This is the 2024 version of this that has been updated to include 2020 results. For reasons to be explained later I also added additional results going back to 1976. In the places where 6 results are mentioned this is the original 2000-2016 election range plus the 2020 update. In places I have decided to run things with all three data ranges. The original, the original and 2024, and all years from 1976 to 2024. Some of the original text has not been changed.
Elizabeth Warren is a presidential candidate of the Democrat party. One of the issues that she is running on is the elimination of the electoral college in favor of popular vote for presidential elections. It is an interesting idea. Hilary Clinton lost the last presidential election in the electoral college despite a popular vote majority. Al Gore lost the 2000 presidential election also with a popular vote majority. Actually, he lost the election by a Supreme Court decision mostly made by Ronald Reagan appointed justices.
On the face of it nothing could really be fairer than popular vote. With a two party system it should be one person one vote, right? Especially as far as the Democrats are concerned since they won the popular vote in 4 of the last 5 elections while only putting one president into office.
Why the Electoral college anyhow? The issue of eliminating the electoral college was recently discussed on the MSNBC Hardball program. One guest repeated that the electoral college was set up to protect small states. While my thoughts had been that this might effectively be what it was accomplishing, my memory was that this was not the reason it was created. I recalled that it was enacted to facilitate elections when vote tallies weren’t as easy to put together as they are in more modern times. I was recalling incorrectly, the Hardball guest was correct, as shown in the above (Why the Electoral college) link.
Another reason for establishing the Electoral College given in the above linked article was…
The founding fathers were afraid of direct election to the Presidency. They feared a tyrant could manipulate public opinion and come to power.
This seems to have somewhat backfired in the last presidential election.
The basis of my contention that switching to popular vote is not a good idea, is that the current two party system considerably aligns along population lines. You hear this frequently, and I have myself mentioned it previously here and there. The Democrats tend to win in cities and large urban areas while republicans control larger space in more sparsley populated areas. The Democrats have what amounts to an edge in strictly popular vote elections.
Let’s see if anything seems to corroborate that population is reflected in presidential elections. First, consider the number of states that the Democrats won in presidential elections since 1976.
## [1] "1976 Carter vs. Ford 24"## [1] "1980 Carter vs. Reagan 7"## [1] "1984 Mondale vs. Reagan 2"## [1] "1988 Dukakis vs. George H.W. Bush 11"## [1] "1992 Bill Clinton vs. George H.W. Bush 33"## [1] "1996 Bill Clinton vs. Dole 32"## [1] "2000 Gore vs. Bush 21"## [1] "2004 Kerry vs. Bush 20"## [1] "2008 Obama vs. McCain 29"## [1] "2012 Obama vs. Romney 27"## [1] "2016 Hilary Clinton vs. Trump 21"## [1] "2020 Biden vs. Trump 26"## [1] ""## [1] "Total: Dem 253 / Rep 359"Notice that while Gore and Clinton received more popular votes they still won fewer states. I think this is an indicator that the states they won were more populous. The Obama elections were different, as a Democrat he still won more states than his opponents. I consider this a good exception for him and not a refutation of the Democrat’s states being more populous.
What if we just total up the wins across all six elections?
## [1] "For all six elections..."## [1] "Dem wins 144"## [1] "Rep wins 162"Here we see that despite the Obama wins the Republicans have still won more states during the last six elections. I’ve retained this with just the last elections as it looks slightly better for the democrats if you eliminate the 80’s massacres.
I now have the census populations for every decade from 1970 to 2020 so we can summarize this as…
## Year D_States D_Elec R_States R_Elec D_Popular R_Popular D_Pop R_Pop
## 1 1976 24 297 27 240 40831881 39148634 4153218 4698836
## 2 1980 7 49 44 489 35480948 43897869 2995429 4672223
## 3 1984 2 13 49 525 36777147 54455074 2335154 4528071
## 4 1988 11 112 40 426 41807979 48881155 2065831 5095542
## 5 1992 33 370 18 168 44927432 39052486 5335913 4034708
## 6 1996 32 379 19 159 47402357 39321374 5650066 3574094
## 7 2000 21 266 30 271 50999897 50455982 6093654 5115172
## 8 2004 20 251 31 286 59028109 62028285 5805534 5332620
## 9 2008 29 364 22 174 69498516 59948323 6252193 4550378
## 10 2012 27 332 24 206 65915794 60933504 7080410 4898936
## 11 2016 21 233 30 305 65853516 62984825 6420318 5797296
## 12 2020 26 306 25 232 81268908 74216146 7032768 5943892D_States / R_States = Number of states won.
D_Elec / R_Elec = Number of electoral votes.
