Benford's law assumes anything about the underlying distribution
oh it absolutely does. the best case scenario for benford's law is a natural growth process that varies over several orders of magnitude. think something like, "how tall is this plant" or "how fast does this stream flow"

if you're going to apply benford's law to the analysis of a dataset, you really have to demonstrate that there's good reason to think that dataset would follow benford's law in the first place. for example, OP is trying to make the case that "number of people who voted for X candidate" is a dataset that Benford's law applies to. however, the number of votes a candidate gets is not, in the American system, a freely growing value independent of other votes. almost all votes go to one of two candidates in a zero-sum manner. i don't believe benford's law makes any real sense to use in this context

[Anonymous No.12313786]

>>12313783
How fast a stream flows isn't freely growing or independent of other stream flows, and Benford's law I believe should still hold of complementary zero-sum datasets, provided that they complement up to another dataset which follows Benford's such as population size

[Anonymous No.12313790]

>>12313603
The county data can be found here:
https://electionlab.mit.edu/data

[Anonymous No.12313791]

>>12313786
obviously there's few perfect cases, and if you look at enough streams, it's close enough.

as to the second point, it's up to the people posting the graphs to show their work on whether they normalized the data. I doubt they did but i could be wrong. also I haven't put more than a minute or two of thought into it but my gut is that I don't agree that zero-sum dataset comparisons should still fit benford's law.

[Anonymous No.12313800]

>>12313791
What do you mean normalized the data? Pretty sure that would kill Benford
There definitely are zero-sum datasets which do fit Benford, i.e. if some distribution satisfies benford then that distribution divided by 2 also satisfies it, which would be a zero-sum dataset with itself. On the other hand the dataset itself and the constant zero would be zero-sum as well and that definitely fails Benford. Of course the second scenario is less random/typical than the first.
My intuition tells me that the zero-sum aspect doesn't add much since you could choose to look at just one candidate's results without any reference to the other -- why shouldn't just that one candidate's results follow Benford's?

[Anonymous No.12313803]

Does it bother anyone else just how little public data there is on our elections? I work at a retailer as a data scientist and we record everything, and I mean absolutely EVERYTHING. Every time a customer so much as comes into range of one of our wifi networks, we have the exact timestamp, mac address, and location where it happened. Every time each item is scanned at a cash register we have that down to the millisecond, plus 100 other columns of metadata. It's all anonymized too so there's no security risk.

Why aren't our elections public, /sci/?

[Anonymous No.12313805]

>>12313800
>provided that they complement up to another dataset which follows Benford's such as population size
implies a normalization procedure

you're right that a 1/2 a benford distribution should also be benford, but I don't know if that applies to an unevent split process applied across polling locations. is a 3/4s to 1/4 split also benford's?

and ultimately the only real way to determine whether fraud existed with benford's is to find other data that is almost exactly the same data type, but which is completely independent cases of it. that's how the whole enron thing people brought up worked - you can look at other similar companies' balance sheets and compare. there is no real "almost exactly the same but independent" comparison for this election.

[Anonymous No.12313807]

>>12313803
it literally ran by retards they fine off the streets.

[Anonymous No.12313856]

>>12313727
>being delivered in person
You just know it was a hard drive instead of actual physical ballots, too.

[Anonymous No.12313864]

ATTENTION DATA SCIENCE ANONS:
>http://s000.tinyupload.com/index.php?file_id=08686811427934550100
>https://pastebin.com/L7daqrw4
Here is a full dataset of the county level election results. Good luck.

[Anonymous No.12313865]

>>12313651
Which rows correspond with Milwaukee? Did a quick search and couldn't find anything.

Fucked around a bit in R and statewide data doesn't significantly deviate from Memeford's law for either candidate. Seems like dredging or misuse desu but I'd have to read up on it.

[Anonymous No.12313866]

>>12313865
I have no idea but you should probably take a look at this one: >>12313864

[El Arcón No.12313899]

>[math]p=10^{-27}[/math]

[Anonymous No.12313924]

>>12313805
>implies a normalization procedure
I don't follow. What I'm saying is that, for example, population size across cities follows Benford's (according to Wiki), and I see no reason to think why the population size of people who like ketchup across cities would not also follow Benford's. But also, I see no reason why the population size of people who do not like ketchup wouldn't follow Benford's.
There's nothing special about zero-sum. If you have a sub-dataset (data-subset?) then you have a zero-sum situation, that subset and its complement.
Look at the graphs in >>12313644. Veracity aside, assuming they are true, would that be evidence of fraud for you?

