Political Prediction Thread

[Bootstoots]

If you rank the states from largest Dem margin to largest Rep margin, it's the state that provides the 270th electoral vote to the winner. It puts the winning candidate over the top, and any further states won don't have any real effect on the outcome, they just add to the victor's total. So winning the tipping point state is equivalent to winning the election.

This is somewhat different than the concept of a swing state because, in a blowout election, the tipping point state need not be particularly close. Nate Silver introduced the concept here.

For instance, in 2008 the closest states were Indiana, North Carolina, and Missouri; Obama narrowly won the first two and very narrowly lost MO. So all of those were swing states, but he did not need any of them to win the election. The tipping point was Colorado in both elections, and Obama carried it by 8.95% in 2008 and 5.37% in 2012.

The reason I've been arguing that Pennsylvania should be treated as a swing state by both campaigns is that it's near the tipping point (being more Dem than the actual tipping point state by 0.02% in 2012 and 1.3% in 2008) with a marginally redder trend than the nation as a whole. Winning it is one of the most likely ways that the Republicans can win the election, although at the moment this doesn't look especially likely.

 

[onejayhawk]

What do you mean by "tipping point state"?... I'm inclined to disagree but it occurs to me that I would be asinine to disagree with a prediction I don't even really understand.

Pennsylvania is a hot choice currently. Previous years it has been Florida, Colorado or Ohio.

J

 

[sophie]

A tipping point state is the state upon which a candidate secures their election. It doesn't necessarily have to be a battleground state or contested state. It's just the point at which a candidate's accumulated EVs crest 272. If a candidate has so much support that they can secure an election without needing any of the contested states then it's conceivable that a fairly secure state could be the tipping point.

eg:

http://www.electoral-vote.com/evp2012/Pres/EVP_average.html

Notice that for Obama the tipping point state, Ohio, was a leans blue state. Obama's support, due to demographics and the general course of the election meant that Obama could win without needing tossup states or leans-red states like Florida, whereas Romney utterly needed 4 blue-leaning states and couldn't win without them.

 

[onejayhawk]

Notice that for Obama the tipping point state, Ohio, was a leans blue state. Obama's support, due to demographics and the general course of the election meant that Obama could win without needing tossup states or leans-red states like Florida, whereas Romney utterly needed 4 blue-leaning states and couldn't win without them.

That is a problem with a lot of the analysis. They treat 2012 as an even starting line when it actually favors the Democratic candidate by 3.5% to 4%. That's the popular margin between Obama and Romney. Normalizing only flips Florida but other states become very close. You can see some of this when the polls give the result of a generic ballot.

J

 

[Sommerswerd]

If you rank the states from largest Dem margin to largest Rep margin, it's the state that provides the 270th electoral vote to the winner. It puts the winning candidate over the top, and any further states won don't have any real effect on the outcome, they just add to the victor's total. So winning the tipping point state is equivalent to winning the election.

I get it, so in football terms (I love football terms), Denver scored the "tipping point" field goal in the 2nd quarter of the Superbowl, and the rest of the game didn't matter after that, because Denver already had all the points they needed to win? Or is your point more like a mercy rule, where the election is basically over when the Eastern time zones finish reporting?

In either case, I don't think the tipping point concept is particularly relevant or meaningful, because this isnt a primary, where each state votes on a different day. All the states vote at the same time/day so the "tipping point" is nothing more than an imaginary concept with no impact on the election. In other words, PA being "the tipping point state" wont have any impact on the election. All you're basically saying is that this year Karl Rove will be melting down when they call PA rather than OH, is that right?

 

[Sommerswerd]

A tipping point state is the state upon which a candidate secures their election. It doesn't necessarily have to be a battleground state or contested state. It's just the point at which a candidate's accumulated EVs crest 272. If a candidate has so much support that they can secure an election without needing any of the contested states then it's conceivable that a fairly secure state could be the tipping point.

eg:

http://www.electoral-vote.com/evp2012/Pres/EVP_average.html

Notice that for Obama the tipping point state, Ohio, was a leans blue state. Obama's support, due to demographics and the general course of the election meant that Obama could win without needing tossup states or leans-red states like Florida, whereas Romney utterly needed 4 blue-leaning states and couldn't win without them.

