United States 2020 Election Probabilities

[AspieAlly613]

I would want to ask the following three questions:

 

1. What do you think is the likelihood of a Democrat being elected President in 2020?
2. What do you think is the likelihood of the Democrats controlling the Senate in 2021 and 2022?
3. What do you think is the likelihood of the Democrats controlling the House of Representatives in 2021 and 2022?

 

However, because most people are absolutely terrible at estimating probabilities, I'm going to ask the following related questions instead:

 

1. What do you think is the most likely percentage-wise vote margin of popular vote victory/loss for the Democratic candidate in the 2020 Presidential election?
2. What do you think that margin would need to be, accounting for any electoral college advantage that either side may have?
3. Assume that the only Senate seats with a significant likelihood of changing parties are those of Senator Jones (Democrat from Alabama), Senator Collins (Republican from Maine), Senator Tillis (Republican from North Carolina), Senator McSally (Republican from Arizona), and Senator Gardner (Republican from Colorado). If President Trump is reelected, Democrats would need to win all 5 of those races to gain control of the Senate.  If it were known that President Trump will be reelected, what would you estimate the most likely *lowest* margin of victory/*greatest* margin of loss for the Democrats in those 5 races?
4. Now assume that it were known that President Trump will not be reelected.  What would you estimate the most likely *second-lowest* margin of victory/*second-greatest* margin of loss for the Democrats in those 5 races?
5. What do you think the Democrats' margin of victory/loss in the House popular vote will be in the 2020 elections?
6. What do you think the Democrats' margin of victory/loss would need to be to retain control of the House after those elections?

 

Also, can you provide some measure of you uncertainty to your answers to questions 1-6?

[Knight of Cydonia]

I think it's way too early to start making predictions as specific as those. Biden still hasn't even announced whether or not he's going to run, and depending on his choice how people will vote will be distributed pretty differently. Can't really say if the Democratic nominee will beat Trump yet because each of the current candidates has totally different electibility against him.

 

All I can say is I hope (and think) Bernie Sanders will win. Back in the 2016 election I predicted that Hillary would lose to Trump if she was nominated over Bernie. Bernie was far more electable against Trump than her. I think he's still favourable against Trump, and this time around he's actually got a ton of visibility. He did incredibly well given he was an unknown going against such an established politician like Hillary and the support of the DNC - and he'll do even better this time.

[Nick2]

Once the toothpaste is out of the tube it is hard to get it back in.

[QuantumEcho]

only thing i can say, Is may god have mercy on the american elections. 

[Back to Avalon]

I'm surprised that no Republican is challenging Trump for the nomination. He's quite unpopular and therefore, I think, vulnerable. Jeff Flake has already said he won't run. I haven't heard that John Kasich is going to run. I'll be a little surprised if he doesn't, but I don't think he has much time left to decide. Campaign season starts earlier and earlier, it seems.

[iff]

On no 2, it is not about the margin of victory to win the popular vote. The last presidency was decided by 50,000 voters in just four states but their electoral votes were very high in the winner takes all. It is the swing states that count.

 

On no 6, on the margin for the democrats, it is generally thought of as 5% needed in order to take control/retain control in the house of representatives because of a republican advantage on seat districting.

[AspieAlly613]

15 hours ago, iff said:

On no 2, it is not about the margin of victory to win the popular vote. The last presidency was decided by 50,000 voters in just four states but their electoral votes were very high in the winner takes all. It is the swing states that count.

Right, but I think we can agree that if either side wins the popular vote by a 40% margin that candidate is more likely than not to win the election.  Somewhere between those unprecedented results there must be a break-even point.

[Sally]

On 4/12/2019 at 6:39 PM, Back to Avalon said:

I'm surprised that no Republican is challenging Trump for the nomination. He's quite unpopular and therefore, I think, vulnerable. Jeff Flake has already said he won't run. I haven't heard that John Kasich is going to run. I'll be a little surprised if he doesn't, but I don't think he has much time left to decide. Campaign season starts earlier and earlier, it seems.