[Anonymous No.12313928]

>>12313866
County level isn't going to lend itself well to a Benford analysis since the sample size is mostly less than 100.
Precinct-level is what you want

[Anonymous No.12313940]

>>12313928
Just figure out what counties belong to each precinct and BADA-BING BADA-BOOM

[Anonymous No.12313942]

>>12313865
milwaukee isn't in michigan

[Anonymous No.12314024]

>>12313928
Precincts have a maximum number of possible votes and benford analysis doesn't work on data sets with a maximum or a minimum number.

https://www.isaca.org/resources/isaca-journal/past-issues/2011/understanding-and-applying-benfords-law

>Examples of data sets that are not likely to be suitable for Benford’s Law include:

>Airline passenger counts per plane
>Telephone numbers
>Data sets with 500 or fewer transactions
>Data generated by formulas (e.g., YYMM#### as an insurance policy number)
>Data restricted by a maximum or minimum number (e.g., hourly wage rate)

[Anonymous No.12314031]

>>12314024
>Precincts have a maximum number of possible votes
This isn't true because the number of voters increases with each election

[Anonymous No.12314040]

>>12308592
You guys dig into the down-ballot deviations yet? Off by an order of magnitude. I've only done michigan so far.

QRD:
>>288364734
>>288364734

[Anonymous No.12314045]

>>12314040
http://archive.4plebs.org/pol/thread/288364734
http://archive.4plebs.org/pol/thread/288364734

[Anonymous No.12314049]

>>12313928
i'm the OP of that dataset. it has every county in the US ~3000 i think

i also have a full data dump of all precinct in georiga. i haven't managed to find a nationwide precinct level result yet.

[KYS No.12314053]

>>12308592
>I am a biologist
>I don't know what is a T-test
Eat my dick nigger

[Anonymous No.12314066]

>>12310801

or a leftist piece of trash.

[Anonymous No.12314068]

>>12313376
This isn't what happened though, and is ridiculous when you pull it apart a little.

Trump had a massive, documented lead in Michigan at midnight.

They simultaneously want you to believe that Biden closed that gap when they counted mail-in ballots last, and that he was in the lead from the pretty much 10pm. Even the predictor gauges on Nate Silver's site were saying Trump was likely to win the state.

[Anonymous No.12314069]

>MAGAchuds shitting and screaming on other boards to bail out their failure of a president
Fuck off.

[Anonymous No.12314076]

>>12314069
I just like data man. The fact that its topical data is a coincidence.

[Anonymous No.12314078]

>>12314069

>chuds

[Anonymous No.12314079]

>>12314031
Here's an example from Michigan on precinct size, other states have similar guidelines:
https://www.michigan.gov/documents/sos/XIV_Establishing_Voting_Precincts_and_Polling_Places_266021_7.pdf
PRECINCT SIZE LIMITS AND VOTING STATION MINIMUMS:
Each precinct must not contain more than 2,999 active registered voters. At least one voting station must be provided for every 300 registered voters.

In addition, this data from 2004 shows that each precinct has, on average, 1,100 registered voters at a time (I couldn't find any updated data). https://web.archive.org/web/20061214025307/http://www.eac.gov/election_survey_2004/chapter_table/Chapter13_Polling_Places.htm
While registered voters go up due to population changes, that just increases the number of precincts since they have to be balanced out to equal a certain number of voters in a given area.

From that 2004 data, the average Michigan size of a precinct was 1200 active registered voters (not actual cast ballots mind you, which would be lower). The size matters here, because the only way to get a first digit "1" in those kinds of precincts to have either a low voter turnout for the candidate or an extremely high turnout for the candidate (1,000 - 1200, 100 - 199, 0-10) anything in the middle makes it look weird. So in large cities that most likely vote D, the numbers will look wonky for the Democrat because they're winning the majority of the vote in a given precinct, while R and independent candidates are getting numbers that fall in the Benford distribution.

[Anonymous No.12314082]

>>12314068

the fact that biden won WI is enough proof that this was fraud.

[Anonymous No.12314105]

>>12314079
>the average Michigan size of a precinct was 1200 active registered voters
I know there were some states with insanely high voter turnout rates this year (+90% in areas that are normally around 65%-70%) so that might have something to do with it.

[Anonymous No.12314112]

>>12308592
This isn't really mathematical on the same level, but there is a large gap between votes for Biden and votes for Democrat representatives in the house and senators in the swing states. If these gaps appear in concurence with the violation of Benfords law (late in the counting) this would be further evidence of voter fraud. This also just happened in the swing states

[Anonymous No.12314125]

>>12310801
We know the Democratic primary is illegitimate, but it is suprising to see it extended to the actual presidential election

[Anonymous No.12314132]

>>12314112
In other words, there were a lot of ballots just voting for Biden, and not for Democrats in House/Senate?

Can you look at this by county?

The counties where Biden had huge turnouts should be able to be correlated with these differences. It would be highly odd to find 'only Biden votes' in just a couple counties, for example.