I went back and counted by percentages... In pure percentages terms, Colorado would be where Pres Obama crosses 270. In other words, he wins without Ohio, Virginia, or Florida.

Trump is going to lose... you can write it down.

 

[onejayhawk]

I went back and counted by percentages... In pure percentages terms, Colorado would be where Pres Obama crosses 270. In other words, he wins without Ohio, Virginia, or Florida.

Trump is going to lose... you can write it down.

This one depends on which numbers are used. Colorado and Pennsylvania were very close in margin. Either can be used. Nate Silver predicts it will be Pennsylvania this year. I predict Trump wins PA.

Clinton is going to lose. You can write it down.

Below is a source where the states are listed in order of margin. Trump will also win Colorado, so I suppose it does not matter.
http://www.democratichub.com/election/8?elr=m&cand=barack+obama&s=a

J

 

[Bootstoots]

I get it, so in football terms (I love football terms), Denver scored the "tipping point" field goal in the 2nd quarter of the Superbowl, and the rest of the game didn't matter after that, because Denver already had all the points they needed to win? Or is your point more like a mercy rule, where the election is basically over when the Eastern time zones finish reporting?

In either case, I don't think the tipping point concept is particularly relevant or meaningful, because this isnt a primary, where each state votes on a different day. All the states vote at the same time/day so the "tipping point" is nothing more than an imaginary concept with no impact on the election. In other words, PA being "the tipping point state" wont have any impact on the election. All you're basically saying is that this year Karl Rove will be melting down when they call PA rather than OH, is that right?

I went back and counted by percentages... In pure percentages terms, Colorado would be where Pres Obama crosses 270. In other words, he wins without Ohio, Virginia, or Florida.

Trump is going to lose... you can write it down.

It doesn't have anything to do with when the state is called, just the margins of victory in each state. The football analogy doesn't quite work because there's a fixed number of EVs, whereas theoretically Carolina could have come back from behind in the second half because there's not a fixed number of points possible in a football game. If there were like 60 possible points in a football game, then the touchdown/field goal/safety that causes a team to have 31 or more points would be the tipping point. And then to complete the analogy the whole rest of the game still gets played for some reason.

I think you understand it in your second post - the tipping point was indeed Colorado. Owen's source is a prediction before the election of how the states were going to vote which had Ohio as the most likely tipping point; in reality, Obama crossed 270 with CO and ended up not needing OH, VA, or FL. If you rank the margins of victory from largest margin for the winner on down, and add up the EVs, the state at which the winner crosses 270 is the tipping point state. In this case it goes:

Margin of victory and total EVs
DC: 83.63% 3 EV
HI: 42.71% 7 EV
VT: 35.60% 10 EV
...
PA: 5.39% 263 EV
CO: 5.37% 272 EV <-- tipping point
VA: 3.87% 285 EV
OH: 2.98% 303 EV
FL: 0.88% 332 EV

 

[Sommerswerd]

It doesn't have anything to do with when the state is called, just the margins of victory in each state. The football analogy doesn't quite work because there's a fixed number of EVs, whereas theoretically Carolina could have come back from behind in the second half because there's not a fixed number of points possible in a football game. If there were like 60 possible points in a football game, then the touchdown/field goal/safety that causes a team to have 31 or more points would be the tipping point. And then to complete the analogy the whole rest of the game still gets played for some reason.

I think you understand it in your second post - the tipping point was indeed Colorado.

Yeah I had to read your and Owen's post again to get where you were coming from, but I still don't see how the "tipping point" concept is relevant.

I'm glad you mentioned how the football analogy isn't on-point, because I think this is also why the tipping-point concept isn't very useful. As you correctly point out, the difference between a football game and the election is that the team that is behind can come back to win... but what is equally true, is the underdog can look like they are behind on paper, but then win the game inexplicably, by outplaying the opponent, rallying, changing strategy in the middle of the game, injuring an opponents key player or getting them thrown out of the game in a key situation... etc.

None of that stuff is at play during an election. In an election we vote for the winner, they don't "play" each other on election day. I mean its covered like a sporting event, but in reality, there is no "ahead" and "behind", there is no "catching up" because they aren't doing anything. It would be like if we simply voted for the Superbowl winner based on who we thought deserved to be champion, given their season performances.

