The claim is that you would always find the negative slope Dr. Shiva highlights (to show that this is very likely the result of a programmed algorithm) due to mathematical constraints. While the constraints are true at certain levels, the assertion that these constraints necessarily lead to data always showing a negative slope is wrong. As an example, take the hypothetical case where individual votes are made in the same proportion as straight party ticket votes in the different districts in the county at different percentages (i.e. some districts lean more republican or more democrat by varying amounts and individual votes are cast in those same proportions). In this case you would see a straight horizontal line crossing the y-axis (Trump - Straight Republican Party) at zero.
Edit: adjusted the hypothetical case thanks to a comment from a more informed Pede. The result is the same in that there are scenarios that can produce a straight horizontal line at zero.
Edit: apologize for potentially calling pedes shills undeservedly. We should have a discussion to get to the truth.
The claim is that you would always find the negative slope Dr. Shiva highlights (to show that this is very likely the result of a programmed algorithm) due to mathematical constraints. While the constraints are true at certain levels, the assertion that these constraints necessarily lead to data always showing a negative slope is wrong. As an example, take the hypothetical case where individual votes are made in the same proportion as straight party ticket votes in the different districts in the county at different percentages (i.e. some districts lean more republican or more democrat by varying amounts and individual votes are cast in those same proportions). In this case you would see a straight horizontal line crossing the y-axis (Trump - Straight Republican Party) at zero.
Edit: adjusted the hypothetical case thanks to a comment from a more informed Pede. The result is the same in that there are scenarios that can produce a straight horizontal line at zero.
Edit: apologize for potentially calling pedes shills undeservedly. We should have a discussion to get to the truth.
Here is what I'd personally like to see: Do the same analysis for a different election cycle, in a different state, in a place which doesn't use the same vote counting machines. Or maybe even in a hand-counted race.
This assumes that Trump doesn't aIready know that they rigged the machines to siphon votes, which they may.
They did a hand count of one of shiva’s districts he ran in and the slope disappeared. It’s only machine counted. Manual recounts would show no such trend, and it would be impossible for them to get an increasing percentage of illegal ballots into the system with such a trend line in multiple counties , it had to be in the tabulators.
Guess you guys missed the part that it is all election machines since 2001
The trend line showed us where to look. I hope to God they are able to show hard evidence that it was programmed this way. Hopefully the whistleblowers have this hard evidence, or a recount can definitively prove it.
They did a hand count version. No slope. Just didn’t release a graph for it
Shiva is a fraud lol... He markets himself as being the "inventor of Email" when it was Ray Tomlinson who actually invented email back in 1971 People who believes this garbage coming from a fraud like Shiva is beyond me.
You can call me a shill or whatever, but it will only take you few minutes to verify his claim is false.
Deboooonked!! Im deboooooking!
I think its correct..but it would be hard to overturn an election with data and statistics.
Impossible is the word you're looking for.....
Those things can only be supporting evidence, but by themselves don't mean a fucking thing in court.
Uh, there have been convictions on nothing but statistical analysis. Yeah they do mean a thing in court.
Trust me, it won't be enough when a Presidential Election is on the line.
Who ever said this is the only evidence we need, or want?
Agreed
They do give you a hint of where to look.
It's actually quite easy. Find the voters, ask them to testify under oath who they voted for. You'll see rather quickly whether the count is accurate. In a truly random sample (trivial to generate thanks to the voted list) you'd only need 100 people to tell if the count us accuratr within 5% or so. If there's a greater variance than that you have scientific proof.
With a population of 10,000,000 (MI, for example) you'd need a sample of about 385 to be statistically significant across a 95% confidence interval with a 5% margin of error.
just have to survey one or two of the precincts that show anomalous results (ones that have high percentage of straight rep ticket voters). Those precincts are small and only have a couple hundred people in each one.
That's a biased sample. You'd need a much higher sample if you're trying to get a statistically significant sample of each district.
Found the stats guy. I studied physics. 385 = 100.
At the right granularity, you'd be correct!
The statistics are the blood hound, the sticks pointed into the leaf piles are the direct evidence of voter fraud.
