538 Model Megathread

[Nym90]

The reason Silver got the primaries wrong was because he was just winging it and had no model. You'll notice his models predicting each indvidual primary were quite good.

That's one of the big ironies of Trump's winning the nomination....the numbers were telling us ever since July of last year that he was the favorite, it's just that no one believed the data. In their defense though, early primary polls had quite often been wrong before, so there was good reason to be skeptical and think that Trump was another Herman Cain or Rudy Giuliani, but it's something to keep in mind. The polls this year have actually been more accurate than normal, which is the opposite of what your intuition might tell you about such an unconventional candidate as Trump.

[Beefalow and the Consumer]

If you sort the states by Trump's %chance to win, then NH is actually the tipping point at 32.8% vs 27.6% for VA. 

So Trump's narrow path ( sorted by % chance to win) is Romney + OH + CO + IA + FL + NH

His percentages seem to update in real-time.  I think there's a constant punch tape loader feeding into his comp-u-tron.

[IceSpear]

Are we really going to pretend that Johnson will get more than 2-3% of the vote at best for this entire cycle?

[Beefalow and the Consumer]

Are we really going to pretend that Johnson will get more than 2-3% of the vote at best for this entire cycle?

Given two major-party candidates with the highest unfavorables ever?  Given that an activist who never held public office running on a leftist fringe ticket won nearly 3% of the vote in 2000, and Johnson was a two-term governor running as a third-way (ish) candidate?

[Ljube]

Are we really going to pretend that Johnson will get more than 2-3% of the vote at best for this entire cycle?

No, of course not. At least I won't.

[Nym90]

Perhaps Clinton going after NE-2 is their backup plan in case Trump wins ME-2. That would be pretty crazy outcome withe the race coming down to NE and ME and their weird rules.  You can bet the NE legislature would be kicking themselves for not going through with the idea of switching to WTA.

Makes sense, though the odds of it coming down to one EV are quite slim. But with a vulnerable incumbent D congressman in NE-02, it's definitely worth spending money there.

[DS0816]

Nate Silver predicts Trump will never win the Republican nomination!

Didn’t Nate Silver throw a fit in May when Donald Trump had some poll leads over Hillary Clinton—and said those polls were too far out from the general election, come November, and must be dismissed?

[eric82oslo]

Nate Silver predicts Trump will never win the Republican nomination!

Didn’t Nate Silver throw a fit in May when Donald Trump had some poll leads over Hillary Clinton—and said those polls were too far out from the general election, come November, and must be dismissed?

That's basically what he said in his last article too, if you read it all. He basically said that every single poll or prediction released before the two party conventions should be taken with several Himalaya + Andes mountains pinches of salt. In the previous two or four campaigns he's always stressed tiredlessly that polls hardly mean anything at all until Labour Day or September 1. This year however, because of the extremely early conventions for both parties, I guess he's frontloaded those dates much closer to August 1. Perhaps August 10th or 15th.

[Sorenroy]

Nate Silver predicts Trump will never win the Republican nomination!

However that was based more on his own beliefs then on data. There is a big difference between Nate Silver the pundit and Nate Silver the statistician.

Also, it is in no way difinitive. Using the 80% confidence interval and the polls-plus model the map could swing as far to the left as:

and as far right as:

(Michigan)

[Angel of Death]

Here's the old forecast from 2012. In particular, I miss the explicit "tipping point state" probabilities this time.

[hopper]

Utah at R+6 seems too dem optimistic and they actually give Gary Johnson an >8% chance of getting an electoral vote because of those junk swing state polls with him getting 10-15% on average.

I don't think so since Mormons in Utah dislike Trump's rhetoric about Latino's and Muslims. Election Projection.com has Trump winning Utah by 7% points.

[pbrower2a]

Here's the old forecast from 2012. In particular, I miss the explicit "tipping point state" probabilities this time.

The Presidential election of 2012 was easy to model because

(1) it was not an open-seat election
(2) the President was predictable enough and perceived to be neither a spectacular success nor failure
(3) the Parties have little ideological overlap
(4) the States were sharply polarized in their partisan affiliation, and this was stable
(5) there were no new causes and there was no pervasive change in cultural patterns


So look at those factors. (1) An open-seat election usually hinges upon who wins contested primaries and caucuses, and it is often difficult to predict who will win that. Incumbents rarely lose nominating processes, so that leaves  the incumbent in an election usually his to lose.  That does not apply this time, of course. 22nd Amendment.