D_Popular / R_Popular = Total popular votes.
D_Pop / R_Pop = Average population of states won.
The above somewhat more completely summarizes information from the 1976-2020 elections. In doing this previously I put some effort into suggesting that democrats won more populous states. For this time with the 1970-2020 census data the last two columns directly indicate the average population of the states won. The Democrats have in fact led this since Bill Clinton in 1992.
I tend to think it has always been the case that the Democrats win the larger population states but going back far enough this wasn’t the case. Carter’s 1976 victory is strange in that he won fewer states with lower average populations but still won both popular vote and in the Electoral College. The rest of the 80’s elections were all lopsided Republican wins that also included the more populous states. Again, this turned around with the Clinton outcome in 1992 where he won with 33 states including the more higher average populated ones. This has continued for the Democrats ever since. Actually, since then the Republicans have only won popular vote once and never again had the higher average popular vote states. Look it up.
For the last 2020 results the average populations of the states won by Democrats across the elections versus the Republican average populations in that year.
## [1] "Average population of states the Democrats won 7,322,408"## [1] "Average population of states the Republicans won 5,642,667"I did this first before getting the more complete historical census information. The numbers differ somewhat. If you do look it up maybe you can let me know which is right. As the Dire Straits song says “Two men say they’re Jesus / One of them must be wrong.” It seems one of the sets of numbers must be off. Industrial Disease - Dire Straits
I tried to fact check some of my other numbers to see if they appear correct. Some of them seem to exactly match Wikipedia reported election results, some don’t. I should try to track down the reason. I had hoped to try and make sure this go round was a little more accurate. But I don’t think I’m going to put in the time required right now. I don’t know how long that would take and I have other things that are of interest to me to look at.
One interesting thing I came across doing this was that, according to Wikipedia, both Trump and Hilary Clinton lost ‘faithless’ electors. Click the electors popup footnote for the Wikipedia 2016 election results. This is the first I remember seeing or hearing anything on that. I wonder if that inspired some of the fake elector controversy in the last election.
Which also made me sort of wonder what happened to the aide Ron Johnson threw under the bus for trying to get a list of fake electors to Mike Pence for my own state of Wisconsin. But again not curious enough to try and track it any further.
So, anyhow, the state populations for Democrat victories are definitely higher in recent times.
Now, consider the average margin of victory for the candidates in the 2020 election.
## [1] "Avg Democrat margin of victory 579,482"## [1] "Avg Republican margin of victory 320,551"Margin of victory is about 50% higher for the Democrats. Even more pronounced than population differences. I will probably consider this in somewhat more detail in a separate document later.
The simulations that are coming up need the standard deviation of the vote totals for each state across the years. Sort of a measure of the spread of the votes. Bigger states should have a bigger vote spread. The average of these spreads should again be yet another indicator of which party has the larger, average, states. I calculate this for three date ranges. Original (2000-2016), then 2020 only, finally from 1976 to 2020.
## [1] "Average dem spread 165,068"
## [1] "Average rep spread 117,535"## [1] "Year 2020 spreads"## [1] "Average dem spread 218,486"
## [1] "Average rep spread 161,118"## [1] "All Year spreads"## [1] "Average dem spread 303,918"
## [1] "Average rep spread 249,928"Hopefully, by now you are convinced that the Democrats tend to win fewer but larger states by larger margins while the Replubicans win more but smaller states with lower margins.
The fact that the Democrats win states with larger populations should be reflected in the Electoral College votes that they have. If the vote margins for these states are proportionally higher or lower than reflected by the population then the Electoral College votes may not reflect that. To try and get a better idea of how this interplay might work out I will attempt to run some simulations.
It seems like saying that the Democrats vote margins outperform their Electoral College population allotment is the same as saying that they have better voter turnout. I did some searching but didn’t find any good source for data on voter turnout by party for the presidential elections. Maybe in a future revision of this, if I have one, I will include that.