Statistics will identify where voter fraud most likely occurred, auditing and canvassing those precincts/counties will verify it.
Note: every legal voter examined must verify they voted at that specific precinct otherwise they could've counted that ballot at many different precincts. So an authentic ballot and voter doesn't necessarily mean authentic vote.
Enough to subpoena the source code of those particular machines under audit and plaintiff expert.
Also enough to make the case, there is no way to audit this result, so it can't be trusted.
The software can also not be trusted....
It is very easy to have code sit on top of other code that deletes itself, after tamping with the outputs of the underlying code.
I actually don't think they're all shills. Accepting something without seeing "the other side" is just bad, and what we have accused the dems of for four years (because it's true). Critical thinking and doubting what we see is a trait that got us to where we are today, otherwise we'd all be mindless NPCs, just like the other side.
I'd personally like to see the numbers ran for Joe too, but so far, nobody has come up with that, I've just heard people screeching about shills. I was initially astounded by what Dr. Shiva showed us, and was 100% bought in. Now I'm skeptical, as I'd like to see more evidence. If i had the capability to do this myself, I would... but I wouldn't know where to begin.
I was trigger happy with the word Shills
What looks fishy about them?
It's not that they look wrong, it's just that I'd always assumed when it comes to maths that you want to see many more examples of what is normal vs what is an anomaly. Surely there's more data out there to reinforce Dr Shiva's argument?
I come from a cyber security background, so anomaly detection works on what we call "baselining"... whereby we have a strong baseline of what is "normal" or expected behaviour, and then anomalies become easier to spot (they don't match the pattern of expected behaviour).
I'd also like to see the same concept but ran as if we were trying to show that votes were stolen from Biden. If it isn't the case, then the slope should not occur for any of his data.
Basically, more maths and more data!
I'm sure it takes time. One of the guys in the team said he had been stuck in a basement for a week putting the numbers in.
I didn't catch that, thanks for info. Here's to hoping there's more coming then
It was the republican that was connected through the phone. Said it towards the end.
This. It may prove something. But he didn't show that it does unless you see more examples of normal. The rough Biden slope is also negative, but higher on the graph.
E.g.. https://m.imgur.com/a/DQsEHdq
Not quite what you're looking for, but someone plotted biden's results on the same graph:
https://thedonald.win/p/11Q8SoArTE/4chan-anon-confirms-dr-shiva-ayy/c/
Isn’t that just the mirrored image of Trump’s results added for visual effect? Look at the lone outlying dots—there’s a corresponding mirrored dot for each.
I thought the same thing too, but they are not a mirror image. Look closely. There is a slight difference, which is probably due to 3rd party votes.
So these do appear to be biden votes plotted, rather than trump's votes mirror imaged.
Seriously, you expect me or others to do the dems job?
When you see it SHUT UP.
https://m.imgur.com/a/DQsEHdq
Here is a rough Biden slope vs a rough Trump slope. Negative curve also but much higher on the graph.
The graphs I saw had a problem not explained by that. It might explain some of the behaviour at the edges but they stay high around 20% to a third in and then nose dive at a far steeper angle than the domain of freedom and are also tightly clustered to that line rather than the full range of freedom.
That's a pattern that looks very biased and systematic. It does not look random according to any natural axis available.
At maximum variance it should be all over the domain of freedom which it isn't. It's not symmetrical. Typically speaking if there's a pattern then it should be at an angle somewhere between the central axis and the domain of freedom and it should be more uniform rather than having a crook. All the graphs they looked at had a crook then a nose dive at an angle of attack too steep and consistent.
What we're really missing are graphs for all regions and showing the D line at the bottom. That would reveal if there's some common underlying influence as you would expect to probably see this pattern on both sides.
If you don't see that then you have to ask why only democrats?
There could be other explanations for the crook and slide but we need to see the data and formulas.
Concur. Signal looks suspect after it goes from an apparently natural noisy spread into a nicely conditioned linear response with no outliers. Through some convos I've been able to dig up scenarios where we would expect to see high y values by high x, yet they never appear. Almost like they're gated or compressed.
Even without a bunch of complicated math, you can explain this to normies very simply.