(2) History will almost certainly rate Barack Obama as an above-average President, even if most people saw him as a mixture of successes and disappointments. He did not expand his coalition as did FDR, JFK, or Reagan. But neither did he make many mistakes. "General Motors is alive and Osama bin Laden is dead" became the unofficial slogan of Democrats in 2012.

OK. Putting an end to the most dangerous economic meltdown in nearly eighty years and whacking the worst anti-American terrorist ever while avoiding scandals and military or diplomatic disasters is one way to get re-elected even if one loses about 20 states by 10% or more in the previous election. Barack Obama should have been re-elected and was.

(3) The 2010 election knocked out the conservative wing of the Democratic Party in the House of Representatives, and the Republican Party has nothing close to a liberal. Ideological choices between candidates are rather stark. Such makes individual choices easy

(4) Barack Obama winning Indiana in 2008 was something of a freak, something that happens when a state has a large industry (recreational vehicles)  whose market can be hit hard by a combination of an economic downturn, a credit crunch, and high gasoline prices, all of which hit in 2008. That is the only big swing in a state that Barack Obama won. The shift of the Deep and Mountain South from D to R between 1976 and 2008 was basically complete, and it now looks irreversible.

Barack Obama did not depend upon freakish conditions that could evaporate quickly to win.

(5) Tea Party? That's the closest thing to social change. There was no great new religious revival to create a new right-wing ascendancy and no leftish populist movement appearing from seemingly nowhere.

So I could come up with a model. First, recognize that seventeen states and the District of Columbia had not voted for any Republican nominee after 1988 and that there was no marked Republican drift in any of them. That's 243 of the 270 electoral votes that one needs for winning the Presidency, nearly 90% of the needed vote. Second, figure that three states had voted  only once for a Republican nominee for President in the same time, and showed no sign of going Republican this time. That's up to 258. That means that either Virginia, Ohio, Florida, or the combination of Colorado and Nevada would win the election/ Some other states?

Obama wasn't going to win Missouri without winning either Ohio or Virginia; he wasn't going to win Indiana without also winning Ohio; he wasn't going to win Arizona without also winning Colorado and Nevada; he wasn't going to win North Carolina without also winning Virginia; he wasn't going to win Georgia without winning Florida, North Carolina, and of course Virginia.  He also wasn't going to win Colorado without also winning Nevada.

So the 2012 election boiled down to four states -- Colorado, Florida, Ohio, and Virginia, any of which would decide the election. Those states are different enough and separated enough that there is no way to make an appeal that could resonate in all four of those states without
shifting America on the whole.

Treat winning any of the four states as  what statisticians call independent events (coin tosses and throws of a die are independent events), and my crude model held that President Obama had one chance in sixteen of losing if each of the states was a 50-50 proposition. I saw President Obama having a 93.75% chance of winning with all four of those states as 50-50 propositions.

We know how the election turned out. Obama was as successful as Mitt Romney had to be lucky in winning all four states.   
 

So what is different this time?

(1) it was not an open-seat election This one is!
(2) the President was predictable enough and perceived to be neither a spectacular success nor failure has done nothing to hurt any Democrat running to be his successor
(3) the Parties have little ideological overlap (and still do)
(4) the States were  remain sharply polarized in their partisan affiliation, and this was is  stable
(5) there were no new causes and there was no pervasive change in cultural patterns(still true)

Donald Trump, having no experience as an elected public official or as a senior military officer , will be the first nominee for President for one of the two main Parties since George H W Bush  who has never won a statewide office (and George H W Bush showed the effect in 1992 and Ford showed the effect in 1976) and has not been a senior officer or a winner of any public office since Hoover in 1928 (Hoover at the least was a Cabinet secretary). If Americans really want a non-politician, then Donald Trump is the choice. Hillary Clinton is trying to run as a "steady hand", "stay-the-course" type.

This is Hillary Clinton's election to lose and potentially Donald Trump's election to lose catastrophically badly.  Should Donald Trump win, he wins on Hillary Clinton's failure to campaign effectively.

