-1

In the 2000 U.S. Presidential Election, Gore won the popular vote by 0.5% and lost the Electoral College vote.

In the 2008 U.S. Presidential Election, Obama won the popular vote by 7% and won the Electoral College vote.

In the 2012 U.S. Presidential Election, Obama won the popular vote by 4% and won the Electoral College vote.

In the 2016 U.S. Presidential Election, opinion polls forecasted Clinton win the popular vote by 3–5%. In reality she won by 2% and lost the Electoral College vote.

As of March 2020, nationwide opinion polls give either Sanders or Biden a margin of about 5% over Trump, only marginally more than the 2016 opinion polls just before election day gave Clinton, but more than Obamas 2016 margin.

Is there any rule of thumb by what margin a Democrat de facto needs to win in practice in order to also win the electoral college vote and become President?

JJJ
  • 39,094
  • 10
  • 121
  • 182
gerrit
  • 48,247
  • 16
  • 142
  • 298
  • Probably a duplicate of How likely is it for a Democrat to win the Presidency if he/she gets 52.8% of the (popular) vote?, which was closed as off-topic. – Rick Smith Mar 04 '20 at 10:50
  • It's curious that you skip the 2004 election and the 1992 and 1994 elections (The latter two Clinton won a plurality of the popular vote (Republicans George H.W. Bush ('92) and Bob Dole ('96) lost in several states because third party spoiler Ross Perot (both elections) split the voters favoring Republican candidates, but not a majority but did win a majority of the Electoral College). Clinton received ~42% and ~49% of the popular vote in the respective years. – hszmv Mar 04 '20 at 15:49
  • @hszmv I focussed on Democratic popular vote wins, because in recent history only Democrats have won popular vote while losing electoral college vote. – gerrit Mar 04 '20 at 15:52
  • @gerrit: But by omitting Republican popular wins and plurality wins, you're ignoring data that supports the A. bakker's answer of no corelation. Clinton was a majority popular winner, but he got 270 Electoral votes, so he's President. Additionally, the lose Popular vote win Electoral scenario has happened in five elections out of the the 58 held over the nations 229 year history (from the adoption of the Constitution) and a disputed possible 6th that will probably never be resolved. 91% of the time the PV and EC have had the same person win. – hszmv Mar 04 '20 at 16:05
  • @hszmv A. Bakkers answer of no correlation is trivially incorrect. Of course there is a correlation between popular vote and electoral vote, but the correlation is smaller than 1. See also the discussion in chat I've had with A. bakker. 2 of those 5 have happened in the last 20 years, and based on recent polling data, it is possible to simulate a million election results to calculate at what PV margin the likelihood of losing EV is below 2 sigma. This depends on which party is ahead. – gerrit Mar 04 '20 at 16:12
  • @hszmv I don't know if anyone has applied such Monte Carlo simulations to model voter dynamics and its impact on EV vs. PV discrepancies, but to base it past on historical election results the large majority of which are entirely irrelevant would be useless — there weren't even any nationwide democratic elections in the USA before 1965! – gerrit Mar 04 '20 at 16:14
  • @hszmv Actually, pollsters must have run Monte Carlo simulations for they came up with probabilities of a Clinton win, which are of course hard to validate (if they said Trump had a 10% chance of winning, were they wrong?). This page cites simulations that were used as evidence that struck down Pennsylvania gerrymandering, such simulations can be used to model how "unfair" the EV (currently) is; this is fully missing from the only existing answer. – gerrit Mar 04 '20 at 16:20
  • @gerrit: So we've been living under a dictatorship prior to 1965? Okay, sure... whatever your trying to argue, it's through the lens of a very negative read and you're clearly not interested in learning but demanding validation for your warped denailist history. I'm done talking with you. Come back when you read a basic history book. – hszmv Mar 04 '20 at 16:48
  • @hszmv Before 1965, the USA (either nationwide or in certain states) restricted (de facto or de jure) suffrage to certain groups. The further back you go, the more severely restricted this was. An election where certain groups are excluded is not a democratic election. I am surprised this would be considered a controversial or debatable statement. Surely you wouldn't call a white-only (longer ago: white men only, longer ago yet: white land owning men only) election democratic?! – gerrit Mar 04 '20 at 16:52
  • @hszmv See Voting rights in the United States for a summary of the history. – gerrit Mar 04 '20 at 16:56
  • @gerrit: And what you fail to realize is that by moving the goal posts like this is throwing out data and making it more difficult to answer this question to what you are claiming is something that can satisfy you. And if you want a million election simulation perfectly modeled so that all issues and cultural zeit geists are accurately reflected for a time period of 4 million years (1 election per year) you're not asking for a reasonable time considering that it not only is a longer period of time than man has walked the earth, but is twice the time that the entire Homo genius has as well.+ – hszmv Mar 04 '20 at 17:11
  • The best you can do is look at several predictive algorithms for this coming election, which predicts Trump winning with more electoral college votes than in 2016 under any turn out that is not historically large by the records. There is no way to simulate accurately better than that, but if you use all available conditions, for every 10 elections, the 10th election will get a PV/EC split in 9 times out of 10. But since all options to get the split are possible in each election, it will come up seemingly more frequently in some times+ – hszmv Mar 04 '20 at 17:16
  • +Consider flipping a coin... we know that it has a 50:50 chance that one face will land up... but when we actually do it, it's possible to get three flips face up in a row because each flip is not dependent on the previous. 2000 and 2004 are WAY different in politics and you can point to a major instance of why just from one book. The similarity is true for every sufferage grant, which will have a wildly different factor to consider, but because states can delegate electors however they please, even unjustly so, it doesn't affect the EC like you think it does. – hszmv Mar 04 '20 at 17:19
  • I'm not moving the goalposts at all. Rick quickly pointed out that my question was a duplicate (thanks!), the linked question has answers that perfectly answer my question, backed up by solid research, the single existing answer to this post does not address my question and is factually incorrect. – gerrit Mar 04 '20 at 18:20
  • As others have mentioned, the popular vote has nothing to do with the electoral college, and, in fact, the amount needed to win the electoral college can theoretically be quite small. –  Mar 04 '20 at 18:24
  • @Chipster The claim that the popular vote has nothing to do with the electoral vote is factually incorrect. The popular vote within each state determines which candidate gets to send people to the electoral college, so mathematically, the overall popular vote must necessarily correlate positively with the electoral vote. To claim anything else is mathematically incorrect. No amount of outliers can mathematically reduce this correlation to 0 or make it negative (but it can be arbitrarily close to 0). – gerrit Mar 04 '20 at 19:17
  • @Chipster Just because it is possible to win the electoral college with 0.00001% of the popular vote does not mean there is no correlation. If you consider all election outcomes that are theoretically possible, then calculate the PV and EV result for all, you will get a positive correlation. – gerrit Mar 04 '20 at 19:20
  • Sure. But my point is that it technically doesn't matter. It is mathematically possible to not even get over 50% of the national popular vote and still win. That is all I'm trying to point out. –  Mar 04 '20 at 20:54