[Anonymous No.12314137]

>>12314132
>>12314112
I saw someone earlier post about the frequency of Michigan split ballots (president and senate votes are different parties on the same ballot). He showed that the split ballots in favor of Biden were an order of magnitude more than the ones for Trump

[Anonymous No.12314144]

>>12314137
I'm suggesting if someone did this by county with erroneously large amounts of Biden only votes, they could focus their analysis on those counties as that is most likely where any fraud took place.

[Anonymous No.12314158]

>>12314112
>>12314125
for me it's bellwether counties
>http://mustreadalaska.com/how-could-the-bellwether-counties-get-it-so-wrong/

[Anonymous No.12314162]

>>12314024
every data set has both a maximum and a minimum number

[Anonymous No.12314169]

>>12314069
there was absolutely fraud done in Biden's favor, and within a month the national conversation will be about why it was necessary

[Anonymous No.12314178]

>>12314162
I'm confused then... how is Benford's Law being used to validate numbers related to the spread of COVID-19 if every country has a maximum and a minimum number of people?

[Anonymous No.12314183]

>>12314178
>how is Benford's Law being used to validate numbers related to the spread of COVID-19 if every country has a maximum and a minimum number of people?
Because it has nothing to do with the maximum and minimum number of people, it has to do with leading digits.

[Anonymous No.12314184]

>>12314183
And it's been used to prove election fraud based on leading digits in the past...

[Anonymous No.12314188]

>>12314184
Benford doesn't work when applied to Biden though, it's been debunked by Snopes

[Anonymous No.12314192]

>>12314178
Because you can apply Benford's law to a dataset with a maximum and minimum. I suppose it's not clear to me what the author means that example, I would think the problem with hourly wage rates is that they don't span many orders of magnitude.

[Anonymous No.12314200]

>>12314192
should be easy to compare it with previous elections and see if it's a strange phenomenon or something totally common?

[Anonymous No.12314205]

>>12314200
It should, and the fact that there aren't a bunch of debunk articles already doing that should tell you something.

[Anonymous No.12314208]

>>12314200
I ran this on some random Michigan data I found for the 2018 republican primary for Governor. I ran it on the second place finisher. p=.000118

[Anonymous No.12314209]

>>12314205
well, to be fair it's started spreading today, maybe they find a real justification, maybe they just ban "Benford" from the internet, let's see how they react

[Anonymous No.12314212]

>>12314192
I think they mean a "hard cap" like the example of the airplane. There are only so many seats in a plane, just as there are only so many voters in a precinct.

[Anonymous No.12314215]

>>12314208
yes, i guess the thing would be running it for many places, for many years, and compare how often discrepancies like the ones displayed by Biden's current election appear in the data, if it's a statistical possibility or if it basically never happened before anywhere

[Anonymous No.12314218]

>>12314209
I think they just don't cover it.
In general this is a good strategy for seeing what has legs. For example there were misleading Wisconsin stats floating around, and there are page after page of google news articles debunking them. Compare that to Benford or downballot gap or bellwether states, nada

[Anonymous No.12314220]

>>12314215
I don't think it applies because precinct data doesn't span enough orders of magnitude for the test to work and you'll just find weird anomalies all over the place when there's probably nothing there.

[Anonymous No.12314222]

>>12314212
But the precincts have different sizes

[Anonymous No.12314227]

>>12314222
They're slightly different but they get adjusted to equal out the number of voters in each precinct. >>12314079

[Anonymous No.12314230]

>>12314220
well, would be good to know how often you get anomalies, so far it seems they somehow only happened in Biden's votes, should be easy to find similar anomalies in places where Trump won, or at least in different years in places where other people won

[Anonymous No.12314236]

>>12314230
The data here >>12314079 shows the average number of precincts per voter was around 1,100. That's just active voter registrations not ballots cast, which means that a 70% turn out rate would only generate numbers between 0 and 770. If you had a bunch of precinct that leans one way, you would be generating more numbers outside of the 100 to 199 threshold thus skewing the distribution.

[Anonymous No.12314241]

So much desperation going on because your orange God lost.

[Anonymous No.12314244]

>>12314236
Precinct data would have to regularly go into into the thousands to generate enough good data to run an analysis on as pointed out here >>12313510

[Anonymous No.12314246]

>>12314236
should be easy to show similar numbers for Trump them? i'm sure they will come up today and debunk this anomaly, it's still early since this numbers come out, so i'm sure we still have time to find why it's normal and not an anomaly at all

[Anonymous No.12314257]

>>12314246
Go back and read what I've got, I've already explained it.

[Anonymous No.12314260]

>>12314241
If it turned out the democrats did cheat, would it affect your opinion of them?

[Anonymous No.12314263]

>>12314236
I mean in any individual precinct this is true, but that's not how this is done

Here is the precinct data for Michigan:

https://gis-michigan.opendata.arcgis.com/datasets/2016-voting-precincts?geometry=26.724%2C43.115%2C-24.252%2C84.203

The average US precinct size is 1100 voters, but that means almost nothing.

