32

Hypothetically, if all states were to allocate their electors proportionally to the votes in their state, who would have won the last election?

That is, let's say that there was a constitutional amendment in place that, instead of abolishing the electoral college, required that the votes be split in each state (somewhat like ME and NE but simpler.) The simplest system would be that each voter would be voting for a specific elector but let's assume a round-half-even approach to splitting up each states electors gives us a approximation of that. Or alternately, assume each citizen gets to vote for one elector for the congressional district and one senator electoral vote (each state split in half.)

Also, let's ignore that such a system could have a profound impact on voting behavior and assume that the proportions of votes per candidate in each state would be consistent to what we saw in the actual election.

Martin Schröder
  • 3,795
  • 3
  • 30
  • 47
JimmyJames
  • 1,138
  • 1
  • 9
  • 15
  • 1
    In theory, it would better match the popular vote...but given the limited number of electors, you'd could still have huge discrepancies from the actual popular vote given demographics (as small populations states still have an advantage over high population states). –  Nov 11 '16 at 18:40
  • Since the popular vote is so close, the answer would depend on whom the rounding errors favor on average, so it is purely random (and hence of doubtful political interest). That should be easy to compute, however, at least when we have all the votes counted. – Joël Nov 11 '16 at 19:00
  • 5
    Excuse-me, my preceding comment was stupid. I forgot that the number of delegates attributed to a state is not proportional to its population. (Too much politics in three days). Now I understand why the question is interesting: in the difference between the result of the electoral college and a direct national vote, we want to understand separately what is due to the fact that states vote with the "majority takes all" system, and what is due to the number of delegates attributed to each state. Interesting, +1. – Joël Nov 11 '16 at 19:11
  • @Joel It's also not proportional to the number of voters so that could also create a discrepancy. – JimmyJames Nov 11 '16 at 19:13
  • 1
    Even for proportional representation, there are many different ways of allocating posts: d'Hondt (https://en.wikipedia.org/wiki/D%27Hondt_method), Imperiali (https://en.wikipedia.org/wiki/Imperiali_quota), Droop (https://en.wikipedia.org/wiki/Droop_quota) and others. They have different results, which give small differences, so one method should be stablished. – SJuan76 Nov 11 '16 at 19:14
  • Based upon the current voting process or a realm where this was applied. If electoral votes were assigned proportionally, the campaigns would be more apt to spend money in states assured to give a majority of its votes to the opposition, e.g. Republicans would spend money in California to maximize the voter turn out, even though the expected to lose there. – Drunk Cynic Nov 11 '16 at 19:44
  • 1
    @DrunkCynic I'm not sure how that relates to the question. It's not that the electors would be allocated any differently, it just that there would be no winner-takes-all system. – JimmyJames Nov 11 '16 at 19:46
  • 1
    If it is restricted to the results from the previous election, the math is possible. If the scope accounts for the resulting change in campaigning strategy, it becomes a large hypothetical problem that can't be answered. – Drunk Cynic Nov 11 '16 at 19:49
  • Would you consider it inside the scope of you question to have electors assigned in accordance with the House of Representative districts, while the other two are assigned in accordance with the popular vote. Both for a super majority from the popular vote, split for results less than a super majority. – Drunk Cynic Nov 11 '16 at 19:51
  • @DrunkCynic I added another option related to congressional districts. As I look at the data, for the rounding approach to work, some assumptions need to be made around 3 and 4th and 22nd candidates so congresional districts would probably make the most sense. I'm not sure what to do with the other two votes. In my state I only get to vote for one senator but I'm don't think that's necessarily a rule. – JimmyJames Nov 11 '16 at 19:56
  • 1
    I wouldn't be at all surprised if the alternative (the approach currently used by Nebraska and Maine) would make for an even more lopsided electoral outcome due to gerrymandering. – David Hammen Nov 11 '16 at 21:22
  • @DavidHammen Good point. I hadn't considered that. – JimmyJames Nov 11 '16 at 21:33
  • Related article: http://fivethirtyeight.com/features/under-a-new-system-clinton-could-have-won-the-popular-vote-by-5-points-and-still-lost/ – Bobson Feb 07 '17 at 20:10

8 Answers8

29

I calculated the vote allocation using the Webster/Sainte-Laguë method (based on results as of November 9, 2016) applied to each individual state:

