Forum Replies Created
-
AuthorPosts
-
sdduuuude
ParticipantI’m pleased to see so many people post comments like “I’m not happy with either choice” on this thread.
Seems like this is the time for a 3rd party to emerge.
Participant[quote=spdrun]Whoever picked the axes for the second-digit analysis of Alleghany County needs to have some decency beat into them with a large slide rule. Lying With Graphs 101.
They used a y-axis range of 14 for Trump’s data (making it look more compressed) and a 4.5 range for Biden’s data.[/quote]
OMG. Yes, I see that. There is a formal way to determine conformity to the expected curve. Having that would help immensely. I’m curious to see if that Michigan prof will ever publish an analysis of this election.
Participant[quote=svelte][quote=sdduuuude][quote=svelte] You lost.[/quote]
You seem to have mistaken me for a Trump fan.[/quote]
You talk like a Trump fan.[/quote]
You listen like a deaf child.
Participant[quote=svelte] You lost.[/quote]
You seem to have mistaken me for a Trump fan.
Participant[quote=spdrun]For what it’s worth, SDNative2 was the one who posted the troll about TDS yip-yap-yip-yap, not sdduuuude. sdduuuude appears ready to listen to reason — why Benford’s Law is useless in this instance.[/quote]
The link you posted shows a 2nd digit analysis that fails and 2nd digit analysis doesn’t require multiple orders of magnitude.
Participant[quote=spdrun]This explains it better than I can … basically, it’s an extension of the size issue that I mentioned previously … [/quote]
Very good, spdrun. Very good.
See, now if Rich had just posted that in the Benford’s Law thread I started specifically to discuss that set of data instead of closing the thread down we wouldn’t have to have all these posts in the Biden Wins thread.
I’d love to see the 2nd digit analysis on these numbers. I bet they clean right up.
Participant[quote=spdrun]I was replying to “SDNative2”, not you, unless you have a sockpuppet account.
[/quote]I had to look that up. No, I don’t.
Participant[quote=spdrun]Benford’s Law depends on the initial data set. Let’s say we have a set of states with 6,000,000 voters that tend to vote close to 50/50. Would their results follow Benford’s Law, or would 2 and 3 be the most common (likely only) first digits, without fraud?[/quote]
Yes, but you don’t have that. The data seemed to come in low level within a county. And, as presented, Trump and the third and fourth and even fifth place followed Benford’s Law.
So the race was close and Trump’s numbers followed Benford’s law but Biden’s didn’t, in the data I saw.
After some reading, it seems that using the second digit avoids the problem you mention.
See – it is interesting, ya ?
Participant[quote=spdrun]There isn’t one close state that would flip the election like in 2000. Do you really think that multiple (3-4) states threw their elections as to favor Biden, while no fraud occurred to favor Trump?[/quote]
Certainly not. And I don’t think I presented the information in a way that suggests it.
I put the probability that there is zero material (i.e. changes the outcome) election fraud in the US at 0 and I wouldn’t suggest either party to be more likely the cause of it than the other.
Participant[quote=svelte]Thank you Rich.
I’m honestly quite confused. How did we get here as a society, where we have a significant part of our society buying into conspiracy theories? [/quote]
You mean – how did we get to a point where data and mathematical laws are mistaken for conspiracy theories ? I can’t imagine – but you and Rich need to show more of an interest in math (that’s supposed to be funny and lighten the mood).
I can’t imagine the courts hold up any of Trump’s lawsuits but I can’t imagine he goes away quietly either.
Even Chris Christie gave him a face-palm on his election night speech.
Participant[quote=Rich Toscano]Haha, I know it’s Benford’s Law (not Theory). I’m not denying the mathematical law. I was denying that proves voter fraud in a way that somehow eluded election experts.[/quote]
I thought it would provoke interesting mathematical discussions. My biggest concern would be – did he use the right data or just make it up – which would be voter fraud fraud.
It’s all pretty damn interesting to me. After browsing through some of those papers, it seems that 2BL is a much better test- that is, checking the distributions of the 2nd digit because you get better distribution of the 2nd digit across data that is not spread across multiple orders of magnitude.
I almost didn’t post the conservative commentary link. Shouldn’t have. Should have just posted the data because it is really fascinating.
It doesn’t really prove fraud, but it certainly suggests an anomaly.