>A 2004 survey by the United States Election Assistance Commission reported an average precinct size in the United States of approximately 1,100 registered voters. Kansas had the smallest average precinct size with 437 voters per precinct, while the District of Columbia had the largest average size at 2,704 voters per precinct.

Precinct sizes vary wildly in Michigan

[Anonymous No.12314265]

>>12314263
It has a hard cap of 2,999 for every precinct per Michigan law, the average being around 1,200.

[Anonymous No.12314277]

>>12314265
What's the average size of a precinct in Detroit?

I didn't find anything on that, but I did find these two articles almost immediately looking for it.

https://www.wxyz.com/news/72-of-detroits-absentee-voting-precincts-didnt-match-ballots-cast

https://www.detroitnews.com/story/news/politics/2016/12/12/records-many-votes-detroits-precincts/95363314/

[Anonymous No.12314279]

>>12314263
The actual cap doesn't really matter, what matters is voter turn out, which for Chicago there were only 9 precincts out 2000+ that had more than 1000 votes and 2 of them had Joe votes that totaled more than a thousand. Since Chicago leans heavily democractic, Joe would be getting votes between 199 and 1000 quite frequently making it look like his data is skewed.

[Anonymous No.12314282]

>>12314277
>State audit: No evidence of fraud in Detroit vote
https://www.freep.com/story/news/politics/2017/02/09/detroit-vote-problems-election-audit/97690514/

[Anonymous No.12314284]

>>12314279
The amount of voters definitely matters.

Here are the unofficial Detroit precinct vote totals

https://detroitmi.gov/webapp/election-results

[Anonymous No.12314285]

>>12314279
sounds too convenient that this only happens in Biden's precincts

[Anonymous No.12314288]

>>12314279
dat second digit though >>12313220

[Anonymous No.12314289]

>>12314282
>no fraud here, says fraudster
uhm...

[Anonymous No.12314291]

>>12314284
From just a cursory look at this data, there are an enormous number of precincts with 3 digit vote totals

[Anonymous No.12314292]

>>12313220
this is suspicious

[Anonymous No.12314295]

Best way that i can describe the graph. i hope you catch my drift.

[Anonymous No.12314298]

>>12314295
what about this? >>12313220

[Anonymous No.12314306]

>>12314291
I downloaded the data and put it into an excel file. There are a large number of precincts showing 0% for some reason.

I was quite surprised to see Trump only won 6.5% of the vote in Detroit. Perhaps this is to be expected. Anyway, someone with more expertise in statistics should look at these numbers more closely.

[Anonymous No.12314310]

>>12314306
My mistake, he only won 5.1% of the vote

[Anonymous No.12314319]

>>12314298
Possible a bad graph?
I made one myself for Philadelphia. I used totals for each county.

[Anonymous No.12314330]

everything in this thread has already been deboonked by a tirbinal of independent deboonkers.

[Anonymous No.12314344]

>>12314330
based and snopespilled

[Anonymous No.12314559]

Data for Polk county in Iowa which has the largest city De Moines. Seems like a lot of fraud going on.

[Anonymous No.12314710]

How do the same counties look like if you look at the 2016 election?

[Anonymous No.12314749]

>>12308592
>stats phd
>this code
>didn't set the degrees of freedom on the chi-square
AHAHAHAHAHA

[Anonymous No.12314818]

>>12314749
That code is the real fraud

[Anonymous No.12314832]

>>12313081
https://county.milwaukee.gov/EN/County-Clerk/Off-Nav/Election-Results/Election-Results-Fall-2020

They call it Ward

[Anonymous No.12314837]

>>12312264
sample size for the wards Trump won is smaller by two orders of magnitude.

[Anonymous No.12314846]

>>12314837
Yes

[Anonymous No.12314849]

>>12313376
kek

[Anonymous No.12314865]

>>12314330
>I have become snopes, debunker of chuds

[Anonymous No.12314903]

https://www.texasattorneygeneral.gov/news/releases/limestone-county-social-worker-charged-134-felony-counts-involving-election-fraud#.X6XXF_hkFjc.twitter
>she committed 134 counts of voter fraud
Checkmate MAGAtards, her fake votes followed Benford’s Law

[Anonymous No.12314935]

>>12313515
>>12313523
None of these graphs follow Benford's law except Jojo in Chicago. Trump actually appears to diverge slightly more than Biden in Miami, actually.

[Anonymous No.12314972]

Ususal fund grabbing bullshit data like the phosphine on Venus

[Anonymous No.12315201]

>>12313924
Ah, then I was just misinterpreting the other post, ignore that then.

