Claim: Election Truth Alliance "proves" voter fraud in Clark County, NV

[Peter Delaney]

Skeptic and stats guy here (psychologist, but I do statistics a lot). Recently my fellow left-wingers have been falling into election denialism and a commonly-cited piece of evidence showing up on reddit is the Election Truth Alliance's conclusion that there was electronic manipulation of voting machines in Clark County, NV. Their main argument is that only for early voting do you see "more stability" in the vote totals and not just random scatter.

I am super sad that Trump won, but their data show a common stats misunderstanding, and no evidence of vote manipulation that I can see. Specifically, this is the "law of small numbers" fallacy. If you measure something more times, your estimate of the true value gets more precise. That's why raising the sample size in studies makes them more trustworthy: you measure the average score more accurately when you measure it more times.

In the Election Truth Alliance Figures, watch the horizontal axis. On election day, the axis is stretched out but there are no machines getting lots of votes (because it's only one day). On the early voting one, the axis includes lots of machines that averaged a ton of votes (because it is adding up over many days when people could vote there). As the number of votes go up, you would expect them to look more like the real voting pattern. This is the same bad math the right wing used and it is hot garbage again.

Kahneman and Tversky in a famous paper asked people to guess where it would be more likely you'd get 60% boys born on a given day: a hospital that has 10 births per day or one that has 100 births per day. The answer is the small hospital, because you get more variability with a small number of births. This is super-counterintuitive to most people so they get it wrong unless they know statistics. You can make it more obvious if you make it more extreme: where is it more likely to get all boys, a hospital with one birth or one with 100? Obviously it's 1, because you flip the coin once and get tails a lot. With 100, you have to flip the coin 100 times in a row and get all tails. That almost never happens.

Same with the Nevada votes. More votes, more accurate measurement of the 60%:40% ratio that people in that county had. (Here, it's not a "fair coin flip" because there are more Trump voters than Harris voters there). Even people who know a fair bit of statistics still make the law of small numbers error, but people who are very good at it shouldn't.

Feel free to share.

 

[yoshy]

Welcome to the website, and nice work! It's always good to get more statistics enjoyers here.

This is the same bad math the right wing used and it is hot garbage again.

This is exactly what I thought of as well. Conspiracy thinking is bipartisan unfortunately (though I am not saying it's in equal amounts!).

 

[JMartJr]

This is exactly what I thought of as well. Conspiracy thinking is bipartisan unfortunately (though I am not saying it's in equal amounts!).

The thing that scares me is that it feels very natural to respond to the excesses of one's opponents with excesses of your own -- and then things can snowball in a series of one-upmanships and we're off to the races. My fellow conservatives having largely fallen into the rabbit hole for the moment, it would be helpful if our friends across the aisle could resist to urge to follow.

 

[yoshy]

My fellow conservatives having largely fallen into the rabbit hole for the moment, it would be helpful if our friends across the aisle could resist to urge to follow.

I think the big difference here is that the Democratic establishment isn't pushing these absurd claims like the Republican party did (e.g. dozens of court cases, bizarre conferences like the pillow man's, and, oh ya, that whole insurrection thing). The conspiracy thinking remains on the fringes of the party for now.

 

[Ann K]

Anecdotally, I was in Nevada in December 2000, when weeks after the election we still didn't have a final decision on who won. There was a discussion in the papers about the integrity of voting machines, and the consensus was that if they wanted to know how to safeguard voting machines, they should ask the people in Nevada who manage to safeguard millions of slot machines from tampering.

 

[Peter Delaney]

Welcome to the website, and nice work! It's always good to get more statistics enjoyers here.

This is exactly what I thought of as well. Conspiracy thinking is bipartisan unfortunately (though I am not saying it's in equal amounts!).

Thanks! I arrived here during the David Grusch/UFO stuff, which seems to have gone even weirder in recent days with no solid new evidence appearing to support his major claim(s).

Statistics is the basis of so many weird misunderstandings, because it's so often counterintuitive. Even good scientists make some intuitive errors now and then. It falls to people who understand it to try to offer a hand up from that rabbit hole to well-intentioned conspiracists. After all, it's math in this case! You can literally prove it! As long as you are dealing with people who believe that evidence is useful, you still have hope of swinging them back from the brink.

 

[JMartJr]

I think the big difference here is that the Democratic establishment isn't pushing these absurd claims like the Republican party did (e.g. dozens of court cases, bizarre conferences like the pillow man's, and, oh ya, that whole insurrection thing). The conspiracy thinking remains on the fringes of the party for now.

Indeed. My internal emphasis is on the "for now," with the hope that "now" extends indefinitely, and the GOP or its successor party decides to move back towards some sort of reality-based approach to politics. Hope is free...

 

[JMartJr]

Anecdotally, I was in Nevada in December 2000, when weeks after the election we still didn't have a final decision on who won. There was a discussion in the papers about the integrity of voting machines, and the consensus was that if they wanted to know how to safeguard voting machines, they should ask the people in Nevada who manage to safeguard millions of slot machines from tampering.

Delightfully ironic in that the house rigs slot machines most carefully to make sure that the house wins.

 

[FatPhil]

Thanks! I arrived here during the David Grusch/UFO stuff, which seems to have gone even weirder in recent days with no solid new evidence appearing to support his major claim(s).

Statistics is the basis of so many weird misunderstandings, because it's so often counterintuitive. Even good scientists make some intuitive errors now and then. It falls to people who understand it to try to offer a hand up from that rabbit hole to well-intentioned conspiracists. After all, it's math in this case! You can literally prove it! As long as you are dealing with people who believe that evidence is useful, you still have hope of swinging them back from the brink.