In these swing states, the more republicans a precinct had, the more likely they were to not vote straight ticket Republican.
Again, more Republicans meant fewer straight-ticket Republican ballots were cast, and those ballots had Republicans in all races except President.
Does that make any sense at all?
Now that you understand that makes no sense, look at the actual numbers. It's a linear decrease that only happened in swing states and only in precincts where Republicans were leading by more than 20%. It didn't happen in safe states, it didn't happen in Democrat precincts, and it didn't happen to Democrat ballots.
Be honest: Doesn't that seem fishy to you?
I think I figured out what they did to get curves like the ones I see in the plots but I'mm need to simulate it later.
Some of the plots Shiva displayed doesn't defy explanation simply because of the crook and steep dive but because if you project those they end up with something like 40% +-10% of straight republican voters consistently voting for Biden. It may be suppressed but a constant rate of 15% +-5% non-straight republican voters going to Trump.
That was one of the plots they had and regardless of the pattern just that base number of Trump under performing against republican candidates among republican bases so much is highly suspect.
I would conclude that the first part is more likely to contain fraud. I.e. if liberals are stuffing Biden-Only votes in the machine, then that would suppress the slope and make it horizontal closer to X=0. As you get into conservative precincts without vote stuffing, the steep angle reflects a more random distribution.
When I simulate 15% of votes on non-straight ballots going to Trump and 30% of ballots on straight republican votes going to Biden then I get a curve that's sort of parabolic.
Though I'm not sure if my equation and simulated data is the same as theirs. You can get a traversal starting high then going low but having it in the form of a horizontal start then a crook and then a linear steep dive isn't something that seems easy to produce.
It could be the other way around, if those other regions are meant to be higher except I'm not quite sure how that effect would push down to linear.
I see a problem. They have Y as the individual vote percent minus the straight party vote percent. That's a potential constant minus X so should be linear as X is linear and should descend after some threshold. That still doesn't explain the crook. The crook weakens to negligible if you extend the line straight or from the bottom. In two graphs the crook isn't actually strictly there despite them drawing it. On the one there appears to be one, it's not as extreme as it looks.
I thought the Y was something else but if it's that then I have to call it BS.
I meam, check my work, but does not that graph quite specifically result from the following psuedo code?
{ const RATE_A = 0.7, RATE_B = 0.15, RANGE_A = 0.5, RANGE_B = 0.5, precincts = [], {random, floor} = Math; for(let i = 0; i < 1000; i++) { const size = 50 + floor(random() * 50), straight = floor(random() * size), other = size - straight, votes_home = floor(straight * RATE_A * (1 + ((random() * RANGE_A) - RANGE_A * 0.5))), votes_away = floor(other * RATE_B * (1 + ((random() * RANGE_B) - RANGE_B * 0.5))), votes_messy = votes_home + votes_away, votes = floor(straight * RATE_A) + floor(other * RATE_B); precincts.push({size, straight, votes: votes_messy}) } const lines = [['index', 'size', 'straight', 'votes', 'difference', 'a', 'b'].join('\t')]; for(const [index, {size, straight, votes}] of precincts.entries()) { const difference = votes - straight, a = straight / size, b = votes && difference / votes; lines.push([index, size, straight, votes, difference, a, b].join('\t')) } console.log(lines.join('\n')) }
It's very rough and lazy but if you do something like this you see you get a curve, not a straight linear line.
Na this is how falsification works, and peer review, except sometimes the peer review is incorrect, but that's good, because it means it's one step closer to a truth.
Given the nature of the topic, I think people are just charged and are acting quickly because they feel a time pressure, and lets face it, there is one.
The guy has 4 MIT degrees. He probably knows math a bit better than a libtard journalist.
This is true. To be fair, the math involved is pretty basic though. Percentages, subtraction, plotting relations on an (x,y) Cartesian plane, and understanding linear relationships. I helped my 8th grade nephew with most of this already in public school.
What's more interesting (and where the appeal to authority / subjective weight of shiva's background in math/comp sci/signal processing/analysis) is the existence of 2 different trends in the different counties. Expected behavior until a certain point, then a deviation into something both unexpected and overly clean in appearance (a strong negative linear relation).