[Anonymous No.12315207]

>>12314162
and every data set will fail to apply to benford's law if it's near its minimum or maximum. the point isn't that benford's law can never apply, but that in this scenario we're far outside the range where benford's law is a good approximation of what you see in nature

[Anonymous No.12315215]

>>12314749
>>12314818
also that apparently "post a screenshot" was their idea of code distribution and not "link your repo"

[Anonymous No.12315241]

>>12315215
https://github.com/cjph8914/2020_benfords

[Anonymous No.12315287]

could an extreme difference between votes for president but not for senator and viceversa, be used to detect fraud?

I posted this on /pol/ but they are busy with memes

data from here:
https://results.enr.clarityelections.com/GA/105369/web.264614/#/reporting

[Anonymous No.12315289]

>>12313220
>and the second digit is not affected by external factors like district size, turnout, etc.
That's completely wrong, the only reason the digits *would* conform to Benford's law is because of the distribution of magnitudes, determined by district size and turnout.

[Anonymous No.12315307]

>>12315287
>The Democrats made up fake votes for Biden, but not for congressional candidates
Why?

[Anonymous No.12315323]

>>12315307
This is one of the worst talking points.
Suppose that a bunch of Congressional wins too. People would be arguing
>Well obviously the democrats were more popular this year
>So you think they rigged not only the presidential race, but all these house and senate race across the country? Take your meds schizo, voter fraud on that scale is impossible

[Anonymous No.12315331]

>>12315307
Well that's the question. Why do hundreds of thousands of people vote for president but not for anything else?

[Anonymous No.12315335]

>>12315287
>>12315307
I think the weird point is there is a voter precinct of Silver Lake in DeKlaub,GA where:
>people voted for the Rep senator but not for Trump
>people voted for Biden but not for the Dem sen

the coincidence of both is the red alarm

[Anonymous No.12315348]

>>12315335
inb4 "moderate republicans support the senator but not Trump", David Perdue is a Trump ally who got in trouble for Trump-like comments about Kamala last month

[Anonymous No.12315350]

>>12315323
>Suppose this thing that didn't happen, happened
And you criticize others for the strength of their argument?

[Anonymous No.12315357]

>>12315287
>>12315335
also the difference is twice as bigger in the absentee vote than in the vote in person

[Anonymous No.12315358]

>>12315350
Am I wrong?

[Anonymous No.12315403]

>>12313803
Two important points. First is that the government cuts corners on things it thinks are unimportant. Elections is one of those things. Given its importance, it should be rolling in funds. Instead it ends up being an improvised mess run mostly by minimum wage workers.
The second is that the voting data has to be anonymized whereas the data for your employer tries to be as highly connected as possible. If we did not have secret ballots, counting and tracking would be much easier and be mostly, but not entirely, fraud proof. The inability to verify much of anything once you've turned in your ballot makes verification that your vote counted and counted the way you cast it, impossible.
One tempting solution would be to get rid of the secret ballot.
Some groups in the Democratic Party have been arguing for this for years. They already did this for union elections with "card check" rules that require public disclosure of how each person voted. Those who vote against forming a union are often visited by certain persuasive individuals that let them know it was in their best interest to have voted for a union. Because no one wants to receive that visit, open voting leads to voting yes on union formation. Now imagine if political votes were open. The Democrats are already openly talking about re-education camps and reprisals against those who voted for Donald Trump. Open voting would make that possible. Free elections would be gone, replaced with whoever had the best militia in an area.

[Anonymous No.12315470]

is it more clear in a Bland Altmann plot?

the difference of patterns between election day and advanced/absentee votes is pretty big

[Anonymous No.12315494]

>>12315358
You're irrelevant

[Anonymous No.12315535]

>>12315470
>Derwin Brown (June 22, 1954 – December 15, 2000) was an American police captain and the sheriff-elect of DeKalb County, Georgia, who was assassinated on the evening of December 15, 2000, on the orders of defeated rival Sidney Dorsey.
https://en.wikipedia.org/wiki/Derwin_Brown
DeKalb is a corrupt cesspool where third world garbage goes on, such as the above mentioned assassinated of the sheriff-elect by the incumbent sheriff. That they'd be involved in election fraud is to be expected. They usually get away with it because no one outside of DeKalb really cares what happens in DeKalb. Now that they're getting national attention, their corrupt ways are being exposed.

[Anonymous No.12315540]

>>12315494
but am I wrong? two chances so far to tell me whether I'm wrong, neither taken

[Anonymous No.12315559]

>>12313598
Did anyone else hear that?

[Anonymous No.12315589]

>>12313605
No he said "areas"

[Anonymous No.12315620]

>>12312768
>Either Trump cheated as well
it's not Trump cheating or Biden cheating it's the GNC and the DNC and if you think they don't act as independently as possible to avoid catching a RICO case you're a retard.

