Okay. I'm an engineer. I've watched this video 4 times now. It doesn't add up. He is plotting mixed ballot Trump vote percentages, minus the straight republican ballot percentage, on the Y-axis. Then plotting straight Republican ballot percentages by themselves on the X-axis. Then he's saying there's an anomaly that is a negative linear slope. Well yes, of course there's a negative linear slope, his equation is subtracting a linear number (X-axis value) from a percentage of Trump voters who filled out a mixed ballot - which should be a fairly constant value after it initially gains from zero. The only way this result stays horizontal (not sloped) is for the percentage of Trump voters casting mixed ballots to equal the corrosponding X-axis value at that precinct. But we should think that there is a limit to the growth of mixed ballot Trump voters no matter where they live. I gotta be honest, unless I'm missing something this data analysis seems pretty quacky. It gets really quacky at the 30 min mark where he does an analysis of 2000 voters in a 45% republican straight ballot precinct - but then somehow forgets that and says that Trump only got 45% of 1000. He then jumps with no explanation to suddenly Trump is losing 25K votes and Biden is adding them. He really lost me there. Some Pede tell me where I'm wrong but this Shiva's logic and numbers aren't adding up for me.
Also engineer, also reached same conclusion. Will copy paste reply from earlier I made.
I'm going to go full doomer here. Doesn't this just mean in precincts where Republicans used the straight ticket option more, there were less available people to vote Trump? Inversely, in places where, let's say 20% of people voted straight R, the remaining 80% of people could have voted Trump for prez then down ballot for dems? Wouldn't that make sense for largely democrat areas where Trump has won people over? In order words, in largely dem areas where no one would use RSP, Trump literally has no where to go but up. If 0% use RPS, the number of Trump-RSP has to be positive or zero. I think that's why these charts start with X<20% and Y being positive.
A downtrend being observed is not surprising at all. If 100% of Trump voters voted RSP, then Trump-RSP would have nowhere to go but down. In fact if 100% of Trump voters voted RSP, Trump then gets NO individual votes which I think would be a -100% score.
Also I caught that trick at 38:50 The X axis only goes to 35% so the effect I'm describing above would not take noticeable effect yet. Compare that chart to other charts when >35% is cropped out.
Doesn't this just mean in precincts where Republicans used the straight ticket option more, there were less available people to vote Trump?
I thought so at first too, but realize he said the independent votes and straight party votes were separate data sets. Each data set should have proportions that add up to 100%.
The X axis expresses "how Republican is the precinct" as inferred by the percentage of precinct voters that chose the Straight Republican option.
The Y axis shows if the Individual Candidate voters were "more or less favorable to Trump". They do this by plotting the difference between these voters' Trump % to the the straight Republican voters' split.
For example, in a precinct which had 80% Straight Republican (vs 20% Straight Democrat) but only 20% Individual Trump (vs 80% Individual Biden), you'd put that precinct as a -60% on the Y axis.
They do this by plotting the difference between these voters' Trump % to the the straight Republican voters' split.
This is the problem right here. He is creating a function with y dependent on x, and then acting surprised that there is a relationship between y and x.
The function for Y is chosen to make the visualization clear. You could also express it as the split between Trump and Biden for Individual Choice voters, and you’d see a trend after 20ish% of X where Y skews towards Biden linearly.
Not an engineer, but took the time to makes sure I followed everything and the problem seems to be that the percentage of Trump votes that are straight republican ticket votes is assumed to be the same as the percentage of Trump votes that are individual filled in. That's what the red line above is showing. Having an assumption as your baseline cannot be good math. It seems that this graph could also be showing that as partisanship increases, voters are more likely to vote "straight ticket"
You're correct that as straight Republican ballot percentages approach 100% then the surplus Trump vote percentage must approach 0%, for the sum of both cannot be greater than 100%.
But that doesn't explain the near uniformity of the slope of the problematic counties, nor the fact that the algorithm looks to be "if republican_straight_ticket > 20%, then force slope to -1/2"
The near uniformity of the slope could be explained by a constant President-Only/Mixed vote for Trump. If that number is assumed to be a constant 50% and orthogonal to the Straight Ballot share, then you'll see a constant 10% decrease for every 10% increase of Straight Ballot R share above 50%... linearly. Also, at below 20%, likely because of small N of straight R ballots.
I see the points being made by the arguments above, but similar to the video they all require that assumptions be made. The truth is that it is damn hard to prove anything either way at this point. If it is fraud, the people doing it aren't stupid and wouldn't make it easy to catch. This is a good resource to get people to question what is going on, maybe even freak out some people who may or may not be "in" on it. It is not a smoking gun. As always, we should keep our minds open to all possibilities in order to come to the best understanding of what is actually going on.
Thank you. I'm also an engineer and came to the same conclusion, but felt like I must be missing something.
TLDR: The guy set his y-axis as y = -x + t, where x is the number of straight ticket voters and t is the number of independent Trump voters. He himself said t should be constant, therefore he is plotting a linear function with a negative slope.
He is finding the difference between two percentages.
Y the percentage of Straight Republican Vote (SRV)
X being the percentage of Trump Individual Vote (TIV) - Y
He argues that in a normal case, the percentage of TIV would be close to percentage of SRV (+/- 7%). So when a precinct has a SRV of 65%, the TIV would be between 60-70%.
There's no reason to assume that the percentage of TIV would be a constant (say 50%), which would be what is required for this to be a linear negative slope.
