election correction

Here's a little calculation I was able to do in a jiffy by plugging an existing table into an Excel spreadsheet and adding a few calculation columns.

In reply to the fact that Clinton received more votes than Trump, supporters of the latter have suggested that the Founding Fathers set up the Electoral College deliberately to favor candidates with wider geographic support.

That's not as strong an argument as it may seem, given that Clinton carried 21 states (including D.C.) and Trump 30; although Clinton's majority can be accounted for by a couple of large states, it's not as if her winning areas were limited to a couple enclaves.

Nevertheless, it's a stronger argument than the ones the Republicans used to offer, where they would present a US county outline map colored in by which candidate won which county. The vast expanses of red couldn't hide the fact that most of those vast expanses were pretty empty. Square miles don't get a vote.

But geographic spread per se was not the Founding Fathers' intent. What they were trying to do was preserve the interests of small states, which they had to do because they were operating in a forum where each state, regardless of population, had one vote. Like, say, giving slave-holders extra seats in both Congress and the Electoral College on account of the non-citizen slaves they held, treating Delaware as equal to Pennsylvania or New York has not held up over time as one of the Founders' wiser plans. (And I can't help cheekily noting that, if we were to return to the Continental Congress era one vote per state system, Clinton won 9 of the original 13.)

But even if we allow the extra weight that the Electoral College gives to smaller states, the unit rule "winner take all" voting system was not the Founders' intent. The Constitution specifies no method by which electors are to be chosen, and in the early days states used a wide variety, of which state-wide vote was one of the less common. Somewhat more frequent, if popular vote was used at all, was to divide the state into districts, each one choosing one elector. (The current Maine-Nebraska system, with two statewide and the others chosen by congressional district, was not used.)

It occurred to me that an idealized approximation of this, and a way to test the theory that geographic spread is important, would be to run a notional electoral college in which the electors in each state were assigned to approximate as closely as possible the popular vote percentage in that state. That would still give the small states extra weight, but it would be an honest reflection of the actual vote in the states, thus testing what the geographic spread actually is, not just a notional take-all winner in each state.

So I multiplied each candidate's percentage of the popular vote by the number of electors in that state, assigned each candidate the closest whole number of electors for the result, and if there was one left over, assigned it to the candidate with the largest remainder.

First I should note that the US is purple. Except for DC which remained 3 Clinton, every state had at least one vote for Clinton and one for Trump. Even West Virginia, which voted only 26.5% Clinton, that share would come to 1.32 of the 5 electoral votes, so she gets 1 vote in my count.

Interestingly, the result came out as an exact tie: 261 votes each for Clinton and Trump. There were 14 votes for Gary Johnson, and one each for McMullin (Utah) and Stein (California). If you adjust it to just the two-party vote, the 16 third-party electors divide up 8 and 8, and it's still a tie.

So it helps Trump, but not quite enough. You'd need some other system, or a tie-spli8tter, to give him an outright victory this way. That geographic spread is not as powerful an argument as his supporters think.