[Anonymous No.12315660]

>>12314066
>read: retard

[Anonymous No.12315668]

>>12314069
>unironically using chuds as an insult

Rope for you, my dear

[Anonymous No.12315760]

>>12315287
>rig an election
>but only POTUS
>not the legislature
come on, federal-level dems are fucking stupid but not THAT stupid

[Anonymous No.12315770]

>>12315760
You're saying if Dems had won both the POTUS and congressional races, this would be evidence of fraud?

[Anonymous No.12315782]

>>12315770
I'm saying that neither scenario works. If dems DID rig the senate/house races, you wouldn't be able to use them as a reference to detect fraud. if there IS a difference between the executive and legislative races in this analysis, then that suggests a contradiction in competency that is, frankly, ludicrous to me, because that would mean they only rigged the one of the least important parts of the government.

[Anonymous No.12315787]

>>12315760
>>12315470
could this be partially explained by people voting a dem president and an independent senator? or something like that?

[Anonymous No.12315792]

>>12315782
by which i mean - obviously POTUS is important, because it sets non-legislative agenda for most federal agencies and it controls foreign policy, but in terms of setting the arc of the country in stone for the next few decades, the courts and the legislature are far more important than who's sitting in the white house

[Anonymous No.12315798]

>>12315782
>>12315792
The point wasn't to score a big win for democrats, the point was to remove the evil fascist existential threat to democracy

[Anonymous No.12315801]

>>12315787
no, independents barely get votes

[Anonymous No.12315810]

>>12315787
i think the explanation is just "this data is a poor candidate for analysis through Benford's Law"

[Anonymous No.12315840]

>>12315810
how does that explain downballot discrepancies

[Anonymous No.12315950]

>>12315798
>evil fascist existential threat to democracy

Liberals unironically believe this. If it was a fascist existential threat, why does Trump allow elections to happen?

[Anonymous No.12315955]

>>12315798
You don't know what fascism is.

[Anonymous No.12315958]

>>12314837
and it still follows Benford's closer than Biden

[Anonymous No.12315967]

>>12315950
>>12315955
I was saying "evil fascist existential threat to democracy" in character as the left
If there were any year that voter fraud would be pulled, it's this one. Most important election of our lifetime, right?

[Anonymous No.12316028]

>>12315810
but >>12315470 has nothing to do with Benford

[Anonymous No.12316067]

>>12315792
Obama had a woody for getting his people into high positions in the military. As a result, he neglected the courts. There were hundreds of empty seats on the federal courts when Trump got elected. He has filled most of them. Any open right now will be filled by Trump and the Senate before January 20th. That's going to be Trump's legacy. Not just his three picks for the Supreme Court, but the federal judges across the country for seats that opened during his and Obama's administrations. He made a point to appoint young judges so there's little chance this can be undone. Court packing is also going to be impossible unless Democrats win big during the midterms (they won't).

[Anonymous No.12316090]

Can someone smarter than me please explain what's going on in this log-log plot of the vote tallying process?

[Anonymous No.12316157]

>>12316090
The red line is jam, the blue dots are ants.
Conclusion: ants like to jam

[Anonymous No.12316235]

>>12313790
Doesn't have 2020 as of yet. Thank you anyway for the link.

[Anonymous No.12316253]

Wisconsin County is no good since 2016.

[Anonymous No.12316301]

>>12313661
>>12313666

You can steal the data directly from their api.


import requests
state = 4
url = "https://www.politico.com/2020-statewide-results/{:02d}/potus-counties.json".format(sta>
r = requests.get(url)
potus_counties = r.json()


4 is Arizona if I remember correctly. You still have to do some manual work since they haven't labeled the states continuously and Trump and Biden have different candidateIDs in every state, Probably can get that from the api too. I'm too tired and lazy though.

[Anonymous No.12316316]

>>12316253
kek

[Anonymous No.12316321]

>>12316253
>>12316316
>Hmm...

[Anonymous No.12316333]

>>12316067
He didn't neglect the courts by choice, McConnell blocked a shit-ton of his court appointments. But yes, the effect Trump's administration has had on the courts will be felt for decades. The republicans won control of the country for the next 40 years in 2016, and it didnt really matter who won in 2020.

[Anonymous No.12316343]

>>12316028
i have absolutely no experience wtih creating or developing bland-atlman plots, so i cannot even begin to talk about it. that's why i didn't mention it.

[Anonymous No.12316616]

>>12310908
Share of votes has nothing to do with Benford's law.

[Anonymous No.12317329]

>>12314559
Now this one just looks like someone typed in 'random' numbers.

[Anonymous No.12317402]

>>12314169
Post your email so I can laugh at you personally in a month

[Anonymous No.12317404]

>>12315540
Yes you're wrong, now take your meds

[Anonymous No.12317416]

>>12312072
>Math isn't the way to go.

