Elections like this, and like the election in 2000, when the popular vote winner is not the same as the electoral vote winner, make people stop and think about the system we have. I have had quite a few discussions about this over the past month, and I am moved by the arguments in favor of the electoral college. Some say it’s antiquated, and that it was a system designed to give more power to states with high slave populations, but without giving slaves the right to vote. True or not, I accept the premise of why the electoral college is important today: it gives a voice to the small states, and helps make sure they are heard. This has been spun as a benefit to republicans, but the evidence supporting that isn’t very strong.

It is absolutely true that the electoral college creates a scenario where an individual vote matters more in a smaller state than it matters in a larger state. If we calculate electoral votes per million voters, we find that places like Wyoming (the smallest populated state in the country) have 5.3 electoral votes for every million people who live there. This is quite different from California (the most populated state), where every million people living there get 1.5 electoral votes. Using these two states as examples, the deep-red Wyoming, where 70% of voters picked Trump, and the blue (but not as deeply blue) California, where 61.5% of voters picked Clinton, it sure seems like a built-in GOP advantage. But if we rank order the states by their proportionate representation in the electoral college, we find a top 10 most over-represented list with, in ascending order, Wyoming, District of Columbia, Vermont, North Dakota, Alaska, Rhode Island, South Dakota, Delaware, New Hampshire, and Montana. That’s 5 states that almost always go GOP, 4 states that almost always go Dem, and New Hampshire that’s harder to predict. If we look at the top 10 under-represented states, we get, in descending order, California, New York, Texas, Florida, Illinois, Ohio, North Carolina, Pennsylvania, New Jersey, and Virginia. California, New York, New Jersey are more solidly blue than the others, but this doesn’t seem to be home to any particular GOP or Dem advantage either.

As any scientist might recognize looking at only the top and bottom 10 ignores the middle, where there might be some importance. Just in case there’s some meat in that range, I coded states that picked Trump as 1, and states that picked Clinton as -1, and ran a correlation analysis on the pick and the proportionate representation of voters in the electoral college. The result was a correlation coefficient of -0.081. This is a very slight, but statistically insignificant, lean toward favoring Clinton, but I imagine that most scientists would conclude that a 0.08 correlation looks like there’s no relationship between the variables.

I had been thinking about this for a while, but it came up again this morning, when I saw an article in the Washington Post, talking about the other ways that presidents can be elected. It’s a great piece, with plenty of good pictures, showing the outcomes of 2000, 2004, 2008, and 2012 if we used different systems.

The article by the Post is interesting, and I strongly suggest people follow the link above and read it through, but it was missing an important ingredient: what would have happened in 2016?

From where I sit, the proportional method described in the article seems like a really smart way to pick presidents. It gives representation to smaller states by the number of electoral votes, but it divides those electoral votes by the proportion of the vote in each state. I ran a couple of scenarios using this system and data from Politico. The proportional allocation of electoral votes isn’t as simple as it sounds at first though, mostly because we can’t have fractions of people, and the percent totals often make the math uneven. Setting that aside, and pretending that we can have fractions of people, using the same allocation of electoral votes per state, but dividing the votes in each state as a function of the percent that went to each candidate, the proportional electoral vote totals would have been 253.4 for Trump and 257.0 for Clinton. Neither would have reached the 270 threshold, putting the decision in the hands of the House of Representatives, which most likely would have voted to elect Trump. If we gave the fractions to the winner of the state, and gave third party candidate votes to the candidate with a plurality, we would have had a tie, 269-269, and, again, the House of Representatives would likely have given it to Trump.

Of course, as much as I don’t like the idea (the reality) of a Trump presidency, I wish it would have gone the other way, but it still seems a lot more fair to me. If the electoral vote of today were balanced by the popular vote of today, the outcomes would cancel each other out, giving us the tie that the proportional electoral vote system gives us. That sits well with me, and seems like a very reasonable way to consider all of the elements at hand.