How I Would Reform the Electoral College

March 1, 2020

An election year is here again. And, given the political leanings of the vast majority of my friends, so too is the trauma of the Electoral College. In 2016, Hillary Clinton won the popular vote by 2.845 million votes (2.1% of total turnout), but lost the Electoral College thanks largely to Pennsylvania, Wisconsin, and my home state Michigan. The trauma is so deep, that most of the people I know voting in the Democratic primary today are voting for the candidate they think is more likely to win the general election in just the three states that tipped the election to Trump last time around. Never mind policy, personality, or winning more votes nationally.

Since 2016, the liberal parts of my social media echo chamber have been complaining endlessly about the Electoral College, and rightfully so. The fact that Democrats won the vote in 2016, but not the election, does feel rather undemocratic. But what I don’t understand is the proposed solution I see most often - the National Popular Vote Bill.

Essentially, the National Popular Vote Bill forces all states to allocate all of their electoral votes to the winner of the national popular vote as opposed to their state’s popular vote 1. This would have resulted in a President Clinton instead of President Trump in 2016 and a President Gore instead of a President Bush in 2000. I get the appeal, but I have a number of issues with it.

First, I think it’s unlikely to ever get enough states on board that it will take effect. This is because there are two groups of states that have no incentive to sign onto it. The first group is rural states that have a disproportionate number of votes in the Electoral College, like Wyoming. The Electoral College gives those states more power than they would otherwise have, but the National Popular Vote Bill would negate that additional power. The second group is swing states, like Michigan. Swing states have an even more disproportionate influence on the presidency than rural states do, because the election largely takes place in less than 10 of them. They can turn this power into preferential treatment from national parties and presidents. Why would they give that up? To this point, we can already see the fate of the bill by looking at states that have passed it - with very few exceptions, they’re all non-competitive states with large urban populations. As long rural states and swings states have 270 electoral votes between them, I don’t think the National Popular Vote Bill will gain any traction.

Second, I think it runs against the intent of the U.S. Constitution 2. Since the founding of the country, U.S. elections have been run independently from state to state. While it’s certainly open for debate whether this is a feature or a bug of the U.S. electoral system, I think it’s a feature. Because we have, essentially, 50 elections in November to determine who becomes the President, that makes it very difficult for either foreign powers or, depending on your level of paranoia, national political parties to interfere with or skew elections in a way that will drastically change the result. If one node in the election network goes down, we still have 49 others that should be fine 3. The National Popular Vote Bill undermines this by collapsing all state elections into one national elections. In 2016, for example, any election mistake or interference that altered 3 million votes anywhere would have been enough for President Trump to win the election without the popular vote. Given the rather weak election security practices in many states, I don’t think this is very hard to imagine.

Finally, I find the whole thing to be hardly more democratic than the Electoral College. The Electoral College set out to give all states an independent voice on who becomes the President, and the National Popular Vote Bill erodes that independence. Moreover, it still fails to accurately reflect the heterogeneity of the state-by-state vote. The map of a national popular vote election doesn’t exactly look more democratic:

Moreover, by automatically handing the presidential election to the popular vote winner, it all but guarantees that no third party will make a serious presidential run. Long term, I think this tends to make partisanship worse, as neither party is incentivized to appeal to more than half the country. And because there will never be a serious third party run, voters are incentivized not to vote for a the politician or party which best represents their beliefs, but rather the one that they’d rather not lose. Thus, the country is trapped in the same circular, but rational, logic that keeps the power of the two political parties entrenched.

Rather than blame the Electoral College, or force all states to conform to the national popular vote, electoral reform should probably be a little more nuanced. The real problem for the our elections is the assumption that electoral votes should be winner-take-all. It wasn’t always this way, and there’s nothing in the U.S. Constitution that mandates it. It’s just something that, as far as I can tell, became a cultural standard that few people have questioned since.

Ways of Allocating Electoral Votes

Rather than use a winner-take-all system, it makes more sense to allocate electoral votes proportionally. For example, Michigan in 2016 would have allotted 9 of 16 electoral votes to President Trump and the other 7 to Hillary Clinton. This is how most Democratic primaries are run, after all. What would that look like?

Allocating a discrete number of electoral votes based on vote share, a continuous variable, is actually more interesting and nuanced than it first seems. For example, consider a hypothetical state that has 11 electoral votes., split between a Republican candidate, a Democratic candidate, and a third-party candidate. In this system, it’s much more plausible the third party candidate gets votes. Consider the results in the following table:

Candidate Vote Share Electoral Votes
Republican 48% 5.28
Democrat 40% 4.4
Third-Party 12% 1.32

In proportional allocation, 48% of 11 votes rounds to 5 votes for the Republican, 40% rounds to 4 vote for the Democrat, and 12% rounds to 1 vote for the third party candidate. This only adds up to 10 total though, so we have one more vote to allocate.