He could still be making it up or have incorrect assumptions. That's why research papers are peer reviewed.
There's a bunch of other people charting the same data and getting the same result https://boards.4chan.org/pol/thread/289949466
So it appears so far that he's not talking out his ass.
Math is not correct or not based on the number of degrees one has.
Dems are openly arguing that math facts are racially biased. Lol.
Exactly. Which us why I will not do their honework for them.
The more the state is blue Trump's and the Straight ticket match The more it is red, the less they match
Republicans don't have a reason to act that differently, especially dumping Trump as they are in more heavily Rep areas
These people are retards which is why they will lose
Unless I'm mistaken it's not that Trump was in the lead in Dem counties/precincts.
He uses straight party voting as a baseline. So, for example, if you have a rep/Dem/ind voter saying I like this Dem for this position this Republican for this one, ie split tickets. Those voters in "left" precinct broke heavily for Trump.
So the data shows, that the more a precinct had straight line Republican voters the more split tickets went to Biden. Now, it showed this consistently and reliably; nature would not produce this, and that only (99.99999) an algorithm would create.
Yes, that is what I am saying: why would Republicans in red counties act differently than Republicans in blue counties
They wouldn't
Not in this fashion
Yes, I think we are in total agreement. Party on pede.
Also, keep in mind that for us to believe dominions algorithm, not only would this piss in the face of nature, but you'd have to believe the highest rated president, by Republican standards, hated Trump, in a simple mathematical formula, the more Trump fans that they lived next to.
Exactly, Trump should have hovered ABOVE 0% difference on the y-axis. This would keep consistent with the number of Dems that voted Trump in the more Democrat precincts. So the estimate of votes given to Biden is the BARE MINIMUM. The data is damning and I would like to hear an argument describing how this trend was natural.
To clarify: the MI machines have a "vote the party" option, which is recorded if selected. That is different from marking all the Republican candidates in each race. The x-axis is determined by people who voted a party. The y-axis can contain people who voted all R on a ballot.
The analysis relies on an assumption that the people using the party vote option are not an overwhelming portion of the sample.
he is comparing
straight republican votes percentage to trump - straight republican votes percentage
the difference between them should be steady or around steady
instead as the county becomes more straight republican, republicans forget about Trump and he loses more and more votes
That's exactly not what the chart says but don't stop saying what you said.
the x axis is percent republicans 0 to the left - 100% to the right
Up to 20% Republican - so 80% Democrat - Trump has a lead (over 0 in the y axis)
Then as the percentage of republicans increases, Trump has less and less votes (under 0 in the y axis)
Where am I wrong?
Wrong.
Close but wrong.
How do you vote in Michigan?
Edit: I am not going to help you see how Dr. Shiva is wrong. I will help you understand his theory so you can explain it to others. I will not give so much detail as to show why it is wrong. I will mock dems who can't do math.
https://patch.com/img/cdn/users/457365/2012/10/raw/f040d3c163e152ad94800bd50130f8f5.jpg
https://media.mlive.com/newsnow_impact/photo/electionjpg-c08121176baf4ca5.jpg
You mean the option for straight party?
Yes.
X axis is the percentage of straight ticket voters who are republican. Y axis is the difference of the percentage of non-straight ticket votets who for trump minus the y axis.
I will say no more.
compared to straight ticket ballots. That's the only part you're missing. Basically, the more Republican a precinct is, the more likely they were to vote for every Republican except Trump, which makes no sense whatsoever. You would expect the percentage of people doing that to either remain static despite the precinct or for that to be more likely in heavily Democrat precincts. Instead it was happening in Republican precincts and happening in a very clear pattern that was duplicated across multiple large precincts in swing states.
Why would the Republicans in Redder counties act differently than Republicans in Bluer counties?
I don't think you understand what he has plotted and you are mistaken.
Look at the plots again. They clearly show that Republicans change behavior based on how red their county is, which doesn't make sense
Your left side of the graph point is ALSO correct, and staggering.
In precincts with less than 20% straight-R ballots, Trump OVERPERFORMS those ballots.