Hillary Clinton has more political experience, and it is apparent that she has learned much by being a First Lady. She is linked to the most successful aspects of the Obama Administration (foreign policy) and makes few gaffes on public policy.

At this point, Donald Trump is behind Mitt Romney in consolidating likely votes for himself. Mitt Romney still lost.
 

 

[Dr. Arch]

Here's the old forecast from 2012. In particular, I miss the explicit "tipping point state" probabilities this time.

The Presidential election of 2012 was easy to model because

(1) it was not an open-seat election
(2) the President was predictable enough and perceived to be neither a spectacular success nor failure
(3) the Parties have little ideological overlap
(4) the States were sharply polarized in their partisan affiliation, and this was stable
(5) there were no new causes and there was no pervasive change in cultural patterns


So look at those factors. (1) An open-seat election usually hinges upon who wins contested primaries and caucuses, and it is often difficult to predict who will win that. Incumbents rarely lose nominating processes, so that leaves  the incumbent in an election usually his to lose.  That does not apply this time, of course. 22nd Amendment.

(2) History will almost certainly rate Barack Obama as an above-average President, even if most people saw him as a mixture of successes and disappointments. He did not expand his coalition as did FDR, JFK, or Reagan. But neither did he make many mistakes. "General Motors is alive and Osama bin Laden is dead" became the unofficial slogan of Democrats in 2012.

OK. Putting an end to the most dangerous economic meltdown in nearly eighty years and whacking the worst anti-American terrorist ever while avoiding scandals and military or diplomatic disasters is one way to get re-elected even if one loses about 20 states by 10% or more in the previous election. Barack Obama should have been re-elected and was.

(3) The 2010 election knocked out the conservative wing of the Democratic Party in the House of Representatives, and the Republican Party has nothing close to a liberal. Ideological choices between candidates are rather stark. Such makes individual choices easy

(4) Barack Obama winning Indiana in 2008 was something of a freak, something that happens when a state has a large industry (recreational vehicles)  whose market can be hit hard by a combination of an economic downturn, a credit crunch, and high gasoline prices, all of which hit in 2008. That is the only big swing in a state that Barack Obama won. The shift of the Deep and Mountain South from D to R between 1976 and 2008 was basically complete, and it now looks irreversible.

Barack Obama did not depend upon freakish conditions that could evaporate quickly to win.

(5) Tea Party? That's the closest thing to social change. There was no great new religious revival to create a new right-wing ascendancy and no leftish populist movement appearing from seemingly nowhere.

So I could come up with a model. First, recognize that seventeen states and the District of Columbia had not voted for any Republican nominee after 1988 and that there was no marked Republican drift in any of them. That's 243 of the 270 electoral votes that one needs for winning the Presidency, nearly 90% of the needed vote. Second, figure that three states had voted  only once for a Republican nominee for President in the same time, and showed no sign of going Republican this time. That's up to 258. That means that either Virginia, Ohio, Florida, or the combination of Colorado and Nevada would win the election/ Some other states?

Obama wasn't going to win Missouri without winning either Ohio or Virginia; he wasn't going to win Indiana without also winning Ohio; he wasn't going to win Arizona without also winning Colorado and Nevada; he wasn't going to win North Carolina without also winning Virginia; he wasn't going to win Georgia without winning Florida, North Carolina, and of course Virginia.  He also wasn't going to win Colorado without also winning Nevada.

So the 2012 election boiled down to four states -- Colorado, Florida, Ohio, and Virginia, any of which would decide the election. Those states are different enough and separated enough that there is no way to make an appeal that could resonate in all four of those states without
shifting America on the whole.

Treat winning any of the four states as  what statisticians call independent events (coin tosses and throws of a die are independent events), and my crude model held that President Obama had one chance in sixteen of losing if each of the states was a 50-50 proposition. I saw President Obama having a 93.75% chance of winning with all four of those states as 50-50 propositions.

We know how the election turned out. Obama was as successful as Mitt Romney had to be lucky in winning all four states.   
 

So what is different this time?

