Distributing Electoral Votes Proportionally
In US presidential elections, nearly every state allocates Electoral Votes on a winner-take-all basis. If a candidate wins a state such as North Carolina by 4% of the vote, that candidate would win all 15 of North Carolina’s Electoral Votes. But what if instead, Electoral Votes were distributed proportionally? If a candidate won 60% of the vote in a state, that candidate would win only 60% of the Electoral Votes, instead of 100%.
Results
Here is what the 2016 election results would look like under a more proportional system. Below, I will go into detail about how arrived at this result.
No candidate won the required 270 votes needed to become president; this race would have been decided by The House of Representatives. It’s also interesting that the “major” third party candidates all managed to snag at least one Electoral Vote (Stein picked up one in California, McMullin in Utah).
A quick sanity check: Overall, Clinton won more votes than Trump, but did not win 50% of the vote. So, it makes sense that she would receive more Electoral Votes than Trump, but less than 270.
If you’d like to jump directly to the data, it’s available as a Google Sheet.
Quirks of the Electoral College
Each state (and D.C.) has a minimum of 3 Electoral Votes. These states with only enough population to send one member to The House of Representatives get a disproportional say in The Electoral College. For example, Delaware has a population of around 950,000 and gets 3 Electoral Votes. Idaho has a population of around 1,683,000 and gets 4 Electoral Votes. Despite having around 1.7 times the population, Idaho only has 25% more Electoral Votes than Delaware.
Because of this disconnect between population and Electoral Votes inherent in the system, we should expect the proportional allocation of Electoral Votes by state by population to be slightly different than the popular vote totals.
Differences from the Current System
I’m not going to make any conclusions as to whether or not this system would be better than the current system, or any other variation. That’s not the point of this blog post. But we can draw some factual conclusions based on this data, compared to the actual results of the 2016 election.
Proportionally allocating Electoral Votes gives a boost to third-party candidates. This is better for third-parties than the current system, but is still less beneficial than going off the popular vote without The Electoral College. For example, in a state with 3 Electoral Votes, a candidate would need to get 33% of the vote to earn 1 Electoral Vote. However, in a state with 10 Electoral Votes, the same candidate would only need 10% of the vote to earn 1 Electoral Vote. This gives third-party candidates a chance in larger states, but they would still face an uphill battle in smaller states.
This proportional allocation system also makes it much more likely that a candidate would not get the majority (270 Electoral Votes) required to win outright, sending the race to The House of Representatives instead. That would have been the case in the 2016 election had a system such as this been in place. For the record, the last time The House of Representatives decided a presidential election was in 1825.
Methodology
During the process of calculating these numbers, there were several methodological decisions I had to make. I’m not claiming to have made all the right ones, but in this section, I will discuss some of the tradeoffs that had to be made, and why I made the choices I did.
Whole Numbers?
The first major question I tacked was: Do I use whole (rounded) numbers of Electoral Votes, or fractional votes? When doing a proportional allocation, the more decimal place precision, the more accurate the numbers. Here are the results if we go to 2 decimal places:
Yet I decided to go with whole numbers instead for 2 reasons:
- It’s not clear that fractional electoral college votes are constitutional
- The “Other” votes
The Constitutional Question
For one thing. The constitution (specifically the 12th amendment) says:
The Electors shall meet in their respective states, and vote by ballot for President and Vice-President, one of whom, at least, shall not be an inhabitant of the same state with themselves; they shall name in their ballots the person voted for as President
The intention was always for The Electoral College to be a body of people, and we cannot have fractional people. Much of the total, especially for the third-party candidates, comes from picking up a few tenths of an Electoral Vote in many states, then aggregating them to a larger final value. But 0.17 persons from Oregon cannot cast a ballot for Jill Stein as required by the 12th amendment. Adopting this approach will likely require a constitutional amendment.
“Other” Votes
When I tabulated these results, I divided the votes into 6 categories. Votes for Clinton, Trump, Johnson, Stein, McMullin, and Other. The Other group is especially problematic here. Other represents other minor third-party candidates, write in candidates, etc. Other does not represent bad ballots or voters who left the ballot blank. However, each state calculates Other in different ways, so the numbers I aggregated into one column are not always consistent from state to state.