Meaning people who voted for D downballot (or, at least not straight-R) voted for Trump about 7% more than 50-50.
Meaning, in districts where there were 10% straight-R ballots, Trump got 17% of the vote.
Staggering.
And likely close to the true nature of the electorate nation wide.
it is exactly what the chart says, you are a moron
While mods are asleep???
There are no rules about calling out issues with data...
Wtf exactly do you think the mods would do about it? This isn't reddit......
And some of the people you're claiming this bullshit about are very well established and well known pedes around here...... What they say carries weight.
What they say carries no weight if they are mathematically wrong and spread things they either don’t understand or are using to demoralize. F off
I don’t have time to look at that data but I’ve been around forever and have known about that dude since 2016 - and I’m skeptical.
Shiva didn’t gather the data.
You are missing the point.
No I’m not. Shiva wasn’t the only engineer doing the study.
Bennie has been doing his own fraud studies for years.
The weight of idiocy.
Then by all means, report this supposed travesty to mods and see how quickly they don't give a fuck.... Let me know how it goes for you....
The pedes calling out his data have proof.... They aren't just saying "the Dr is wrong and you can totally believe and trust us!".......
What proof?
What proof?
I have no issue with people questioning the data, it needs to stand up under inspection.
This helps get to the truth.
Trumps suits are not going down this path. More breach of constitutional path
Thank you. Your arguments are logical and make sense based on some assumptions. So it comes down to whether individual candidate voters replicated the mix between straight R and D. The other question is what percent in each precinct votes straight party. If it's low then it would not be impossible or even too difficult for straight Republican party votes to be 80% and individual candidate votes to be higher, especially with high first time voter turnout.
Furthermore, conservative areas have conservative candidates from both parties and vice versa, so it could be the case that individuals feel comfortable not casting straight party votes there at a higher rate than in more competitive voting areas.
Just to be precise, Y must not be greater than zero when X is zero; it can be equal to zero as well. And Y must not be less than zero when X is 100; it can be equal to zero as well. If you have two people and one voted straight R and the other for Trump individually you get (100,0) coordinates; unlikely but possible.
Happy to hear if I'm missing something.
I just wanted to point out Dr Shiva made that video for normies (like me) to understand the jist of it. Maybe he’ll make another more technical video that gets into the nuts and bolts.
Dr. Shiva is working with BSEEs and CS guys. You're not going to see more rigor. You'd need math or physics guys for that.
And I'm not going to help the dems.
The burden of proof is much lower in court than in science.
Yeah, I was expecting something more - but I am glad it was easy enough to understand so that it is shareable! As a physics major, this kind of analysis is basic (which does not mean that it is wrong).
Not wrong doesn't mean right though. I have tons of questions about possible reasons to see a line. But when I see data like that, it looks too perfect.
I will say no more.
You are correct, it does not mean right either. Obviously there could be other explanations for it, but like you said it is too perfect and it is not our job to disprove it - let's leave that to the democrats. Though a healthy amount of scientific push back might be appreciated by a fellow scholar, it's probably not the best time for it.
It's one thing to convince pedes who love Trump. Another to convince a judge not only that fraud occurred, but that it was significant enough to have changed the result, in court, where those claims will be challenged by experts.
This analysis won't convince a judge to invalidate an election; it is used to convince a judge to order a hand recount or to get access to other evidence (ballot receipts, source code, etc.). The results of the recount or other evidence would then determine the election result.
Thank you, that's a great point. If a judge grants discovery, better evidence may be attainable.
My main issue is the lack of normalization of population across the wards (let alone acknowledgement of population sizes at all). Locations with lower likelihood to vote straight republican tend to have higher populations (cities).
Is the population of the ward that votes 80% straight ticket republican statistically comparable to the population that votes 20% straight republican? Does the within-sample variance contribute to the compounding effect we see in Shiva's graphs? Could an increasing variance of a field that both x and y are dependent on affect the trend we see? While the downward trend could be explained by a 50/50 split, and a straight graph would indicate increasing support as you vote straight ticket, is it also possible that relating the axes in an ambiguous way has introduced assumptions that may not be statistically relevant?