  • Clinton 263
  • Trump 262
  • Johnson 10
  • Stein 2
  • McMullin 1

In the spirit of the Electoral College giving less populous states a boost in the vote, I altered the formula to award 2 votes per state to the winner of the popular vote of that state, and the remainder allocated via Webster/Sainte-Laguë:

  • Trump 269
  • Clinton 259
  • Johnson 7
  • Stein 2
  • McMullin 1

For comparison, here I applied Webster/Sainte-Laguë to the entire United States population without splitting them based on state:

  • Clinton 256
  • Trump 255
  • Johnson 17
  • Stein 1
  • McMullin 1
  • Other 8 (these were not separated in the data source)
Brandon Bonds
  • 466
  • 3
  • 6
  • 2
    This is the kind of thing I was looking for. Great answer. – JimmyJames Nov 14 '16 at 18:05
  • 1
    For your second breakdown, "I altered the formula to award 2 votes per state to the winnder of the popular vote" You mean the popular vote in each state? – sturgman Feb 06 '17 at 15:47
  • 1
    @sturgman Yes, thank you. I have added that clarification. – Brandon Bonds Feb 07 '17 at 16:16
  • 1
    Fascinating nobody reaches the 270 to claim a majority (only a plurality). – Jesse C. Slicer Mar 06 '20 at 19:07
  • 1
    @JesseC.Slicer It's one of the features that strongly encourages a national two party system. We had plurality-only results early in the country's history, which were all such a mess that it resulted in constitutional amendments to try to fix it, and political bickering and rage that'd actually compete with modern day behaviors. – zibadawa timmy Oct 17 '20 at 05:05
22

Okay, I have made some excel computations. Giving each candidates a number of delegates proportional to its share of the vote in the state, without rounding (I know, if we're talking of living human delegates, it will be very cruel and unusual to sends 2.34 delegates to Washington, but abstractly why not?), I get 256 delegates for Clinton, 252 for Trump, the rest for small party candidates.

My interpretation: a result quite close to the national popular vote. So the non-proportionality of the number of delegates doesn't seem to be, in this election at least, biased toward any of the candidates. (Sure, voters of Wyoming are over-represented, but so are voters of Vermont.) What advantaged Trump seems to be the majority-take-all system with the way his electors were shared among states, with a short majority in many of the major swing states (nothing new or surprising, in other words).

You can find my excel file here

Joël
  • 1,210
  • 11
  • 19
  • currently 270 electoral votes are required to assume the presidency so that would go into the contingency procedure – JimmyJames Nov 11 '16 at 20:12
  • 1
    But isn't this entirely hypothetical? You do not know how many more votes would have been cast, and who they would have gone to, in states like California and New York, if people had known that their votes would count in the end result. – WS2 Nov 12 '16 at 09:30
  • 1
    @WS2 Yes, but that affects the entire question. Might more people have voted third-party? They would have gotten electors. Might less have, because they knew a major party candidate needed their vote more? Might more people have turned out? Trying to model all that is just asking for trouble. – Bobson Nov 14 '16 at 18:26
  • @Bobson I accept that it affects the entire question. And it indicates the irrelevance of the total number of votes cast. The same sort of thing happens in Britain's parliamentary system. The party which has the highest number of votes is not necessarily the one with the most number of seats. And nowadays with strong third, fourth and fifth parties, having the most number of seats does not even guarantee a party that they will form the government. – WS2 Nov 14 '16 at 20:55
  • 1
    I agree with everything WS2 says. My computation does not show what would have happened if we had another system of vote, there is no way to know (we can make some guesses, some slightly more educated than others, but that's all). It is just (based on the clever idea of the OP) a way to evaluate separately the two respects in which the Electoral College is different from a direct national election, where the candidate with the most votes nationally is elected. Those two ways are: (1) small states have more electoral votes than big states relatively to their population, (2) the countries is... – Joël Nov 14 '16 at 21:39
  • divided into districts (here, states) each of which is attributed to the majority rule. Obviously, one of those features,or both has to have benefited Trump over Clinton since he won the Electoral College but lost the popular vote (again, that doesn't mean that Clinton should have won, or would have won if the election system was national -- it is ridiculous to claim that without further arguments as many do). The computation I have done shows that Trump was not advantaged by the over-representation of small states. He simply, as convention wisdom has it, benefited from higher participation... – Joël Nov 14 '16 at 21:42
  • where participation matter with our system, namely in swing states (including states like Wisconsin that were Swing states but not recognized as such by the Clinton campaign). If anything (but that's pure speculation here), this seems to indicate that Trump's voter were more motivated and that if the election was national, he would have won it as well. Another argument in this direction is the fact tat Republican won handily the national popular vote in the House of Representative elections. – Joël Nov 14 '16 at 21:44
12