Calling a state the "tipping-point" implies that the state has some special relevance or significance, like a "swing-state" or "battleground-state", maybe there is something I am missing?

 

[Lexicus]

What you're saying makes sense, but I think the tipping point thing makes sense insofar as it's when we become aware that a candidate has won the election. It isn't like, an ongoing process of contesting the win, it's just a vagary of how elections are reported. But no less valid a concept for that.

 

[AdrienIer]

Yeah I had to read your and Owen's post again to get where you were coming from, but I still don't see how the "tipping point" concept is relevant.

I'm glad you mentioned how the football analogy isn't on-point, because I think this is also why the tipping-point concept isn't very useful. As you correctly point out, the difference between a football game and the election is that the team that is behind can come back to win... but what is equally true, is the underdog can look like they are behind on paper, but then win the game inexplicably, by outplaying the opponent, rallying, changing strategy in the middle of the game, injuring an opponents key player or getting them thrown out of the game in a key situation... etc.

None of that stuff is at play during an election. In an election we vote for the winner, they don't "play" each other on election day. I mean its covered like a sporting event, but in reality, there is no "ahead" and "behind", there is no "catching up" because they aren't doing anything. It would be like if we simply voted for the Superbowl winner based on who we thought deserved to be champion, given their season performances.

Calling a state the "tipping-point" implies that the state has some special relevance or significance, like a "swing-state" or "battleground-state", maybe there is something I am missing?

Theoretically the democrats could only campaign in the 5 states leading to the tipping point. They could (although of course they won't) avoid Florida Ohio and Virginia, because the tipping point in 2012 was Colorado. If the order of most democratic/most republican states stays the same from 2012, the democrats basically have to make sure they keep Colorado Pennsylvania NH Iowa Nevada and Wisconsin to win.

Any state the democrats win beyond that is just overkill (in the context of 2012).

 

[Bootstoots]

Yeah I had to read your and Owen's post again to get where you were coming from, but I still don't see how the "tipping point" concept is relevant.

I'm glad you mentioned how the football analogy isn't on-point, because I think this is also why the tipping-point concept isn't very useful. As you correctly point out, the difference between a football game and the election is that the team that is behind can come back to win... but what is equally true, is the underdog can look like they are behind on paper, but then win the game inexplicably, by outplaying the opponent, rallying, changing strategy in the middle of the game, injuring an opponents key player or getting them thrown out of the game in a key situation... etc.

None of that stuff is at play during an election. In an election we vote for the winner, they don't "play" each other on election day. I mean its covered like a sporting event, but in reality, there is no "ahead" and "behind", there is no "catching up" because they aren't doing anything. It would be like if we simply voted for the Superbowl winner based on who we thought deserved to be champion, given their season performances.

Calling a state the "tipping-point" implies that the state has some special relevance or significance, like a "swing-state" or "battleground-state", maybe there is something I am missing?

It's a concept like "swing state", but I would argue it's a more important one. The probability of a state putting a candidate over 270 is the key factor that decides what states are important to campaign in. In the 1984 election with Reagan vs. Mondale, the closest states were Minnesota and Massachusetts. You could argue that they were swing states because they were very close, but they were irrelevant to the overall outcome because they just determined how enormous the landslide was, not who the winner was. (FWIW, Reagan came within 0.2% in Minnesota of winning all 50 states. )

When I say that Pennsylvania will be at or near the tipping point, I'm arguing that if the national popular vote is nearly tied, then Pennsylvania has a good chance of deciding the election. Ohio and Florida likely go Republican under those circumstances, and it's down to Pennsylvania along with a few others with similar margins (VA, CO, IA, and NH, going by 2012) to decide the election. Of those, PA alone could swing it.

If on the other hand Clinton goes into the election lead of 6 points nationally, then PA won't even be especially close and the map will look like one of Obama's maps. Then the "swing states" will be places like North Carolina or even Missouri, Indiana, Arizona, Georgia, etc. which may be close if Trump loses nationally by 6, but would just serve to run up the score for Clinton if she were to win them.