Either way, the entire Other category aggregated together never receives enough votes in any state to win enough to round up to 1 Electoral Vote. Yet the totals matter. Add it all up and Other would take home 4.41 Electoral Votes. In any real system, we would need to break down “Other” into each individual candidate, including write in candidates that only receive a couple of votes because they may still get a few tenths of an Electoral Vote at the end.
We still want to include these votes for “Other” candidates because it impacts the overall percentage of the vote for the more popular candidates. Throwing these votes away would be undemocratic. In a whole numbers system, we are not throwing them away, but rather rounding them down to zero. Their impact is still felt in the totals of the major party candidates, and how those values are ultimately rounded up or down.
Rounding
Having decided on using a whole number of Electoral Votes for each state, I now needed to figure out how to round the values. This is actually fairly easy in 2 party races, we round up and down and the results work out. But when you enter in third party candidates, this becomes much harder.
Rounding Border Case #1: Too many Electoral Votes
Let’s take the Michigan. Michigan has 16 Electoral College Votes to give. Here are the fractional results:
Now, under a straightforward rounding system, this would round to:
We have a problem. This gives us 17 Electoral Votes from Michigan, but we were only supposed to have 16. I solved this by defining a rule: If the rounded total would exceed the Electoral Vote total for the state, do not simply round for every candidate. Instead, use whole numbers and award the remaining votes based on the highest fractional part of the raw Electoral Vote number for each candidate.
In the example of Michigan, Trump has the highest fractional part at 0.6, so he gets one Electoral Vote. This brings us to 15 of 16 total Electoral Votes, so the remaining Electoral Vote is given to the next highest fractional part: Johnson with 0.57. The end result is:
This is not the only way to do this type of rounding. For example, we could not round at all and always give the popular vote winner of the state the remaining Electoral Votes. I found this approach however, negatively impacted third party candidates in a way that was excessively unfair. In my reading of the data, Johnson deserved the rounded up 1 vote for Michigan based on his vote totals. But this alternative rounding scheme would have given Trump 9 votes to Johnson’s 0.
As with many methodological choices I’ve made, I think there is room for debate here, but hopefully I’ve explained and justified why I chose this approach, and how I got to the totals I presented above.
Rounding Border Case #2: Too few Electoral Votes
Now let’s look at Arizona and its 11 Electoral Votes:
If we round these numbers, we get:
This only gives us 10 of the 11 Electoral Votes we want from Arizona. In the border case above, we used a method where we distributed the fractional parts in order from largest to smallest, but in this case, that would give Gary Johnson the last Electoral Vote. This is unusual; we generally would not round such that someone with less than 0.5 would end up getting rounded to 1. That arguably overvalues the votes Johnson received from Arizona.
Instead, I defined a rule to handle this scenario: If the rounded Electoral Vote numbers are less than the total Electoral Votes for the state, give the remaining Electoral Vote to the winner of the popular vote for that state. Under this approach Electoral Votes for Arizona are distributed as follows:
Of course, there are downsides to this approach. I have effectively rounded up from 5.35 to 6, when just paragraphs ago I argued against rounding up from 0.45 to 1. But I also think it is reasonable to give Trump more Electoral Votes in a state where he won, a sort of popular vote bonus where there was an extra Electoral Vote to give. I didn’t take this approach earlier because I though it unfairly penalized third party candidates. But I would argue that is not the case here, as Johnson did not earn enough votes to win an Electoral Vote under a classic rounding system.
As with many methodological choices I’ve made, I think there is room for debate here, but hopefully I’ve explained and justified why I chose this approach, and how I got to the totals I presented above.
Rounding Roundup
The two border cases above are practical cases I had to deal with to make the 2016 election data work with in a proportional Electoral Vote distribution system. That doesn’t mean there won’t be other pitfalls, or that my solutions would make sense for all data sets.
What I hope you take away from the above analysis is that some of these problems do not have straightforward solutions, and we need to consider that if we ever want to seriously discuss a system like this as a matter of policy.
Equal Vote
This blog is not a response to the Equal Votes movement recently launched by Lawrence Lessig. I’ve been working on this for some time. However, the movement is worth mentioning as it advocates a system very much like the one I explore here. Hopefully this blog can add some context on the details of how such a system might work.
Final Notes
This was something I did for fun. If I made any mistakes in the data or in my calculations, please let me know, and I will correct them.