As @SJuan76 pointed out, there are multiple algorithms that deal with representing votes proportionally (that don't require dividing the elected in parts). In addition to Joël's work I also did some exceling, which you can see here. I decided to use Hare quota, which I consider most intuitional (it can lead to absurd results, though, such as in the Alabama paradox). The achieved result was:

Clinton 263
Trump 262
Johnson 11
Stein 1
McMullin 1

Disclaimer: Vote totals are taken from Wikipedia article that doesn't give any sources. I presume them to be incomplete, possibilities of inaccuracies include: not all of the precincts reporting, omission of candidates other than top five, vote-ins not counted (might especially impact McMullin).

Edit: corrected wrong input data, I can't believe that no one bothered pointing that the numbers don't add to 538. Trump's count rose from 260.

Szymon
  • 341
  • 2
  • 10
  • if you simply type "election results" into google right now and then click on "Presidential" tab, you can get a state-by-state drill down from Google itself. – Radically Reasonable Nov 12 '16 at 00:24
5

I used the d'Hondt method and got the following results:

  • Clinton: 263
  • Trump: 250
  • Johnson: 18
  • Stein: 5
  • McMullin: 1
  • Other (possibly Castle or de la Fuente): 1

Applying the same method but to each state individually yields the following results:

  • Clinton: 277
  • Trump: 257
  • Johnson: 3
  • McMullin: 1
Tim B
  • 89
  • 4
2

Five Thirty Eight finally weighed in on this; and the result is quite interesting, including and may be especially the what-if scenarios.

"Under A New System, Clinton Could Have Won The Popular Vote By 5 Points And Still Lost" says:

If every state voted like Maine and Nebraska, Trump would have lost 16 electoral votes

... Clinton would actually have won a few more electoral votes in 2016 had all states used proportional allocation by district (though she’d still fall short of 270).

More interestingly, as the article's title suggests, the sensitivity of the result to the national popular vote changes dramatically in the two methods, and in 2016, not in Clinton's favor (though 538 is wise enough, as usual, to note that this is specific to 2016 and you can't and shouldn't draw generic parallels to 2020 etc..) - to wit, if all states were allocating electors proportionally, Clinton not only would have still lost under the existing 2% popular vote margin in her favor, but, would need over 5% national popular vote margin to win under a new system - while only needing 3% margin to win under existing non-proportional elector allocation.

enter image description here

Community
  • 1
user4012
  • 92,336
  • 19
  • 225
  • 386
1

who would have won the last election?

You wwoukdd think that it would mimic the popular votes . But electoral votes have to come in in whole numbers. So you will have issues of rounding up and rounding down.

Putting that aside, 3 million voted out of 100 plus votes separate. The two. That translates into a max difference of 1 5 votes between the two.

Not enough to push any of them over the 271 threshold.

That means we need new rules.

dannyf
  • 2,619
  • 8
  • 13
0

Almost every one of the scenarios has Johnson with enough votes to tip the scales for either Clinton or Trump.

Finally a way for third-parties to come into the political process!

If these numbers were publicized it would be much easier for the changes in how state allot their Electoral College votes.

Brythan
  • 89,627
  • 8
  • 218
  • 324
Sandee Cohen
  • 11
  • What scenarios? – Drunk Cynic Nov 15 '16 at 23:44
  • True, +1. However, the same objections applies to any reasoning using the result of an actual vote under an actual system to an other virtual system: if the virtual system became the actual system, voters and campaigns would know it and act accordingly and the result could be very different. Moreover, In the scenario of a proportional attribution by states of the members of the electoral college, the small parties candidates would be kingmaker, with the perspective of ugly behind-the-scene negotiations between them and major candidates who needs their supports. – Joël Nov 16 '16 at 15:24
-1

Using the same raw vote numbers as Szymon, I rounded so there were no partial delegates.

In that scenario, no one wins outright, with Clinton at 260, Trump at 257, and 21 going to either third party or being uncommitted.

PoloHoleSet
  • 20,854
  • 3
  • 55
  • 89