I just feel like Shiva's analysis was incredibly half-cocked. (he forgot to do any math)
Shiva essentially said 'heres an assumption, and here's a bunch of examples of that assumption being broken, and now here's one example of neither my assumption, nor my observations are correct (Wayne/Detroit)' All he's done is make assumptions and observations, but has yet to display those assumptions in action (for example, in a reliable district or compared to 2016), and he failed to even plot his trends with a regression, while asserting each district's anomaly has the same slope (contradictions).
I fear this analysis (with phone pictures of graphs, mind you) may be a distraction from people doing real math to find fraud.
I continue to see people assert that this is a conclusive argument, and that the responsibility lies in the user to improve his work, but this analysis is barely preliminary, and he's going to have a much harder time explaining away a lot of the holes in his presentation now. At least in my line of work, a stakeholder would never accept something like this in any way.
No matter how much you'd want to believe him, he's provided very little evidence to support his argument, and it is his prerogative to back up these claims. I support Trump as much as the next guy, but all I've ever seen from Shiva is snake oil.
Source: PhD Data Scientist
At this point in history, after everything we’ve seen over the last five years, if you’re still thinking this all goes south, you ain’t a black piller. You’re a fucking moron.
If you want Dr. Shiva's claims to stand don't think too hard about them. There is an explanation but the dems won't find it. Don't do their homework for them. I won't say anything more.
I am not a fucking shill. Check my history. I also was a 2016 primary voting OG deplorable.
Dr. Shiva's analysis is BAD. Check my comment history for an explanation.
As per my second edit, apologies for undeserved shill calling.
Shiva is a fraud lol... He markets himself as being the "inventor of Email" when it was Ray Tomlinson who actually invented email back in 1971 People who believes this garbage coming from a fraud like Shiva is beyond me.
You can call me a shill or whatever, but it will only take you few minutes to verify his claim is false.
I'm not trying to discredit is work. But then again I am not a shill.
Seriously, the only people you can discredit that shit to are the group that thinks 2+2=4 is racist. Shit.. so they'll discredit it to the entire left. Damnit!
Well, you wouldn't if you assume independents vote 50/50. just questioning the details.
I’m assuming no independents to make a mathematical point. But lets add independents and assume they all voted for Biden you would still get a straight line at zero. Basically, questioning Dr. Shiva is saying the more republican the district the more people defected from Trump. BS
Edit: addressed changed scenario to be individuals votes were cast in the same proportion as straight R or D causes the horizontal line at zero.
No you wouldn't get that straight line pede, the graph compares straight republican vote and percentage of individual votes that went to trump. Or to be more specific, it compares straight republican vote and the difference thereof with the second point mentioned. So basically anything above the line symbolizes a precinct where trump got a higher percentage of individual votes, than republicans got straight party votes. It doesn't say anything about his success overall.
(Just another example to make it extra clear: Say in a precinct we have 99 republicans, 1 democrat and 1 independent. Now let's say 100% of republicans voted straight party line, 100% of democrats voted straight party, and the independent voted for Biden, Trump would have won that precinct by ~99%. The dot would be at -99%, since that would be the difference in percentage of individual trump votes and straight party line votes)).
It’s so confusing my wife and I watched it over and over and the thing that convinced me finally is the statement the one guy makes “even if you argue that the higher percentage republicans areas hated Trump, they wouldn’t hate him in such a straight line” and this is the realization that 1. I don’t understand statistics and 2. I fully believe that there are precincts in that plot where they would have had high straight party votes AND high individually cast ballets for Trump. Also think about how ludicrous it is to think that drown ballet republicans would ever get more votes than trump!
In any case it should be random and close to a line, but we have various counties with the same strange slope.
Actually the dot would be at 0%, because the Trump vote exactly matches republican straight party vote. -99% would suggest that 99% of republicans voted for other candidates but not for Trump.
Doesn't the y-axis show the difference between straight republican votes and individual trump votes? (individual Trump votes - straight republican votes) Edit: Otherwise, how would anything below 0% be possible? It's kind of impossible to vote for every republican and at the same time not to vote for trump
From my understanding, which might be wrong, I happen to understand the 0% line as a basis of republican candidates support, and individual in this case represents how well Trump fares in contrast to republican candidates support. >0% means Trump is doing better than other republican candidates on the ballot, <0% is the opposite.