[2024 NOTE] I’m not sure the above is entirely correct. A better vote margin might be demographics. For example, it might mean an older state with more eligible voters. Rather than a younger, growing state with more people below voting age. If you’re curious, searching again I did come across this Voter Turnout Charts, which apeears to take into account the Voting Age Population (VAP). I even see a Wikipedia now, Voter turnout in United States presidential elections. Although only includes 2020 by state. Yes, Wisconsin rocks the vote. Including Milwaukee.
What happens if we extend these results to consider the question of whether the electoral college or popular vote should be used?
First, the caveats. We are talking about events that occur only once every four years. I am basing what I am doing on only the last five of those. This is the classic computer cliche of ‘insufficient data’. A handful of atypical events to try to determine something typical about. These elections could be argued to be more atypical than most. You had Gore-Bush with hanging chads, Florida and Nader, and finally the Supreme Court. The Obama elections were frequently described as historic. The last electiono had Russian hackers, the Comey letter, and again a win with less popular vote.
The approach I am taking is to assume a ‘normal’ - bell shaped curve - distribution of state votes. To the best of my admittedly limited knowledge this is a valid enough approach for this sort of thing. However, again to my understanding, this is probably not based on enough data to draw confident conclusions from. Based on these distributions I use a Monte Carlo type approach (sort of controlled randomize the parameters and do a lot of trials) and simply hold 10,000 elections. But to repeat one more time, all 10,000 of these are all still based on only five real elections.
[2024 NOTE] As mentioned I am now considering more election results than the five original. Sometimes six, sometimes twelve.
Spoiler alert, keep in mind that the Democrats actually won four out of five of these elections on popular vote.
## Iteration 1000
## Iteration 2000
## Iteration 3000
## Iteration 4000
## Iteration 5000
## Iteration 6000
## Iteration 7000
## Iteration 8000
## Iteration 9000
## Iteration 10000
## [1] "Elections without an electoral college winner 109"
## [1] "Dem: wins both 7463"
## [1] "Rep: wins both 572"
## [1] ""
## [1] "If split decision decided by Electoral College"
## [1] "Dem: wins EC 335"
## [1] "Rep: wins EC 1521"
## [1] "Rep: percentage this occurs 15.21%"
## [1] ""
## [1] "If split decision decided by popular vote"
## [1] "Dem: wins pop 1521"
## [1] "Rep: wins pop 335"
## [1] "If no EC just decided by popular vote"
## [1] "Dem: wins pop 9081"
## [1] "Dem: percentage this occurs 90.81%"
## [1] "Rep: wins pop 919"## Iteration 1000
## Iteration 2000
## Iteration 3000
## Iteration 4000
## Iteration 5000
## Iteration 6000
## Iteration 7000
## Iteration 8000
## Iteration 9000
## Iteration 10000
## [1] "Elections without an electoral college winner 87"
## [1] "Dem: wins both 7186"
## [1] "Rep: wins both 764"
## [1] ""
## [1] "If split decision decided by Electoral College"
## [1] "Dem: wins EC 405"
## [1] "Rep: wins EC 1558"
## [1] "Rep: percentage this occurs 15.58%"
## [1] ""
## [1] "If split decision decided by popular vote"
## [1] "Dem: wins pop 1558"
## [1] "Rep: wins pop 405"
## [1] "If no EC just decided by popular vote"
## [1] "Dem: wins pop 8815"
## [1] "Dem: percentage this occurs 88.15%"
## [1] "Rep: wins pop 1185"## Iteration 1000
## Iteration 2000
## Iteration 3000
## Iteration 4000
## Iteration 5000
## Iteration 6000
## Iteration 7000
## Iteration 8000
## Iteration 9000
## Iteration 10000
## [1] "Elections without an electoral college winner 78"
## [1] "Dem: wins both 3743"
## [1] "Rep: wins both 4115"
## [1] ""
## [1] "If split decision decided by Electoral College"
## [1] "Dem: wins EC 1042"
## [1] "Rep: wins EC 1022"
## [1] "Rep: percentage this occurs 10.22%"
## [1] ""
## [1] "If split decision decided by popular vote"
## [1] "Dem: wins pop 1022"
## [1] "Rep: wins pop 1042"
## [1] "If no EC just decided by popular vote"
## [1] "Dem: wins pop 4801"
## [1] "Dem: percentage this occurs 48.01%"
## [1] "Rep: wins pop 5199"I had at first intuitively expected that including the last 2020 election results would result in less chance that there would be a victory in the electoral college despite trailing in popular vote. Since the 2020 result had been a clear victory in both categories. This was not evident in the simulation. The chances for a Republican win in my simulation still remain at about 15%-16%. So, I thought I’ll add still more years where this doesn’t happen to demonstrate that this changes the probability as would be expected.