Participant[quote=Rich Toscano]I’m shutting down this Benford’s Law nonsense right now, as described here: https://www.piggington.com/benfords_law_and_voter_fraud#comment-292046%5B/quote%5D
It isn’t “Benford’s Theory.”
It is “Benford’s Law”
Dismissing it outright as a conspiracy theory is not right.By the way, here is a syllabus to mebane’s class at Michigan:
http://www-personal.umich.edu/~wmebane/ps485/ps485_syl/ps485_syl.html
And that includes references to several studies, papers and lectures regarding the application of Benford’s Law to elections. Admittedly, I haven’t read them all, but they are every much academic papers as the one you cited:
Pericchi, Luis Raúl and David Torres. 2011. “Quick Anomaly Detection by the Newcomb-Benford Law, with Applications to Electoral Processes Data from the USA, Puerto Rico and Venezuela.” Statistical Science 26 (Nov, 4): 502-516. (in file STS0703-006R4A0.pdf).
Mebane, Walter R., Jr. 2014. “Can Votes Counts’ Digits and Benford’s Law Diagnose Elections?” In Steven J. Miller, The Theory and Applications of Benford’s Law, Princeton UP, 206-216. (in file miller13.pdf).
Mebane, Walter R., Jr. 2013. Election Forensics, chapters 9, 10 and 12. (in files Chapter9.pdf, Chapter10.pdf and Chapter12.pdf).
Mebane. 2007. “Election Forensics: Statistics, Recounts and Fraud,” Presented at the 2007 Annual Meeting of the Midwest Political Science Association, Chicago, IL, April 12-16. http://www.umich.edu/~wmebane/mw07.pdf
Wendy K. Tam Cho and Brian J. Gaines. 2007. “Breaking the (Benford) Law: Statistical Fraud Detection in Campaign Finance.” The American Statistician, 61 (August): 218-223.
Mebane. 2008. “Election Forensics: The Second-digit Benford’s Law Test and Recent American Presidential Elections.” In R. Michael Alvarez, Thad E. Hall and Susan D. Hyde, eds., Election Fraud: Detecting and Deterring Electoral Manipulation. Washington, DC: Brookings Press, 2008, pp. 162-181.
http://www.umich.edu/~wmebane/fraud06.pdfMebane. 2006. “Election Forensics: Vote Counts and Benford’s Law,” Presented at at the 2006 Summer Meeting of the Political Methodology Society, UC-Davis, July 20-22.
http://www.umich.edu/~wmebane/pm06.pdfParticipantAnd if you look even further, you would see that there is a rebuke to that article by Walter Mebane, at the University of Michigan.
““Benford’s Law and the Detection of Election Fraud” raises doubts about whether a test based on the mean of the second significant digit of vote counts equals 4.187 is useful as a test for the occurrence of election fraud. The paper mistakenly associates such a test with Benford’s Law, considers a simulation exercise that has no apparent relevance for any actual election, applies the test to inappropriate levels of aggregation, and ignores existing analysis of recent elections in Russia. If tests based on the second significant digit of precinct-level vote counts are diagnostic of election fraud, the tests need to use expectations that take into account the features of ordinary elections, such as strategic actions. Whether the tests are useful for detecting fraud remains an open question, but approaching this question requires an approach more nuanced and tied to careful analysis of real election data than one sees in the discussed paper.”
I agree with your “luinatic” edit, though.
Since the Michigan guy says it is an “open question” I’ll change my statement from “proof positive” to “worth looking at.”
It’s data. Someone brought data. I don’t think it should be ignored and I don’t think you should have closed the thread.
ParticipantFrom the “Biden Wins” thread:
[quote=svelte] Because, you’d think that if they had fed fake ballots through the system, that would have conformed to Benford’s Law and ended up with a 1 in the first digit most of the time.
[/quote]No, that is not correct. When you feed a random number of fake ballots into the system, it makes it a system that does not conform to Benford’s Law.
Benford’s Law applies to numbers that are counted or accumulated.
The law works because as you count, when you get to a section of numbers beginning with “1” you need a 100% increase to move on to a set of numbers beginning with the number 2. But, when you get to a section of numbers beginning with “9” you only need an 11% increase.
Random numbers don’t have this problem. They can land anywhere. So, if a random number is introduced into the system, the numbers can fall anywhere. You end up with a mix of a Benford set of numbers and random numbers.
Benford’s Law isn’t really ever violated but if you are looking at a system you would expect to follow the law and it doesn’t, it means the system is not what you think it is.
-
AuthorPosts