We could avoid this issue by always rounding up, so that the Republican candidate rounds up to 6 votes and the Democrat to 5 votes, but that isn’t exactly fair - the third-party candidate doesn’t get the one vote they deserve because the 11 votes are spent by the time we get to allocating their votes.

The obvious solution is to round to the nearest delegate, and then if there are some left over afterward, go through each candidate again and allocate the remaining votes, rounding up. This results in the Republican getting 6 votes, the Democrat getting 4 votes, and the third-party candidate getting 1 vote. While this is the most fair, it does demonstrate a behaviour of the electoral college you can’t get around - allocating a relatively small integer number of candidates based on the vote share leads to oddities. A popular/electoral vote split is possible even without winner-take-all just based on integer rounding.

The other interesting thing about proportional allocation is that it establishes a very clear minimum vote share required to qualify for an electoral vote, much like the Democratic primary, which has a 15% threshold. In this case, the threshold is 1/2*N, where N is the number of electoral votes for that state. In a small state with three electoral votes, a candidate needs to get 1/6th of the vote. But in a state like California, with 55 delegates, a candidate only needs to get 0.9% of the vote to qualify.

Even with proportional allocation, smaller states still have a disproportionate number of votes. Because of the lower floor needed to win delegates in populous states, it could give an even wider advantage to the candidate that wins small states, because it harder for the minority candidate to get votes in small states than large ones. It’s also likely a field of battleground states will still emerge as campaigns do the math on the percentage of the vote they need to shift to their column to get the maximum number of votes possible. The battleground will be wider, but there are still states that candidates are not incentivized to visit. For example, Democrats in 2020 would have no real reason to visit the 3-vote states in the West (Wyoming, Montana, and the Dakotas) because any vote share between 16.7% and 50% will get them one electoral vote.

Nonetheless, allocating electoral votes proportionally does make the electoral college more representative. But it also makes it less likely that a major party candidate gets a majority of the electoral votes. For example, Bill Clinton won 69% (370) of the electoral votes in 1992 despite only getting 43% of the popular vote. In proportional allocation, he would have only gotten 234 electoral votes, and failed to clear the 270 needed to win. This means that Ross Perot would have clearly played spoiler in 1992 and the election would have gone to the House of Representatives. This demonstrates the increased viability of third party candidates in a proportional system.

Unfortunately, that means some kind of Proportional Allocation Vote Bill would probably run into the same fate as the National Popular Vote Bill. Because both major political parties are better off without third parties, they’re never going to change the Electoral College to a system that makes them more viable.

But I think there’s a compromise that would make the Electoral College more representative, give third parties more of a chance, but still have a strong enough institutional bias toward the two major parties to the point they’d perhaps be willing to consider it. I call it Hybrid Allocation.

The number of electoral votes each state gets is equal to the number of seats they have in Congress. That’s 2 for every state from the Senate, plus their number of seats in the House of Representatives. Senate seats are always statewide, so why not make the 2 electoral votes from the Senate be winner-take-all, and allocate the electoral votes from the House of Representatives proportionally? 4 This decreases the number of electoral votes effectively in play for third-party candidates by 2 in each state 5, therefore also increasing the vote share required to qualify for electoral votes. It also skews the electoral votes more in the favour of the state’s popular vote winner, which is almost always going to be a major party candidate.

But compared to winner-take-all, it still increases the electoral battleground beyond the half dozen or so swing states, makes third parties more viable than they are now, and makes the Electoral College more closely represent both the national popular vote but the geographic distribution of the vote.

Running Alternative Histories

So what would this Hybrid scheme look like in practice? To see, I grabbed state-level election data for every Presidential election since 1976 from the MIT Election Data + Science Lab. Because it has results at the state level, it’s pretty easy to run the elections with different allocation schemes and see what the results would have been 6. You can see the results for each election below.


This is a good demonstration of the interesting dicontinuties of the Hybrid allocation. Becuase there are fewer votes to allocate proportionally in small states, President Ford got 4 extra from states like New Hampshire, Nevada, Wyoming, North Dakota, and Alaska. While he still wouldn’t have won the election, it likely would have been a long election night, with the margin in the West Coast states being crucial.


This election was not close, no matter how it’s spun. However, it is notable that third parties combined for 31 electoral votes in the proportional scheme and 26 in the hybrid scheme, which is the most of any election that didn’t include Ross Perot. That suggests that the election could have played out much differently if states didn’t use the winner-take-all system, which strongly incentivized voters to vote for one of the two major candidates. Without the incentive to vote for a major party, is it possible President Raegan would have only won a plurality instead of majority?