Trump is a revengeful guy.  If I were a Republican (which I can't really imagine, but...) I would not want to challenge him and lose.  

[Evren]

On 4/12/2019 at 9:39 PM, Back to Avalon said:

I'm surprised that no Republican is challenging Trump for the nomination. He's quite unpopular and therefore, I think, vulnerable. Jeff Flake has already said he won't run. I haven't heard that John Kasich is going to run. I'll be a little surprised if he doesn't, but I don't think he has much time left to decide. Campaign season starts earlier and earlier, it seems.

They are actually, though he isn't well known.

https://www.nytimes.com/2019/02/15/us/politics/bill-weld-trump-2020.html

[iff]

18 hours ago, AspieAlly613 said:

Right, but I think we can agree that if either side wins the popular vote by a 40% margin that candidate is more likely than not to win the election.  Somewhere between those unprecedented results there must be a break-even point.

I would have said in 2016 that 2% margin in popular vote would be enough to win the electoral college but I'd have been proved wrong.

 

Maybe 3%?

[AspieAlly613]

It's about time I actually tried to answer some of my own questions:

 

1) Given that President Trump's current net approval rating is -10%, that we can expect some reversion to the mean in the next year and a half, and that a Democratic candidate would be subject to some negative campaigning, I'm tempted to say that the maximum likelihood margin of popular vote victory would be 5% for the Democratic candidate.  (Margin of error difficult to calculate, but given that an election-day polling average has a 2% probable error, I'd guess a 5% probable error this far in advance, giving President Trump a roughly 25% chance of winning the popular vote.)

 

2)  Here's probably my most surprising estimate:  If the popular vote is tied, I predict a 50% chance of either major candidate winning.  Looking at the tipping-point margin vs. popular vote margin since 2000, the average net difference is only 0.2%...and it favors the Democrats.  One could try to find some patterns (such as performance in the four states that share a border with Mexico, none of which switched parties between 2012 and 2016, but where to combined vote margin grew from 1,700,000 votes in favor  of the Democrats in 2012 to 3,500,000 votes in favor of the Democrats in 2016) but I'm hesitant to cherry-pick data from select states.  The root-mean-square deviation from 0 over those 5 elections was a 1.8% tipping point vs. popular vote difference.  The probable deviation is roughly 2/3 of a root-mean-square deviation giving a probable error of 1.2%.  I believe changes in electoral college advantage to be independent of changes in popular vote margin.  Because probable deviations of independent sources sum in quadrature, combining this with the estimate for 1), I estimate that President Trump has approximately a 25% chance to be reelected.

 

3)  I believe the most Republican-leaning seat that I mentioned is that of Senator Jones, whom I would normally predict would lose by 20%.  (In a politically neutral year, with a republican incumbent, that seat leans around 30% Republican, but this is likely to be a blue-ish year, and there is a Democratic incumbent.)  If it were *known* that President Trump would win, then my assessment that it is likely to be a blue-ish year would be proven wrong, and I would change my estimate to Senator Jones losing by 25%.  (Margin of error would be way less than 25%, so I would consider Senator Jones' defeat all-but-inevitable if we knew that President Trump would be reelected.

 

4) Estimating the margin of victory in the *second*-most-Republican-leaning seat would be harder.  I know order statistics don't work this way, but I'm going to do this analysis anyway.  I predict that Senator Collins will be reelected by a 10% margin if President Trump is not reelected.  With a 5% probable error that would lead me to give her a roughly 90% chance of getting reelected.  I believe Senator Collins to have a greater chance of Reelection than Senators McSally or Gardner do.

 

5&6) I see no reason to expect ticket-splitting to go one way or the other.  I'd give the Democrats a 5% edge in the popular vote.  I'd also give them a 5% Homophily-induced disadvantage.  (Gerrymandering has little to do with it at this point, the REDMAP initiative was based on the results of the 2010 census and has had less and less of an effect in subsequent elections.)  This would give the Democrats a 50% chance to maintain control of the House.