This isnt correct. If its all you have, it can work. Theres been 9 cases of Judges review of elections because widespread fraud. Judges ordered a revote in 7 of them but in the 2 where the Judge declared a winner without a revote, math anomallies were used to show fraud benifited one party and the judge declared victory for the other party. Will link soon. But if I forget, Heritage Foundation report hosted at whitehouse.gov the cases are pg 36 and 69

[Anonymous No.12317420]

>>12308592
It's not really proof of anything, as he doesn't have a big enough sample size.

[Anonymous No.12317430]

>>12317416
Heres the fraud cases, look at pg 69 for instance of Judge declaring somebody a victor after anomalies helped expose fraud. I misremembered the details of case on pg 36. Search “judicial finding” to see the other 7.

https://www.whitehouse.gov/sites/whitehouse.gov/files/docs/pacei-voterfraudcases.pdf

[Anonymous No.12317438]

>>12317420
May make a false postive, but theres a larger question some of us are wondering. Did fraud become widespread in other states/cities? If so it would be hard to trust elections. If theres a positive in one county we can probably ignore it, but if theres 4 or 5, then I imagine the chances of all of them being false positives is low.

[Anonymous No.12317445]

>>12310817
Benfords law works in non random variables midwit.
You are falling into midwit trap.
Try to understand it first

[Anonymous No.12317449]

>>12312255
Dunning–Kruger Level of effert

[Anonymous No.12317451]

>>12316253
>related

Hey guys, rate my amateur analysis. I made this while neck deep in Dunning-Krueger.

Serious question I have: When 100% is a hard limit, do distributions push hard that direction if close?

[Anonymous No.12317457]

>>12317420
Cope.

[Anonymous No.12317548]

>>12317451
Try doing this in the other states that have slow counting counties, with 2016's primary and general along with 2020. Then add random counties from blue and red states to contrast. I suspect you'll find odd trends like this in the former but not the latter.

[Anonymous No.12317550]

>>12314188
O that's right, pack it up boys. Time to go home.

[Anonymous No.12317600]

>>12317451
And why exactly must the underlying distribution be gaussian? There is absolutely no guarantee that a random population is normal and single moded. In fact what you can say is that because a normal distribution has tails that would be non-zero at non-physical turnout values (i.ie >100% and negative values), you don't expect a gaussian.

Secondly if you're going to plot any comparison then use the same limits, and don't just draw an interpolation by hand like a child. Your interpolation is clearly wrong for 2016, you've ignored the bit where it clearly shows you neither population is gaussian.

[Anonymous No.12317628]

[Anonymous No.12317631]

[Anonymous No.12317731]

>>12308592

> What does it mean that the p-values are so small

The p-values stand for compatibility with the probability distribution, as another post already answered.

However, since the rest of the thread has zero stats-trained people, it is sorting out what that means, instead of spotting something much more obvious:

> statsguyphd
> reads p-values like effect size

I guess "unemployedCSneckbeard" was already taken as a Twitter handle.

Rather sad to come back here after all those years and realise that no one, absolutely no one with any intellectual skill is still hanging around. 100% cancer.

[Anonymous No.12317736]

>>12317628
>>12317631
why are tally amounts different? count the amount of votes they won in each county regardless of whether they won or not

[Anonymous No.12317768]

>>12308592
looking for differences in votes between president and senate in GA

I've plotted the Bland Altmann plot of all the voting stations in the 10 biggest counties fof Georgia. There are big differences between Election Day vote and Absentee by mail vote.

>in election day people tend to vote for both the same party president and senate, for both parties
>in absentee vote, people tend to overvote VASTLY for Biden and undervote Trump
>there is one data point in Ponce de Leon deKalb, GA with 1554 absentee votes for Biden and only 876 absentee votes for Ossoff (D). That's 800 votes of difference in just one polling station.

[Anonymous No.12317804]

>>12317768
Not sure about the Ponce precinct lines but that area is where young hipster neighborhoods that have gentrified butt up against a massive set of Section 8 housing developments believed to be the largest cluster of Section 8 in the entire country. These are probably the two groups most likely to be motivated to engage in voting fraud but good luck actually proving it.
The real problem with the statistical analysis is that it's not actionable. Even if we know there was fraud, there's not much that can be done about it. I've never heard of a court throwing out an entire precinct's votes and ordering a do-over and without the ability to pinpoint specific ballots that are fraudulent, there's no way to throw those specific ballots out. This might end up being the perfect crime, where we know it happened but cannot do anything about it.

[Anonymous No.12317861]

>>12317804
that's the point
you cant auction the entire election, but you can auction the points where the analysis guides you

[Anonymous No.12317922]

>>12317768
another graph

any suggestions?

[Anonymous No.12317946]

>>12313567
You know exactly why.