I agree with Nate Silver that Trump has a ~20% chance of winning the election. In the 20% of scenarios where he does win, the following is almost certainly one of the likeliest:

Spoiler Disturbing content. Viewer discretion advised. :

(EV count 273-265 for the GOP. Feel free to flip any of VA, CO, IA, or NH too, but obviously it doesn't matter)

 

[Gucumatz]

IMO I think if you were going to somehow come up with a Trump victory, I think you would have to assume at this point a loss in Florida regardless thanks to demos, even with a magic Republican tide. Michigan looks more competitive than Florida does at this point with a said magic Republican tide

 

[Bootstoots]

No it doesn't. The RCP average is at Clinton+3.7 in Florida. There's nothing magic about a 4-point swing in 4 months, nor is the average of Michigan polls particularly close to Florida ones. I do agree that assuming the states will come in the same order with the same margins in 2016 is rather suspect: I expect Florida and Virginia will be somewhat more Democrat than they usually are relative to the national average, while Pennsylvania comes in slightly more Republican; not sure about Ohio, Iowa, New Hampshire, or Colorado at this point.

 

[Hygro]

It's a concept like "swing state", but I would argue it's a more important one. The probability of a state putting a candidate over 270 is the key factor that decides what states are important to campaign in. In the 1984 election with Reagan vs. Mondale, the closest states were Minnesota and Massachusetts. You could argue that they were swing states because they were very close, but they were irrelevant to the overall outcome because they just determined how enormous the landslide was, not who the winner was. (FWIW, Reagan came within 0.2% in Minnesota of winning all 50 states. )

When I say that Pennsylvania will be at or near the tipping point, I'm arguing that if the national popular vote is nearly tied, then Pennsylvania has a good chance of deciding the election. Ohio and Florida likely go Republican under those circumstances, and it's down to Pennsylvania along with a few others with similar margins (VA, CO, IA, and NH, going by 2012) to decide the election. Of those, PA alone could swing it.

If on the other hand Clinton goes into the election lead of 6 points nationally, then PA won't even be especially close and the map will look like one of Obama's maps. Then the "swing states" will be places like North Carolina or even Missouri, Indiana, Arizona, Georgia, etc. which may be close if Trump loses nationally by 6, but would just serve to run up the score for Clinton if she were to win them.

I agree with Nate Silver that Trump has a ~20% chance of winning the election. In the 20% of scenarios where he does win, the following is almost certainly one of the likeliest:

Spoiler Disturbing content. Viewer discretion advised. :

(EV count 273-265 for the GOP. Feel free to flip any of VA, CO, IA, or NH too, but obviously it doesn't matter)

basically... predicting which state is the cheapest state to campaign in?

 

[Bootstoots]

Well, the goal is to maximize (change in probability of winning election)/(change in dollars spent). dP/d$, if you will. So there's not much point in campaigning in Indiana, Missouri, Georgia, or Arizona, because those states are Republican enough that they won't swing the election to one side or another. Likewise with, say, Michigan, New Mexico, or New Jersey for the Democrats. If a candidate does win one or more of these sorts of states (e.g. Obama won Indiana and almost won Missouri in 2008) it just means that they won the election by a wide margin. One or both campaigns may still spend money in that sort of place if they have extra money to blow; for instance, Obama successfully did this in Indiana in 2008 when he had enough money and lead in the polls that he could afford to do it and run up the score, but he didn't bother in 2012 when it was slightly tighter.

To me a swing state is a state that could actually swing the election, not just run up the score against someone who would already lose anyway, like when Reagan won Massachusetts and Rhode Island in 1984. This is a lesser-used definition of swing state, but IMO it's a more useful one.

 

[Sommerswerd]

It's a concept like "swing state", but I would argue it's a more important one. The probability of a state putting a candidate over 270 is the key factor that decides what states are important to campaign in. In the 1984 election with Reagan vs. Mondale, the closest states were Minnesota and Massachusetts. You could argue that they were swing states because they were very close, but they were irrelevant to the overall outcome because they just determined how enormous the landslide was, not who the winner was. (FWIW, Reagan came within 0.2% in Minnesota of winning all 50 states. )

When I say that Pennsylvania will be at or near the tipping point, I'm arguing that if the national popular vote is nearly tied, then Pennsylvania has a good chance of deciding the election. Ohio and Florida likely go Republican under those circumstances, and it's down to Pennsylvania along with a few others with similar margins (VA, CO, IA, and NH, going by 2012) to decide the election. Of those, PA alone could swing it.

