My understanding of the charts was that a certain percentage of dislike for Trump would've appeared as a flat line below the 0% threshold.
Instead what we're seeing is that the percentage increase in "not voting for Trump" increased linearly with the percentage increase in Republican voters.
That isn't saying "Oh, 10% of Republicans didn't like Trump", it's saying that deep red areas... somehow... didn't like Trump, but light red or even blue areas did. That's what doesn't make sense.
If 5% R went away consistently, then consider at only 10% R straight ticket, 9.5% of the non-straight ticket R votes Trump.
9.5% - 10% = -0.5%
At 20% R straight ticket, losing 5% of the R vote still
19% - 20% = -1%
At 40% R straight ticket, still losing 5%
38% - 40% = -2%
At 60% R straight ticket
57% - 60% = -3%
So it is a very gradual line, but the way it is set up naturally leads to a line with a negative slope when anything less than 100% party support.
I agree, 5% is the upper limit, and even then it barely goes down, so I think they clearly siphoned off votes (a straight percentage of R, with some fuzz added in), but the Romneys and Kamalas can say "I guess your polling is wrong, 20% of Rs in MI can't stand Trump!"
And their base will buy it.
You're right and they even address this exact issue in their presentation. The poster above you is wrong.
Certainly not all Republicans will vote for Trump (although he does have historically high approval ratings). However, we would not see a perfectly linear decrease that grows as the number of Republicans in a district increases. Besides, the downward slope is too perfect which allows Dr. Shiva to precisely predict the number of votes transferred to Biden.
I thought like you, then I started to do the math and realized one of my friends was right, it does naturally go down when set up that way in the situation that a percentage of R voters consistently don't vote Trump. Look at my other comment to see numbers worked out.
Obviously all those voters will not be in the straight ticket, so they will go to the y-axis, and it will gradually slope down. It isn't 1-1 straight ticket R to non-straight ticket R, because anyone who doesn't want Trump (or Biden for D) will always be distributed into the non-straight ticket section that is graphed in the y-axis.
The problem is how quickly it sloped down, clearly higher than 5%.
I think that slope is about how the same data would look sorting by Biden voters with no siphoning, but the principle holds: if a stable portion of dems walked away, with this graph set up, the more dems in a district, % D biden - % D straight ticket
would trend linearly down.
What?
My understanding of the charts was that a certain percentage of dislike for Trump would've appeared as a flat line below the 0% threshold.
Instead what we're seeing is that the percentage increase in "not voting for Trump" increased linearly with the percentage increase in Republican voters.
That isn't saying "Oh, 10% of Republicans didn't like Trump", it's saying that deep red areas... somehow... didn't like Trump, but light red or even blue areas did. That's what doesn't make sense.
Precincts that voted....
40% Republican: ~5% Trump siphoned off
50% Republican: ~10% Trump siphoned off
60% Republican: ~15% Trump siphoned off
70% Republican: ~20% Trump siphoned off
80% Republican: ~25% Trump siphoned off
That makes no sense at all.
If 5% R went away consistently, then consider at only 10% R straight ticket, 9.5% of the non-straight ticket R votes Trump. 9.5% - 10% = -0.5%
At 20% R straight ticket, losing 5% of the R vote still 19% - 20% = -1%
At 40% R straight ticket, still losing 5% 38% - 40% = -2%
At 60% R straight ticket 57% - 60% = -3%
So it is a very gradual line, but the way it is set up naturally leads to a line with a negative slope when anything less than 100% party support.
I agree, 5% is the upper limit, and even then it barely goes down, so I think they clearly siphoned off votes (a straight percentage of R, with some fuzz added in), but the Romneys and Kamalas can say "I guess your polling is wrong, 20% of Rs in MI can't stand Trump!" And their base will buy it.
You're right and they even address this exact issue in their presentation. The poster above you is wrong.
Certainly not all Republicans will vote for Trump (although he does have historically high approval ratings). However, we would not see a perfectly linear decrease that grows as the number of Republicans in a district increases. Besides, the downward slope is too perfect which allows Dr. Shiva to precisely predict the number of votes transferred to Biden.
Look at the video 44m in for a discussion. https://www.youtube.com/watch?v=Ztu5Y5obWPk&ab_channel=Dr.ShivaAyyadurai
I thought like you, then I started to do the math and realized one of my friends was right, it does naturally go down when set up that way in the situation that a percentage of R voters consistently don't vote Trump. Look at my other comment to see numbers worked out.
Obviously all those voters will not be in the straight ticket, so they will go to the y-axis, and it will gradually slope down. It isn't 1-1 straight ticket R to non-straight ticket R, because anyone who doesn't want Trump (or Biden for D) will always be distributed into the non-straight ticket section that is graphed in the y-axis.
The problem is how quickly it sloped down, clearly higher than 5%. I think that slope is about how the same data would look sorting by Biden voters with no siphoning, but the principle holds: if a stable portion of dems walked away, with this graph set up, the more dems in a district, % D biden - % D straight ticket would trend linearly down.