There are not fewer Republicans voting for Trump in those precincts moving left to right in his graph. There are more. His sloping graph is a result of subtracting the x-axis value from the mixed ballot Trump voters. Of course that result will go negative as x increases.
I'm not following this.
Precinct A: 20% vote Republican in other races and Trump receives 25% of the vote = +5% (net difference from potentially lost Rep and gained Dem, Ind voters)
Precinct B: 75% vote Republican in other races and Trump receives 55% of the vote = -20% (net difference from potentially lost Rep and gained Dem, Ind voters)
In a heavy Rep area, why would Trump lose so many Rep voters to go below neutral?
Each dot on Shiva's plot is a precinct. The dots coordinate has an X & Y value. X is the percentage of straight Republican ballots (voters who voted Republican all the way down the ballot). All of Shiva's numbers are percentages (I can't emphasize this enough) not number of votes. Y is the PERCENTAGE of voters who filled out a mixed ballot but voted for Trump (meaning they didn't vote straight Republican all the way down the ballot (those include Reps, Dems and Independents - likely mostly Independents)). In heavy Rep areas he is NOT losing Rep votes. Shiva's trend line is going down because his equation for Y looks like this: y=c-x where c is the PERCENTAGE of mixed ballot Trump voters (heavily Independents) MINUS the PERCENTAGE of straight ticket Republican voters. As you move left to right the value of x increases linearly. Try it yourself, plug those increasing x-values into that equation (y=c-x) and see what "c" needs to be to keep "y" constant (not go into a downward slope). Remember "c" represents the percentage of mixed ballot votes for Trump. In order for the trendline to stay flat and constant "c" would have to increase as you move into more Republican precincts. And "c" essentially is a look at how Independents voted. I doubt they linearly increase their vote for Trump as you get more Republican. I would expect that we'd see a fairly constant percentage of Independent support for Trump across the County - and that's what the data is showing.
TL;DR The percentage of Independents who voted for Trump stays very consistent and steady between heavy Dem precincts to heavy Rep precincts. That seems pretty logical to me. Shiva is saying Independent voters should go for Trump in an increasing percentage of mixed voters as you move into heavy Republican precincts. That's the assumption he made and I think it's flawed.
Your arguement hinges on the assumption that the percentage of Trump only voters should not track with the percentage of GOP straight ticket voters. Most Trump only voters are likely to be disgruntled GOP voters. It seems highly likely that precincts with higher GOP turnout include both higher straight ticket and higher Trump only voters.
There are not fewer Republicans voting for Trump in those precincts moving left to right in his graph. There are more. His sloping graph is a result of subtracting the x-axis value from the mixed ballot Trump voters. Of course that result will go negative as x increases.
I'm not following this. Precinct A: 20% vote Republican in other races and Trump receives 25% of the vote = +5% (net difference from potentially lost Rep and gained Dem, Ind voters) Precinct B: 75% vote Republican in other races and Trump receives 55% of the vote = -20% (net difference from potentially lost Rep and gained Dem, Ind voters)
In a heavy Rep area, why would Trump lose so many Rep voters to go below neutral?
Each dot on Shiva's plot is a precinct. The dots coordinate has an X & Y value. X is the percentage of straight Republican ballots (voters who voted Republican all the way down the ballot). All of Shiva's numbers are percentages (I can't emphasize this enough) not number of votes. Y is the PERCENTAGE of voters who filled out a mixed ballot but voted for Trump (meaning they didn't vote straight Republican all the way down the ballot (those include Reps, Dems and Independents - likely mostly Independents)). In heavy Rep areas he is NOT losing Rep votes. Shiva's trend line is going down because his equation for Y looks like this: y=c-x where c is the PERCENTAGE of mixed ballot Trump voters (heavily Independents) MINUS the PERCENTAGE of straight ticket Republican voters. As you move left to right the value of x increases linearly. Try it yourself, plug those increasing x-values into that equation (y=c-x) and see what "c" needs to be to keep "y" constant (not go into a downward slope). Remember "c" represents the percentage of mixed ballot votes for Trump. In order for the trendline to stay flat and constant "c" would have to increase as you move into more Republican precincts. And "c" essentially is a look at how Independents voted. I doubt they linearly increase their vote for Trump as you get more Republican. I would expect that we'd see a fairly constant percentage of Independent support for Trump across the County - and that's what the data is showing.
TL;DR The percentage of Independents who voted for Trump stays very consistent and steady between heavy Dem precincts to heavy Rep precincts. That seems pretty logical to me. Shiva is saying Independent voters should go for Trump in an increasing percentage of mixed voters as you move into heavy Republican precincts. That's the assumption he made and I think it's flawed.
Your arguement hinges on the assumption that the percentage of Trump only voters should not track with the percentage of GOP straight ticket voters. Most Trump only voters are likely to be disgruntled GOP voters. It seems highly likely that precincts with higher GOP turnout include both higher straight ticket and higher Trump only voters.