I decided the easiest way to get the data was from my Election Games app. I put in some effort for a couple election cycles to have election data. I think just about the time that I finally dropped support I was looking at getting that data into an SQL database for anyone who wanted to do their own queries against it. For anyone interested, nostalgic maybe, a couple years ago I did use the current java application packager to have current working versions (cross-platform) of the application. Election Games 2012.
So, anyhow, there are now also results going back to 1976. What did this do this do for the chances of a Electoral College win with popular vote loss? Originally nothing. This was due to a bug in what I was doing. Although adding the 2020 data didn’t seem to move the needle at all on the likelihood of a Electoral College only win, going back to 1976 does. The chances go from 16%-17% to about 10%. When I still had the bug I thought looking at swing states might provide some indication as to what was going on with what seemed to be a fixed percentage chance for the Electoral College only win.
Notice in what follows that what is shown for the full date range differs considerably. This appears to indicate a definite shift in the electorate. I believe this makes it less useful for what is happening now. The following discussion will mainly concern the more recent date ranges.
I believe including swing states is still of some use. They have direct bearing on any presidential election. They are the easiest electoral votes to pick up without completely reversing a state that always votes against your party. Given the states that do remain almost always constant, the swing states are what determine the election outcome.
## State EV dwins_per rwins_per
## 46 VA 13 52.25 47.75
## 10 FL 29 47.73 52.27
## 36 OH 18 43.43 56.57
## 13 IA 6 56.64 43.36
## 6 CO 9 57.14 42.86
## 34 NV 6 60.68 39.32
## 28 NC 15 34.42 65.58
## 31 NH 4 65.79 34.21
## 4 AZ 11 28.34 71.66
## 49 WI 10 73.17 26.83## State EV dwins_per rwins_per
## 10 FL 29 47.52 52.48
## 13 IA 6 46.48 53.52
## 34 NV 6 56.84 43.16
## 46 VA 13 57.92 42.08
## 6 CO 9 61.11 38.89
## 4 AZ 11 38.67 61.33
## 28 NC 15 38.50 61.50
## 36 OH 18 37.34 62.66
## 39 PA 20 67.24 32.76
## 31 NH 4 67.31 32.69A couple other states frequently mentioned as “battleground” or swing (might already be included from last random run)…
Michigan
## State EV dwins_per rwins_per
## 23 MI 16 79.43 20.57## [1] "Swing # 15"I remember Nicole Wallace MSNBC saying Biden was winning on the ‘blue wall’, aka ‘rust belt’?
Pennsylvania
## State EV dwins_per rwins_per
## 39 PA 20 67.24 32.76## [1] "Swing # 9"Formerly(?) included in the blue wall
Georgia
## State EV dwins_per rwins_per
## 11 GA 16 32.33 67.67## [1] "Swing # 11"Still controversial from last election.
Wisconsin
## State EV dwins_per rwins_per
## 49 WI 10 68.33 31.67## [1] "Swing # 12"Another brick in the wall.
Another Brick in the Wall - Pink Floyd
All years data from 1976 on
## State EV dwins_per rwins_per
## 3 AR 6 49.95 50.05
## 48 WA 12 49.95 50.05
## 41 SC 9 49.89 50.11
## 39 PA 20 49.85 50.15
## 15 IL 20 49.84 50.16
## 43 TN 11 50.18 49.82
## 13 IA 6 49.77 50.23
## 9 DE 3 50.24 49.76
## 23 MI 16 49.68 50.32
## 34 NV 6 50.39 49.61The results are very similar, usually there are only some changes in ordering, the percentages also usually differ slightly. This does suggest a consistency, at least in more recent elections, that might be what carries over to the frequency of the electoral college only wins. Not that many states are really close, by the end of the top 10 it’s already about 3/4 in favor of the states consistently going one direction only. Although there have been notable reversals there recently, Arizona for one, Georgia, Pennsylvania.