If on the other hand Clinton goes into the election lead of 6 points nationally, then PA won't even be especially close and the map will look like one of Obama's maps. Then the "swing states" will be places like North Carolina or even Missouri, Indiana, Arizona, Georgia, etc. which may be close if Trump loses nationally by 6, but would just serve to run up the score for Clinton if she were to win them.

I agree with Nate Silver that Trump has a ~20% chance of winning the election. In the 20% of scenarios where he does win, the following is almost certainly one of the likeliest:

Spoiler Disturbing content. Viewer discretion advised. :

(EV count 273-265 for the GOP. Feel free to flip any of VA, CO, IA, or NH too, but obviously it doesn't matter)

PA won't flip regardless of how close it is. NH, IA, CO, maybe even VA, NV, ME in a nailbitter close race (which is the only kind any Republican would have a chance at winning), are the only realistic possibilities... along with the obvious necessities of FL, OH, IN, MO, NC, some of which are already dubious for Trump. But PA?... Just... No

 

[Bootstoots]

To whatever extent it can be believed (there aren't that many recent state polls, for one), the RCP average is at Clinton+2.3 in PA, compared to Clinton+4.6 nationwide, so I don't see why PA would be especially unlikely to flip in a situation where the national vote is tied or has a very slight Republican edge. It did go for Kerry in the Reps' only popular vote win since 1992, but Trump's running a campaign that is unusually popular among white Rust Belt inhabitants, and the current numbers suggest that PA will be about as tight as the other battleground states.

Granted it probably won't flip because the election will probably feature a significant Clinton edge nationwide, but I don't see why you're so confident about it.

 

[Hygro]

But if you predict your tipping point state correctly, campaign there, and then win by a wider margin than the next one, which is your tipping point state?

Doesn't this whole thing assume that each state has voters of equal cost to swing, with each voter as a percent of the margin of victory being marginally more expensive as the last, for the concept to matter?

 

[onejayhawk]

If on the other hand Clinton goes into the election lead of 6 points nationally, then PA won't even be especially close and the map will look like one of Obama's maps. Then the "swing states" will be places like North Carolina or even Missouri, Indiana, Arizona, Georgia, etc. which may be close if Trump loses nationally by 6, but would just serve to run up the score for Clinton if she were to win them.

I agree with Nate Silver that Trump has a ~20% chance of winning the election. In the 20% of scenarios where he does win, the following is almost certainly one of the likeliest:

Spoiler Disturbing content. Viewer discretion advised. :

(EV count 273-265 for the GOP. Feel free to flip any of VA, CO, IA, or NH too, but obviously it doesn't matter)

If the vote was Tuesday, Clinton would probably win. This is nothing more or less than an acknowledgment that she has a lead. That is Silver's Now-Cast number. His best guess is called Polls-Plus, which is a little more lenient to Trump's chances. Today he has it at 28.4%.

I agree that Trump winning via PA is the most likely outcome. The part I question is PA before VA. That said, it's the way the polling is going.

PA won't flip regardless of how close it is. NH, IA, CO, maybe even VA, NV, ME in a nailbitter close race (which is the only kind any Republican would have a chance at winning), are the only realistic possibilities... along with the obvious necessities of FL, OH, IN, MO, NC, some of which are already dubious for Trump. But PA?... Just... No

It's going to happen. Trump will crack some of the ice near Philly and own upstate.

The Republicans can certainly win a blowout, a la 1980 - 1988. Reagan won with southern Democrats jumping ship. Trump has also been pilfering registered Democrats.

But if you predict your tipping point state correctly, campaign there, and then win by a wider margin than the next one, which is your tipping point state?

Doesn't this whole thing assume that each state has voters of equal cost to swing, with each voter as a percent of the margin of victory being marginally more expensive as the last, for the concept to matter?

Which brings us back to the whole idea of swing states. The big states that are near the center of the curve are FL, OH, PA, VA. You could add CO and NV together as a fifth. That is where it will be won or lost.

One thing that bears remarking. Jimmy Carter looked good into October of 1980. His polls numbers never did fall apart, except the ones at the ballot box.

J