Ok let me start from 0. There are two types of voting:
In each one Trump gets a certain percentage of the vote. That gives us two percentages that we can compare.
As an example: we have 100 people. 99 republicans. 1 democrat. all republicans vote straight party. no democrats vote straight party. trump gets 100% of the straight party method vote. the democrat votes by the individual method, and he votes for Biden. in this category trump got 0% of the vote.
the y axis symbolises the difference in percentage between the two types of voting. So we subtract 100 from 0. we now have -100. the x axis symbolises the percentage that republicans got in the straight party method. in this case 100%. so we now would have a dot at (100 | -100).
Am I mistaken somewhere?
In yet another case, we have that single person vote for Trump (indie or cross party lines, or just a 99% R who didn't support one or more candidates diwn ballot or abstained), now y is 0 (as good as 100% RSP). This performance even if rare never occurs in the signal. After 25% or so, the outliers are gradually elimated and performance by Trump always negative.
Im not 100% sure about this honestly, just what I got from the video. In this case democratic vote is irrelevant unless they cast vote for Trump as well. If 100% of republican voters cast their vote straight down the ballot for every R, its at 0% or neutral line. If the one democrat(1% of total votes) voted only for Trump and didnt vote for other R candidates on the ballot, Trump result would be +1% invividually. Again, Im not an expert on statistics so dont take my word for it.
You are correct. 0% is the baseline. And the number of Republicans voting for other Republicans and not Trump must be TINY. So, any assumption of negative slope is bullshit. Maybe an assumption of -1-2% at the extremely Republican precincts, but I still think he would be in the positives even at super Republican precincts.
You are correct. I adjusted the scenario to show that a straight horizontal line at zero would happen when individual votes come in at the same proportion to straight R / straight D. My point still stands that a negative slope is not a rule. A more realistic scenario is that independents vote in similar proportion for the President as the straight party voting.
Is it? Have no data to back this up, I would assume though, that independents remain more or less consistent, meaning because of their consistency they would tip the percentage away from trump in the individual vote, the more republicans vote straight party. But I will be glad to see otherwise.
It doesn't have to be consisten across counties. I could see how behavioral factors, like herd mentality, could tip the independents towards the precinct party percentages. For example, if you see more yard signs for Trump or Biden in your neighborhood you might be influenced. It certainly seems to happen across states where an independent in California is more likely to be democratic leaning than one in Alabama. Not to mention democrats in red states were looking at Trump more favorably due to fear of the communists within the party.
Oh crap, you might be right.
It depends on how exactly they define the Y axis.
If they define it as "the % of non-straight-line-R votes that go to Trump", then you're right, the point is -99% (ie minus 99) on the Y-axis.
If it's defined as "% votes for Trump - % votes for straight-line R" then the point is at Y-axis -.01% (minus .01)
[see below, EDIT 3] And Egads, if it's the former, then it WOULD naturally be a straight line. There are fewer and fewer non-straight line ballots cast as you move down the X-axis so the right, so every ballot will be a bigger percent of the overall total, linearly.
EDIT: from the video: "Y-axis: Difference of %Trump individual candidate Votes minus %RSP (Republican Straight Party) votes
Yeah. Unfortunately he doesnt define well "Individual Candidate Votes"
EDIT 2: "Straight-ticket voting (also called straight-party voting) allows voters to choose a party’s entire slate of candidates with just a single ballot mark. Voters make one mark or selection on the ballot in order to vote for every candidate of that party for each partisan office on the ballot.
In 2020, a total of 6 states will allow or offer straight-ticket voting (STV). With a few exceptions, the straight-ticket option is available in all general elections, and applies to all partisan offices on the ticket, including federal, state and local races.