[Anonymous No.12317967]

>>12315403
The amount of political push to try and eliminate secret ballots is, frankly, terrifying

[Anonymous No.12318063]

>>12317804
I checked the actual precinct boundaries and my earlier post was incorrect. The area I was thinking of is on the Fulton County side of the Ponce area. The DeKalb area is upper middle class/lower upper class, mostly white households earning $200,000+. These are urban white liberals so them voting for Biden isn't too much of a surprise. It is however very much out of character for this type of group to ignore down ballot races. They tend to be straight party voters for Democrats.
Biden was going to win this precinct no matter what but it is extremely odd that Ossoff didn't get the votes from this group. Perdue is a rural redneck and the cousin of Trump's Secretary of Agriculture, Sonny Perdue (former governor of Georgia). It makes no sense that Ossoff didn't get the Biden vote. Very strange though I guess one could argue this is as much evidence of Perdue cheating as it is of Biden cheating.
Pic is a typical house for the area.

[Anonymous No.12318068]

>>12317967
That might be the real purpose behind making the cheating this time so obvious. Cheating always happens but usually it is kept to a small enough scale that it isn't noticeable and only make a difference in a few local races. This time they're almost shouting it from the mountain tops. The long term plan might be to goad conservatives into demanding a way of tracking the votes of individuals, at which point ideological corrections mobs can be sent to the homes of those who vote incorrectly.

[Anonymous No.12318083]

>>12317922
sorry I've got the senate data from DeKalb wrong
dont use this one I've got to correct it

[Anonymous No.12318088]

>>12317922
>>12318083
the corrected one doesnt have the Ponce outlier, fuck
still suspicious tho

[Anonymous No.12318094]

>>12308592
Has this been debunked yet?

[Anonymous No.12318108]

>>12318094
republicans prefering a more conservative leader - biden.

[Anonymous No.12318122]

>>12317922
>>12318088
corrected with color
Ponce was a fake outlier
I hate when this happens

[Anonymous No.12318127]

>>12308592
biologists should know statistics...

[Anonymous No.12318132]

>>12318094
Jon Ossoff is a pencil neck geek. He's the type who would have been a hall monitor in middle school and enjoyed snitching on everyone. In an earlier losing campaign he declared his love for San Francisco.
The numbers are odd but maybe there are people who don't want Trump but also don't want a male feminist dork as their senator. Or if there was cheating, yes, this would be a way to force a runoff for the Senate. The Democrats here are pushing hard for 17 year olds who will turn 18 before the runoff election in January to register to vote. Those votes will almost all go to Ossoff, the Democrat.
I suspect we'll never know what really happened in this election. There's simply little tracking that can be used as definitive proof unless someone comes forward with video evidence.

[Anonymous No.12318135]

>>12318122
That makes more sense. No way people in households making $200k or more would forget to complete their entire ballot.

[Anonymous No.12318156]

>>12318135
still 173 voting precincts two std dev above with absentee votes, that's over 12k suspicious votes

and the next one on the list is Smyrna 7A with 151 votes of difference between President and Senate
Smyrna is a rich town according to wiki

[Anonymous No.12318209]

Reminder

[Anonymous No.12318234]

>>12318122
now with annoying data labels

[Anonymous No.12318306]

>>12318156
Smyrna used to be lower middle class whites but in recent years spill over from the trendy upper middle class neighborhood called Vinings has made Smyrna increasingly prosperous. Both are known for having large populations of single childless white career women. They're exactly the type who campaign for people like Jon Ossoff and mayor Buttboi so it wouldn't be likely they'd not vote for Ossoff.

[Anonymous No.12318312]

>>12318234
Very interesting. Most of the outliers are areas with more single white women than the rest of the metro. The further out from Atlanta, the more likely they are to be divorced with children while closer in they tend to be childless and never married.

[Anonymous No.12318592]

>>12318108
>>12318132
https://www.pewresearch.org/politics/2020/10/21/large-shares-of-voters-plan-to-vote-a-straight-party-ticket-for-president-senate-and-house/

[Anonymous No.12318652]

>>12310817
Brainlet back to /pol you go
Shill elsewhere

[Anonymous No.12318680]

>>12315950
Realistically this

[Anonymous No.12318720]

>>12318234
>>12318122
>>12318088
>>12317922
>>12317768

Can you explain how to read these graphs?

[Anonymous No.12318886]

Looking through everything posted here, there really doesn't seem to be anything unusual from a statistics/data science perspective. It all looks legitimate, or based on misapplication of certain tests. Is there anything in particular the rest of you find more convincing?

[Anonymous No.12319115]

>>12318886
>Why would anyone lie on the internet?

[Anonymous No.12319200]

>>12310908

>I heard something on the news just believed it, so I don't actually have any evidence DURRRRRRR

(SAFIRE has nothing to do with this thread I just hate them and use this image)