1 Answers1

3

The popular vote matters nothing concerning the election of the president either Democratic or Republican. What matters is the electoral college which can be won with (seemingly) a minimum of 23% of the votes according to this article. But of course the percentage will change regularly due to population increasing/declining in certain areas.

And when we look for a rule of thumb there doesn't seem to be one seeing John Quincy Adams won the 1824 election with a margin of -10.44% and Warren Harding winning the 1920 election with a margin of 26.17% (and if we only include Democratic presidents the difference in margin is still between 0.17% for JFK in 1960 and FDR in 1936 with a margin of 24.26%)

So simply said, there is no rule of thumb.

A.bakker
  • 2,079
  • 7
  • 13
  • I know the legal rules, but I wrote "rule of thumb". The theoretical minimum is irrelevant in practice. De facto, popular vote correlates with electoral college vote. – gerrit Mar 04 '20 at 10:35
  • @gerrit edited my answer. – A.bakker Mar 04 '20 at 10:40
  • And 1824 or 1960 results are hardly relevant for todays demographics. – gerrit Mar 04 '20 at 11:30
  • 1
    You are asking for a rule of Thumb... This would require data analysis if you exclude everything outside the current demographic (so let's say nothing beyond 1970) you will only have 5 democratic wins and 2 losses where they got the popular vote...this isn't enough data to do anything with other then stating the conclusion "There is no real correlation between popular vote margins and the electoral vote" – A.bakker Mar 04 '20 at 11:41
  • Of course there is, there is very extensive (exit) poll data per state, categorised by many different demographic factors, and plenty of models based on these data. – gerrit Mar 04 '20 at 11:54
  • 1
    Your question is "Is there any rule of thumb by what margin a Democrat de facto needs to win in practice in order to also win the electoral college vote and become President?" To this the demographic doesn't matter nor does the state. All that matters is the total count which is just to low on (re presentable) data to give an mathematically sound answer. It is simple...there is NO CORRELATION WHATSOEVER between the popular vote margin and the chances on becoming president. The only thing you can say is that the popular vote gives a slight larger probability. – A.bakker Mar 04 '20 at 12:17
  • You're contradicting yourself. You state that there is no correlation (which is false), then you state that "the popular vote gives a larger probability", which means there is a correlation (which is true). – gerrit Mar 04 '20 at 12:30
  • The popular vote gives a larger probability(if they get the PV they will likely have the EV but the PV does NOT increase the chances so no correlation), the margin matters next to nothing with the limited set of data available seeing this data indicates that there is no feasible correlation due to people losing with larger margins then winning with them. If it were true correlation then every time the PV winner should have won the EV or every time a party has larger then X Margin PV should have won the EV. – A.bakker Mar 04 '20 at 12:41
  • If the popular vote gives a larger probability, then there is, by definition, a correlation larger than 0. Your final sentence is factually incorrect for correlations smaller than 1. The Pearson correlation can be between 0 and 1 even with large outliers in both directions, and the Spearman correlation can be 1 even when the Pearson correlation is much smaller than 1. – gerrit Mar 04 '20 at 12:45
  • Let us continue this discussion in chat. – A.bakker Mar 04 '20 at 12:47
  • The 23% figure is wrong because you don't need >50% to win a state. Trump won FL, MI, PA, WI all with less than 50% of the vote. – Alec Mar 04 '20 at 23:03