The states with STV are: Alabama, Indiana*, Michigan, Kentucky, Oklahoma, and South Carolina."
from: https://www.ncsl.org/research/elections-and-campaigns/straight-ticket-voting.aspx
EDIT 3: Ok you uncovered something that I didnt see before. There is something that increases linearly with increasing republican straight-party vote percent. Since there are fewer and fewer non-RSP votes available, each ballot DOES make a bigger and bigger difference as you move right across the graph. That should make the spread wider, but it doesnt get wider.
But ...some examples for sanity check:
If a district votes 20% RSP, then 25% (ie +5% on the graph) of the individual ballots (not including the straight-D ballots) go to Trump
If a district votes 30% RSP, then 25% (ie -5% on the graph) of the individual ballots (not including the straight-D ballots) go to Trump
If a district votes 40% RSP, then 30% (ie -10% on the graph) of the individual ballots (not including the straight-D ballots) go to Trump
If a district votes 50% RSP, then 35% (ie -15% on the graph) of the individual ballots (not including the straight-D ballots) go to Trump
If a district votes 60% RSP, then 40% (ie -20% on the graph) of the individual ballots (not including the straight-D ballots) go to Trump
So Trump independent votes stills rises with increasing RSP, but not as much as the RSP itself. It lags. Why?
still thinking...
Great work pede! Yep, pretty much this is what I have been thinking about too. If you have virtually no Republican support it becomes easier for Republicans to do worse, than Trump with independents. Take a heavy dem state with some disaffected Bernie voters, that just vote Trump as an f u to the establishment. This would explain trump being above the RSP line in the beginning.
As far as the lagging behind of independents, I think you could say, that this is due to the nature of independents. They are more or less undecided. So it might be easier for a republican to vote straight-party, than it is for an independent to vote for trump. Note that this is just speculation and in order to really find an answer to this, one needs to check all sorts of details.
Voting behaviour of independents in relationship to political leanings of their county, voting behaviour of both parties, frequency of straight-party voting and so on, just to list a few.
You can get the impression from the video that the more 'republican' a county was, the worse trump did. From what I understand, that is misrepresentative because it compares two different types of voting and percentages, not absolute numbers.
Does anyone have a good link to his findings?
This goes right over my head but I'm pretty excited about it
Not a shrill I promise, but if hypothetically Republicans were twice as likely to vote straight ticket than Democrats, wouldn’t that explain the results he is seeing.
Not necessarily. And that's the point, it should be more random rather than a tight fit around a negative slope. At a given level of straight R support, some precincts' individual votes would favor Biden and some Trump, causing data to hover around zero at the horizontal.
Right, but I believe he is assuming the % Republican by straight ticket % that is Republican. So if hypothetically republicans are 3 times as likely (not twice as I stated earlier before doing the math) to vote straight ticket but only 60% of the precinct is Republican then the x axis would be 80% and y would be -20%.
For example, if 30% Democrats vote straight ticket and 90% Republicans vote straight ticket and the precinct is 60% Republican, the x axis would be .6x.9 /(.6x.9+.4x.3) = 60%.
This works for any % of Dem straight ticket voters. Of course that can not be greater than 33% because then Republican % could not be 3 times higher. Would be interested to see if the percent of total straight ticket votes increased as precincts got more red.
You wouldn’t see the exact same slope in different counties at different times (mail-in vs Election Day). All other 2020 data shows differences with mail-in vs Election Day.
Indeed. The deep state propagandists and their cohorts are being busy bees.....
I wasn't necessarily referring to this in particular. Take a chill pill.
I gave it some thought and this is why the video is problematic:
If 100% of Straight party votes go to Republicans Trump cannot do better than 0 on the y axis. If 100% of Straight party votes go to democrats, trump can ONLY do better than 0.
You will always get a negative slope.
We need to look at the source code of the machines.
that fix nothing if they was connected to internet.
Disagree. Probability is what cases are built on. DNA forensics, for example, are consistently used as strong evidence because they show that there's only a tiny chance that someone else would leave the same markers/material behind.
The same basic structure of argument would work here, but the probabilities are in fact much more incredible. If these data analyses are accurate, and they're happening in multiple places, the chances that this is not fraud quickly approaches 0. And it would absolutely hold up in court, but it would need to be in the right court with the right jurisdiction, remedy, etc.. That's the difficult part as I see it.