(1) it was not an open-seat election This one is!
(2) the President was predictable enough and perceived to be neither a spectacular success nor failure has done nothing to hurt any Democrat running to be his successor
(3) the Parties have little ideological overlap (and still do)
(4) the States were  remain sharply polarized in their partisan affiliation, and this was is  stable
(5) there were no new causes and there was no pervasive change in cultural patterns(still true)

Donald Trump, having no experience as an elected public official or as a senior military officer , will be the first nominee for President for one of the two main Parties since George H W Bush  who has never won a statewide office (and George H W Bush showed the effect in 1992 and Ford showed the effect in 1976) and has not been a senior officer or a winner of any public office since Hoover in 1928 (Hoover at the least was a Cabinet secretary). If Americans really want a non-politician, then Donald Trump is the choice. Hillary Clinton is trying to run as a "steady hand", "stay-the-course" type.

This is Hillary Clinton's election to lose and potentially Donald Trump's election to lose catastrophically badly.  Should Donald Trump win, he wins on Hillary Clinton's failure to campaign effectively.

Hillary Clinton has more political experience, and it is apparent that she has learned much by being a First Lady. She is linked to the most successful aspects of the Obama Administration (foreign policy) and makes few gaffes on public policy.

At this point, Donald Trump is behind Mitt Romney in consolidating likely votes for himself. Mitt Romney still lost.
 

 

Great analysis

[wolfsblood07]

Now I have 1 more reason to hope Trump wins.  Would be great to see Nate Silver be wrong again, and never hear about him again.

[President Johnson]

It's definitely too early to make a serious prediction. I think the election will be won within 2-4 points.

[Beefalow and the Consumer]

So the 2012 election boiled down to four states -- Colorado, Florida, Ohio, and Virginia, any of which would decide the election. Those states are different enough and separated enough that there is no way to make an appeal that could resonate in all four of those states without
shifting America on the whole.

You must be logged in to read this quote.

There has been movement in party affiliation due to demographic shifts.  Not seismic, but incremental.  Virginia looks a little bit more like Maryland.  North Carolina looks a little bit more like Virginia.  Georgia looks a little bit more cosmopolitan and less Southern every year, though this change is not enough to put it in play just yet.  The industrial Midwest is getting whiter, and more Republican, while the Sunbelt is getting more diverse, and more Democratic.

Combine this with the type of candidate Donald Trump is, and how he built his appeal, and the model may need tweaking.  Florida and Virginia may be too diverse to be within Trump's reach, and Ohio may be too white and industrial to be within Clinton's reach - in a close election.  I also believe Nate Silver's model for PA is built too heavily on outlier polls (+11 and +13 polls when there is a preponderance of tied or +1-4 polls). 

I think if there are four states that decide the election, they are now Colorado, Iowa, North Carolina, and Pennsylvania.  It still is, of course, Clinton's to lose, and Trump's to lose spectacularly, as you say, but those four states are now the critical path.

[hopper]

So the 2012 election boiled down to four states -- Colorado, Florida, Ohio, and Virginia, any of which would decide the election. Those states are different enough and separated enough that there is no way to make an appeal that could resonate in all four of those states without
shifting America on the whole.

You must be logged in to read this quote.

There has been movement in party affiliation due to demographic shifts.  Not seismic, but incremental.  Virginia looks a little bit more like Maryland.  North Carolina looks a little bit more like Virginia.  Georgia looks a little bit more cosmopolitan and less Southern every year, though this change is not enough to put it in play just yet.  The industrial Midwest is getting whiter, and more Republican, while the Sunbelt is getting more diverse, and more Democratic.

Combine this with the type of candidate Donald Trump is, and how he built his appeal, and the model may need tweaking.  Florida and Virginia may be too diverse to be within Trump's reach, and Ohio may be too white and industrial to be within Clinton's reach - in a close election. I also believe Nate Silver's model for PA is built too heavily on outlier polls (+11 and +13 polls when there is a preponderance of tied or +1-4 polls).  

I think if there are four states that decide the election, they are now Colorado, Iowa, North Carolina, and Pennsylvania.  It still is, of course, Clinton's to lose, and Trump's to lose spectacularly, as you say, but those four states are now the critical path.

I don't think Virginia will ever have the Black Population that Maryland has but I will give you that NOVA looks more like Maryland than most of rest of Virginia.

I think North Carolina and Virginia always vote pretty similar in Presidential Elections.