It occurred to me that for the times that Electoral only wins happen that looking at which states flipped from the usual might also be somehow consistent. I use the orginal data to set a base line for which way the states normally go. Then when a result is an electoral college only win I look at what states flipped.
States that flipped in Electoral College only victories for the prior data with 2020
## State EV flipped bush00 trump16
## 13 IA 6 1144 - +
## 46 VA 13 1024 + -
## 34 NV 6 949 + -
## 39 PA 20 892 - +
## 6 CO 9 883 + -
## 49 WI 10 779 - +
## 31 NH 4 697 + -
## 24 MN 10 604 - -
## 23 MI 16 588 - +
## 4 AZ 11 575 + +States that flipped in Electoral College only victories 1976 on
## State EV flipped bush00 trump16
## 31 NH 4 1108 + -
## 33 NM 5 1080 - -
## 13 IA 6 1069 - +
## 6 CO 9 1068 + -
## 22 ME 4 1068 - -
## 49 WI 10 1067 - +
## 50 WV 5 1061 + +
## 43 TN 11 1053 + +
## 32 NJ 14 1049 - -
## 47 VT 3 1048 - -For the ‘bush00’ and ‘trump16’ (Electoral only win) columns a ‘+’ indicates a state they won, a ‘-’ a state they didn’t. The 2020 results tend to show swing states included. The all years since ’76 data shows many flipped and not so many swing states. Although the current run does show that both Bush and Trump won both Ohio and Florida. Florida of course a bit controversial in the Bush election. (More nostalgia - Election 2000 Florida, Florida, Florida). My final run to generate a web page may not show this.
The more recent results do usually include swing states but usually only match 5 or 6 Bush or Trump wins. Probably swing state wins are required but I don’t think the simulations indicate some clear prerequisites for this type of victory. There are probably many ways this can happen with the fifty states.
[2024 Opinion] It is my opinion that these considerations were the main criteria in the Trump selection of JD Vance as his running mate. The choice was based on state for election day. Locking down Ohio which had been a swing state although more recently has seemed to trend Republican. Florida with Rubio would, of course, also been a good swing state choice. Vance might be seen to be a more MAGA choice. Having Rubio on the shortlist might still help him with the Florida Cuban voters, so still getting a boost in that state. I don’t think the North Dakota Governor was a serious option with that states electoral votes. I believe that was just a shoutout to his rural Republican base. Ohio - Crosby, Stills, Nash & Young
As a quick aside. During every election coverage they mention the possibility that there might not be a decisive result in the Electoral College. The election could end up in an Electoral vote tie. This has never happened but simulations seem to show that the odds against it aren’t astronomical. In my testing the “Elections without an electoral college winner” value is usually around 100 or about 1% of the time.
For comparison of election systems we can ignore elections where there is both an Electoral College and popular vote winner. The system selected doesn’t matter. But, in this case I think how many of these elections are won by Democrats rather than Republicans needs to be mentioned. Based on the last five elections it is very difficult for a Republican candidate to get any kind of popular vote win. The results show that they win both categories much less often.
Relative to the others maybe Obama’s victories were resounding enough to severely skew these results. That can be checked…
## [1] "sum of Obama's margins of victory (popular) 14532483"## [1] "sum of Republican margins of victory (popular) in the rest 400419"## [1] "Obama's national victory margins are 36 times the Republican"Remember also, that the Democrat’s won on popular vote in two out of three of the other elections besides Obama’s. The one Republican popular vote win was the incumbent Bush’s victory following 9-11. So, based on data generated from these results, my simulations are bound to considerably favor the Democrats in popular vote. However, I think that the results showing that the Democrats also win so many of these simulated elections in the Electoral College is an indicator that there is no significant Electoral College bias against them.
My simulations have shown about a 20% chance that there won’t be a clear winner in both categories. Historically, this is rather high. Given 2 out of 5, 40%, of the last actual elections ended this way, it is rather lower. This may reflect a parity in the current electorate making closer elections more likely?