Even in the case where the SRV approaches 99%, the TIV should also be corespondent - because ultimately it's tracking the same population.
Okay. I'm an engineer. I've watched this video 4 times now. It doesn't add up. He is plotting mixed ballot Trump vote percentages, minus the straight republican ballot percentage, on the Y-axis. Then plotting straight Republican ballot percentages by themselves on the X-axis. Then he's saying there's an anomaly that is a negative linear slope. Well yes, of course there's a negative linear slope, his equation is subtracting a linear number (X-axis value) from a percentage of Trump voters who filled out a mixed ballot - which should be a fairly constant value after it initially gains from zero. The only way this result stays horizontal (not sloped) is for the percentage of Trump voters casting mixed ballots to equal the corrosponding X-axis value at that precinct. But we should think that there is a limit to the growth of mixed ballot Trump voters no matter where they live. I gotta be honest, unless I'm missing something this data analysis seems pretty quacky. It gets really quacky at the 30 min mark where he does an analysis of 2000 voters in a 45% republican straight ballot precinct - but then somehow forgets that and says that Trump only got 45% of 1000. He then jumps with no explanation to suddenly Trump is losing 25K votes and Biden is adding them. He really lost me there. Some Pede tell me where I'm wrong but this Shiva's logic and numbers aren't adding up for me.
Also engineer, also reached same conclusion. Will copy paste reply from earlier I made.
I'm going to go full doomer here. Doesn't this just mean in precincts where Republicans used the straight ticket option more, there were less available people to vote Trump? Inversely, in places where, let's say 20% of people voted straight R, the remaining 80% of people could have voted Trump for prez then down ballot for dems? Wouldn't that make sense for largely democrat areas where Trump has won people over? In order words, in largely dem areas where no one would use RSP, Trump literally has no where to go but up. If 0% use RPS, the number of Trump-RSP has to be positive or zero. I think that's why these charts start with X<20% and Y being positive.
A downtrend being observed is not surprising at all. If 100% of Trump voters voted RSP, then Trump-RSP would have nowhere to go but down. In fact if 100% of Trump voters voted RSP, Trump then gets NO individual votes which I think would be a -100% score.
Also I caught that trick at 38:50 The X axis only goes to 35% so the effect I'm describing above would not take noticeable effect yet. Compare that chart to other charts when >35% is cropped out.
I thought so at first too, but realize he said the independent votes and straight party votes were separate data sets. Each data set should have proportions that add up to 100%.
The X axis expresses "how Republican is the precinct" as inferred by the percentage of precinct voters that chose the Straight Republican option.
The Y axis shows if the Individual Candidate voters were "more or less favorable to Trump". They do this by plotting the difference between these voters' Trump % to the the straight Republican voters' split.
For example, in a precinct which had 80% Straight Republican (vs 20% Straight Democrat) but only 20% Individual Trump (vs 80% Individual Biden), you'd put that precinct as a -60% on the Y axis.
This is the problem right here. He is creating a function with y dependent on x, and then acting surprised that there is a relationship between y and x.
The function for Y is chosen to make the visualization clear. You could also express it as the split between Trump and Biden for Individual Choice voters, and you’d see a trend after 20ish% of X where Y skews towards Biden linearly.
Not an engineer, but took the time to makes sure I followed everything and the problem seems to be that the percentage of Trump votes that are straight republican ticket votes is assumed to be the same as the percentage of Trump votes that are individual filled in. That's what the red line above is showing. Having an assumption as your baseline cannot be good math. It seems that this graph could also be showing that as partisanship increases, voters are more likely to vote "straight ticket"
You're correct that as straight Republican ballot percentages approach 100% then the surplus Trump vote percentage must approach 0%, for the sum of both cannot be greater than 100%.
But that doesn't explain the near uniformity of the slope of the problematic counties, nor the fact that the algorithm looks to be "if republican_straight_ticket > 20%, then force slope to -1/2"
The near uniformity of the slope could be explained by a constant President-Only/Mixed vote for Trump. If that number is assumed to be a constant 50% and orthogonal to the Straight Ballot share, then you'll see a constant 10% decrease for every 10% increase of Straight Ballot R share above 50%... linearly. Also, at below 20%, likely because of small N of straight R ballots.
I see the points being made by the arguments above, but similar to the video they all require that assumptions be made. The truth is that it is damn hard to prove anything either way at this point. If it is fraud, the people doing it aren't stupid and wouldn't make it easy to catch. This is a good resource to get people to question what is going on, maybe even freak out some people who may or may not be "in" on it. It is not a smoking gun. As always, we should keep our minds open to all possibilities in order to come to the best understanding of what is actually going on.
Thank you. I'm also an engineer and came to the same conclusion, but felt like I must be missing something.
TLDR: The guy set his y-axis as y = -x + t, where x is the number of straight ticket voters and t is the number of independent Trump voters. He himself said t should be constant, therefore he is plotting a linear function with a negative slope.
That's not quite right.
He is finding the difference between two percentages.
Y the percentage of Straight Republican Vote (SRV) X being the percentage of Trump Individual Vote (TIV) - Y
He argues that in a normal case, the percentage of TIV would be close to percentage of SRV (+/- 7%). So when a precinct has a SRV of 65%, the TIV would be between 60-70%.
There's no reason to assume that the percentage of TIV would be a constant (say 50%), which would be what is required for this to be a linear negative slope.
Even in the case where the SRV approaches 99%, the TIV should also be corespondent - because ultimately it's tracking the same population.