Another welcome to you, good sir. What might loosely be described as "statistics" is a very fruitful ground for misunderstandings. My absolute personal favourite by a country mile is Simpson's paradox (whereby, for example, despite hospital A having better in-patient care than hospital B, and also having better out-patient care, the combined data tells you that hospital B gives better care - https://en.wikipedia.org/wiki/Simpson's_paradox ), because you have to be really on your toes to recognise when it might be happening, and it might be happening deliberately for nefarious reasons. Some are quite shallow, such as the elevator paradox (the direction you more often encounter an elevator moving in depends on what floor you're on - simple-ish, as you're not sampling without bias - https://en.wikipedia.org/wiki/Elevator_paradox ), some require a bit more of a perspective shift to understand, like the friendship paradox (your friends typically have more friends than you do - because you're less likely to know the ones with few friends, and the average can be boosted by just a few highly gregarious friends - https://en.wikipedia.org/wiki/Friendship_paradox ). Of course, as someone interested in game theory, I can still get some entertainment out of other people discussing things as trivial as the Monty Hall problem (it's about evaluating probability, so I'd say it's in the field as defined loosely), as popcorn's cheap.

 

[Ann K]

Delightfully ironic in that the house rigs slot machines most carefully to make sure that the house wins.

Not the house. The state. But outside would-be fiddlers don't stand much of a chance.

 

[Kyle Ferriter]

Reminds me of the couple weeks after the US 2020 Presidential election where a ton of conspiracy influencers who'd never heard of "Benford's Law" before then were totally convinced that it proved statistically that Biden's votes in some counties were fraudulent, even though they didn't really understand what the prerequisites for that "law" are or when it is applicable or what it can "prove". They were reaching for whatever justification they could find to back up their existing beliefs even if it didn't really make sense when you look a little deeper.

Thankfully I think the pushback against that misinfo was enough that it never recirculated again after that, like some other bits of misinfo have.


Source: https://youtu.be/etx0k1nLn78

 

[goodgravy]

More votes, more accurate measurement of the 60%:40% ratio that people in that county had.

The official Clark County totals are 50.4% Harris and 47.8% Trump. Not ~40% Harris and ~60% Trump.

If you measure something more times, your estimate of the true value gets more precise

Agreed. So for Early Voting, as the "Total Votes per Machine" (x-axis) gets larger, you'd expect Trump to get closer to 47.8%. Instead he gets closer to 60%.

Edits: replaced the huge quote block with two smaller quote blocks. So many blocks.

 

[goodgravy]

Following-up on my previous comment...

ETA hypothesizes that Early Voting (in Clark County, NV) was manipulated, while Mail Voting and Election Day Voting yielded data that looks more natural, and less manipulated. With that in mind:

The attached image shows that Mail Voting in Clark County was ~62% Harris and ~37% Trump. And as the clustering in your image shows, Election Day Voting seems to be ~50% Trump. I believe Lefties favor Mail while Righties show-up on Election Day (but maybe that changed).

Regardless, ETA is asking: "Why does Early Voting show compelling signs of artificial manipulation?" And they believe they have the Evidence to file lawsuits for a full audit. I think it's called a forensic audit (or something like that).

 

[FatPhil]

I believe Lefties favor Mail while Righties show-up on Election Day (but maybe that changed).

It's not hard to believe that, given what Trump was saying about early voting for many years. (I think I read that he did start to tone down his rhetoric in the latter months of the recent campaign, though, but that's only the vaguest of memories.)

Troubling signs for the GOP on mail voting

Republican leaders have pushed for their party to embrace early and mail voting, despite Trump's attacks on it. Their voters appear to be taking their cues from the former president.
September 25, 2024
...

As early and mail voting have begun in some states, new polls show Republicans still aren't planning to embrace these methods. In fact, the partisan gaps look a lot like they did in both 2020 and 2022.

A Quinnipiac University poll released Tuesday showed just 14 percent of Republicans planned to vote by mail or absentee ballot, compared with 34 percent of Democrats.

And surveys from NBC News and YouGov this month each showed 17 percent of Republicans planned to vote by mail, compared with 34 percent and 43 percent of Democrats, respectively.

All three polls show at least a 2-to-1 edge for Democrats in intent to vote by mail. Notably, the gaps (between 17 and 26 points) are similar to what we saw in 2020 and 2022 among voters who cast ballots for each party (26 points and 18 points, respectively, according to data from the Pew Research Center).

-- https://www.washingtonpost.com/politics/2024/09/25/troubling-signs-gop-mail-voting/

 

[Ann K]

Regardless, ETA is asking: "Why does Early Voting show compelling signs of artificial manipulation?" And they believe they have the Evidence to file lawsuits for a full audit. I think it's called a forensic audit (or something like that).

In many (perhaps most) states, mail-in votes are received before Election Day but counted AFTERWARDS. That has caused many people, mostly republicans, to mistrust mailed votes - "Whaddya mean, my candidate had lots more votes and now it's changing?" This is the opposite of the electronic voting suspicions.

 

[JMartJr]

In many (perhaps most) states, mail-in votes are received before Election Day but counted AFTERWARDS. That has caused many people, mostly republicans, to mistrust mailed votes - "Whaddya mean, my candidate had lots more votes and now it's changing?" This is the opposite of the electronic voting suspicions.

Note that in some states (NC for one) they per-process the mail in votes and they can then be counted as soon as the polls close, and are the FIRST votes reported. So we get a "Blue Mirage," where the Democrat candidate seems stronger than they actually are going to be when day-of voting is counted and the final results are known. There are several reasons why this does not lead to such strong reactions from conspiracy mongers, the one I want to mention is that the issue resolves on election night or the next day as day-of voting is counted normally, and so it does not tend to lead to a long, drawn out ordeal as the illusory lead evaporates over some days or weeks, as can happen with late-processed mail votes.