The system selected does matter in the “split decision” elections. The results always indicate that if they are decided on popular vote the Democrats win a big majority. If they are decided by Electoral Vote then the Republicans win a large majority of the elections. A possible indicator that currently with the Electoral College Republicans might tend to win close elections. When they already seem to have an advantage in the majority of elections that aren’t so close, would it be better to switch to popular vote giving Democrats an advantage here?
[2024 Update] I had some popup notification on my phone from somewhere that Biden would win the popular vote but lose the upcoming election in the Electoral College. I decided to look at updating this document to include the 2020 presidential results. I have ended up putting a little more time into this than originally planned. There is additional data. I have done computations against the different election ranges. I show percentages. I have included some new changes related to swing states. I have also looked at what states ‘flip’ when the simulation is an Electoral College only win. I even threw in some splashes of color.
I think the simulations now give strong indications that there have been changes to the electorate since 1976. Although, it is probably still shifting 2020 might of suggested not so much. The simulations show more clearly that if the elections were won only by popular vote the Democrats would have a very large edge themselves. More than the 15% edge indicated by the simulations for Electoral College only wins for the Republicans, which was pretty much unchanged with the 2020 data.
So, I haven’t really changed my view that simply getting rid of the Electoral College is not a solution to a fairer system.
Still, it occurred to me, that if the Democrats get more votes while the Republicans get more Electoral votes this is essentially the definition of Gerrymandering. The Democrats clearly have wasted votes. This has been called an Efficiency Gap and used as a measure of Gerrymandering. I also already looked at Gerrymandering. But not in this context. This was just supposed to be an update of the 2020 election results. But this was where I was going to put in some time on something new. Maybe, finally getting around to it. Teaser.
To what extent given the Electoral College and the prevailing population demographics is the United States of America a gerrymandered nation?
For the Efficiency Gap - Wikipedia
This is not a “scientific” experiment. It doesn’t prove anything. I think it may suggest something though. This is that the Democrats have recently had a popular vote advantage. There is no reason to believe this won’t continue. Possiblly, the Electoral College gives the Republicans a slight advantage in close elections. Maybe, even to the extent of being unfair to Democrat candidates in some of these. But, from what I’ve seen it seems that the Democrats have a considerable advantage based on popular vote in most of the elections to offset this.
Changing the process could have unwanted consequences. It could make a tipping point where the Democrats win almost all presidential elections. Establishing a Democrat dynasty. This is contrary to the point of a two party system. Even if that did not happen it would certainly diminish the importance of smaller states. Presidential election campaigns could come down to candidates going from one major city to another. Issues and concerns of voters in the smaller states would become more unheard.
I’m not sure this is something you want to trifle with lightly. As Churchill said,
Many forms of Government have been tried, and will be tried in this world of sin and woe. No one pretends that democracy is perfect or all-wise. Indeed it has been said that democracy is the worst form of Government except for all those other forms that have been tried from time to time.…
You don’t want to inadvertently turn our system into one of the other forms. It seems like the idea of switching to popular vote is gaining some traction but from the link above, Why the Electoral college,
changing it is very unlikely. It would take a constituitional amendment ratified by 3/4 of states to change the system. It is hard to imagine the smaller states agreeing. One way of modifying the system is to eliminate the winner take all part of it. The method that the states vote for the electoral college is not mandated by the consitution but is decided by the states. Two states do not use the winner take all system, Maine and Nebraska.
I might add that I am against split electoral votes as well. It seems an unnecessary and annoying complication when doing something like this or other presidential election related computer programming. For this I made everyone winner take all. So there Maine and Nebraska.
“One person, one vote” is an important democratic principle. But in a representative democracy so is “equal representation”. The Electoral College I think may to some extent provide this for smaller states. Which means as the demographics have shaken out, to generally be the Republicans.
So again, I think the Hardball guest is still currently correct beyond the founding fathers. The Electoral College does somewhat protect the votes of smaller states, but in a broader sense than just the extra 2 non-population based seats it gets votes for. Their electoral vote is guaranteed and not wiped out by having a small margin of victory compared to larger states with much larger margins of victory.
For the record I tracked down the Hardball segment.
Majority
of Americans support getting rid of the Electoral College
The guest was Michael
Steel