By Paul Butler – November 6, 2018
Twice in recent memory, the Electoral College has picked a president who did not win the popular vote. Each result rekindled the ongoing debate on the merits of replacing the electoral college.
The central contention in that debate is that the system gives voters in less populous states a disproportionately large share of the electoral votes. But this is only one face of the Electoral College. As we will see, even if we could retroactively redistribute electoral votes by population, the 2016 outcome would remain unchanged!
Before we do that, though, let’s look back to an election where the electoral vote distribution did matter: the 2000 presidential election. This chart shows how many electoral votes each state had per million residents back then.
This disproportionate representation is a direct result of how electoral votes are allocated: one for every senator (each state has two), plus one for every congressperson (who are allocated to states by population.) The remaining three electoral votes go to the District of Colombia.
There are other ways you could imagine allocating electoral votes, for example, proportionally to the population of each state. We could even imagine, after the polls close, reallocating the electoral votes according to the voter turnout in each state, in order that each ballot-box vote carries the same weight in the Electoral College. To do this right, let’s assume we can reallocate the votes fractionally; otherwise some states will be short-changed due to rounding error.
Each bar below represents a state, with width proportional to its share of the electoral votes in 2000. The color represents the party that won the state in that year. Use the toggle below the graph to see the how the results change when we instead weight states votes purely by population or ballot-box votes.
Note: To avoid additional complexity I’ve ignored third-party votes throughout this article. This changes the margins, but does not affect who wins.
As you can see, reweighting the states by either population or total votes is enough to change the outcome of the election, provided that voters cast the same votes.
I should emphasize here that it’s impossible to actually know who would have won the election under a different system, since we don’t know how the campaigns or voters would have acted under other voting systems. It’s a different thing to say that a candidate won the popular vote, than it is to say a candidate would have been elected under a popular vote system. My intent here is to understand the dynamics at play when the outcomes of those systems differ.
How does tabulating the votes with different state weights affect the 2016 election results? Here’s the same visualization for 2016. Note that one of the bars has both red and blue: this reflects Maine’s split allocation of its four electoral votes in 2016.
As in 2000, the Democrat (Clinton) received more votes in 2016. This time, though, the outcome is unchanged by reweighting states’ impact by population or number of votes. All this despite the fact that the distribution of electoral votes hasn’t become notably more flat:
If the allocation of electoral votes to states didn’t affect the outcome in 2016, what did?
This brings us to the second way that the Electoral College distorts the vote: winner-take-all effects.
Under the Electoral College, it’s up to each state’s constitution to decide how they allocate their electoral votes based on ballot-box results. Over time, most states (excepting Maine and Nebraska) have decided to allocate all of their electoral votes to the candidate who won the plurality in that state. In this way, each state maximizes its influence on the outcome of the election.
What happens when we give each state the same relative weight as we do in the electoral college, but this time allocate that weight fractionally according to the share of votes actually cast in that state?
We can see that in 2016, Democrats would have won if states were not winner-take-all, even if states continued to have the disproportional weights they do in the electoral college.
Contrary to the popular narrative, Rebublicans won the electoral college not because it underweighted the votes of the Democratic majorities in high-population states, but because it diminished the votes of Democrats in the states where they represented a minority of the vote.
For comparison, let’s see how the 2000 results would have reacted to this system of fractional allocation of votes:
As you can see, the race gets tighter, but still ultimately goes to the Republican.
So, to recap, there are two ways in which the electoral college distorts the vote relative to the popular vote:
We have seen that effect #1 was alone responsible for the discrepancy between electoral and popular vote in 2000, and #2 gets sole credit for 2016.
The most-repeated defenses of the Electoral College system is probably that it prevents a so-called “tyranny of the majority,” where low-population states are overruled by the votes of populous states. There is some truth to this (effect #1), but this argument carries a sort of fractal irony: tyrrany of the majority on the national level is exchanged for tyrrany of the majority on an individual level, in which the political minority in every state has their vote effectively flipped to that of the majority (effect #2.)
If there’s anything to take away from all this, it’s that making the debate over the Electoral College a partisan tug-of-war over the power it gives each state misses the broader picture. This should come as bad news to anyone who hopes that replacing the Electoral College will have a lasting systematic benefit for their party, but good news to electoral reformers who hope to make a bipartisan appeal to the benefits of a more representational system.
Data in this article was obtained from the Federal Election Commission (2000, 2016); the Federal Register (2000, 2016), and the US Census Bureau.