It's illegal to write that kind of thing about Pennsylvania without referencing Clinton's "Pittsburgh and Philadelphia with Arkansas in between" line -- which is actually pretty accurate.
I don't think and electoral college advantage is measured by comparing the popular vote against the percentage in the tipping point state. I don't see a clear linkage in terms of math or statistics, although it has intuitive appeal.
His analysis is statistical, he links the electroal college advantage to "wasted votes", the votes a party has above the number needed to win the state, 3,620,000 in the case of CA. In total he calculates 11,050,000M wasted Dem votes against 5,450,000 Rep.
You can have a 3% EC advantage going in and win at +2 or lose at +4, it depends on state by state vagaries. The EC advantage is a statistical edge, like a coin that weighs slightly more on one side. It cannot be measured by the results of one coin toss, each individual toss could be, for example, 51/49. If the 49% side happens to win a toss that does not mean there was no advantage. Similarly the EC advantage cannot be measured after the fact by subtracting two numbers.
The Electoral College advantage is a function of how different individual states are, on average across a variety of iterations of a model based on historical and fundamental data, from the national electorate. But the composition of both state and national electorates is a function of how the parties campaign. That's why the advantage can shift, significantly.
As a result, it is a true statement that, if you could predict that turnout in California would drop relative to the nation as a whole, you would also expect the Republican advantage in the Electoral College -- statistically speaking -- to drop relative to past elections.
You're right of course that you can't measure the advantage *going in* by what it was last time, but if you're trying to measure what it *turned out to be* in the last election, you would do it exactly the way I did: by comparing the tipping point state margin to the national popular vote margin. Nate Silver does that all the time.
These are all valid points. Yet it still feels like there's a specific sense in which the EC is unusually salient in the era of Trump. He has only been competitive in national politics, despite his lack of popularity, because he performs unusually well in 3 states that Republicans hadn't won since 1988, and "performs unusually well" in this case means "is capable of winning by the skin of his teeth, and no more."
If I disagree with anything, it's your last paragraph; it seems that for most of US history, most campaigns DID at least sort of behave as though the national popular vote mattered, and this was rational when it was generally a near-perfect indicator of who would win the EC. It's a lot easier to try to win over the American people more broadly, rather than try to hack your way into an EC-winning minority coalition by narrowly tipping a few swing states in your favor in exchange for accepting bigger-than-usual blowout margins in states you can't win. Trump presumably stumbled upon his 2016 map by dumb luck -- he too was trying to win a majority of the popular vote and maybe still believes that he actually did win it. But having discovered that "hack", recreating that 2016 map as much as possible is the only plausible path for victory for him that he or anyone else can see.
It's illegal to write that kind of thing about Pennsylvania without referencing Clinton's "Pittsburgh and Philadelphia with Arkansas in between" line -- which is actually pretty accurate.
If memory serves that quote is attributed to James Carville.
I don't think and electoral college advantage is measured by comparing the popular vote against the percentage in the tipping point state. I don't see a clear linkage in terms of math or statistics, although it has intuitive appeal.
Silver Bulletin has a free article about this
https://www.natesilver.net/p/mad-about-the-electoral-college-blame
His analysis is statistical, he links the electroal college advantage to "wasted votes", the votes a party has above the number needed to win the state, 3,620,000 in the case of CA. In total he calculates 11,050,000M wasted Dem votes against 5,450,000 Rep.
You can have a 3% EC advantage going in and win at +2 or lose at +4, it depends on state by state vagaries. The EC advantage is a statistical edge, like a coin that weighs slightly more on one side. It cannot be measured by the results of one coin toss, each individual toss could be, for example, 51/49. If the 49% side happens to win a toss that does not mean there was no advantage. Similarly the EC advantage cannot be measured after the fact by subtracting two numbers.
I don't think we actually disagree.
The Electoral College advantage is a function of how different individual states are, on average across a variety of iterations of a model based on historical and fundamental data, from the national electorate. But the composition of both state and national electorates is a function of how the parties campaign. That's why the advantage can shift, significantly.
As a result, it is a true statement that, if you could predict that turnout in California would drop relative to the nation as a whole, you would also expect the Republican advantage in the Electoral College -- statistically speaking -- to drop relative to past elections.
You're right of course that you can't measure the advantage *going in* by what it was last time, but if you're trying to measure what it *turned out to be* in the last election, you would do it exactly the way I did: by comparing the tipping point state margin to the national popular vote margin. Nate Silver does that all the time.
These are all valid points. Yet it still feels like there's a specific sense in which the EC is unusually salient in the era of Trump. He has only been competitive in national politics, despite his lack of popularity, because he performs unusually well in 3 states that Republicans hadn't won since 1988, and "performs unusually well" in this case means "is capable of winning by the skin of his teeth, and no more."
If I disagree with anything, it's your last paragraph; it seems that for most of US history, most campaigns DID at least sort of behave as though the national popular vote mattered, and this was rational when it was generally a near-perfect indicator of who would win the EC. It's a lot easier to try to win over the American people more broadly, rather than try to hack your way into an EC-winning minority coalition by narrowly tipping a few swing states in your favor in exchange for accepting bigger-than-usual blowout margins in states you can't win. Trump presumably stumbled upon his 2016 map by dumb luck -- he too was trying to win a majority of the popular vote and maybe still believes that he actually did win it. But having discovered that "hack", recreating that 2016 map as much as possible is the only plausible path for victory for him that he or anyone else can see.