Georgia-Yeah I went to the Atlanta Suburbs last year and I expected a culture shock with a bunch of people with Southern Accents but there was nobody really with a Southern accent barely. Basically its a Southern area that's basically Northern now.

The Industrial Midwest is basically is staying D.

Hillary I think will win Ohio. Romney got 55% of the white vote according to Nate Cohn. So Trump would have to get 58% of the White Vote to win the state I think even in a 2012 electoral climate.

[Kingpoleon]

Didn't Nate Silver say Hillary had a >99% chance of winning Michigan in the primaries?

[Devout Centrist]

Didn't Nate Silver say Hillary had a >99% chance of winning Michigan in the primaries?

The model he built did. Rare events happen in state polling, especially in the primaries. Out of 58 primary contests, his model called 52. The other 6 usually didn't have solid polling or, like Democrats Abroad, were simply averaged out because of the lack of information on the electorate.

Basically, you're using one wrong answer to call Nate an idiot. Should we say he's an idiot for calling Indiana wrong in 2008?

Finally, it should be noted, as Nate noted himself, that 80% is not a sure thing. There is still a significant chance Trump could win.

[Gabagool102]

Nate Silver has been wrong about Trump all year, I wouldn't trust him to predict anything.

[100% pro-life no matter what]

In polls plus, Trump is closer to 30%.  Much of it is based on really old polls, too.  Like, his rationale for saying that SC could be close is two polls from November showing modest Trump leads.

[Attorney General, LGC Speaker, and Former PPT Dwarven Dragon]

Didn't Nate Silver say Hillary had a >99% chance of winning Michigan in the primaries?

The model he built did. Rare events happen in state polling, especially in the primaries. Out of 58 primary contests, his model called 52. The other 6 usually didn't have solid polling or, like Democrats Abroad, were simply averaged out because of the lack of information on the electorate.

Basically, you're using one wrong answer to call Nate an idiot. Should we say he's an idiot for calling Indiana wrong in 2008?

Finally, it should be noted, as Nate noted himself, that 80% is not a sure thing. There is still a significant chance Trump could win.

The worst part with Nate Silver and the 2016 primary is that he 1) continued to construct a "with endorsements" model for R's months after it had been proved useless, and 2) didn't integrate any sort of fundamentals into his models - there was no boost to Cruz in evangelical states, no boost to Sanders in white-dominated states, etc. It was literally just RCP with a boost to whoever was leading in national polls and/or endorsements. A five year old could have recreated his model.

[Devout Centrist]

Didn't Nate Silver say Hillary had a >99% chance of winning Michigan in the primaries?

The model he built did. Rare events happen in state polling, especially in the primaries. Out of 58 primary contests, his model called 52. The other 6 usually didn't have solid polling or, like Democrats Abroad, were simply averaged out because of the lack of information on the electorate.

Basically, you're using one wrong answer to call Nate an idiot. Should we say he's an idiot for calling Indiana wrong in 2008?

Finally, it should be noted, as Nate noted himself, that 80% is not a sure thing. There is still a significant chance Trump could win.

The worst part with Nate Silver and the 2016 primary is that he 1) continued to construct a "with endorsements" model for R's months after it had been proved useless, and 2) didn't integrate any sort of fundamentals into his models - there was no boost to Cruz in evangelical states, no boost to Sanders in white-dominated states, etc. It was literally just RCP with a boost to whoever was leading in national polls and/or endorsements. A five year old could have recreated his model.

Point 1 is obvious, but point 2 is patently false. Silver created a demographic model that gave Sandersan edge in solidly white states and Cruz a boost in states with more evangelicals. His model projected a close race in Michigan, yet the polls showed otherwise. I think the demographics mode put Michigan at either +2 Clinton or +2 Sanders.

[Attorney General, LGC Speaker, and Former PPT Dwarven Dragon]

Didn't Nate Silver say Hillary had a >99% chance of winning Michigan in the primaries?

The model he built did. Rare events happen in state polling, especially in the primaries. Out of 58 primary contests, his model called 52. The other 6 usually didn't have solid polling or, like Democrats Abroad, were simply averaged out because of the lack of information on the electorate.

Basically, you're using one wrong answer to call Nate an idiot. Should we say he's an idiot for calling Indiana wrong in 2008?

Finally, it should be noted, as Nate noted himself, that 80% is not a sure thing. There is still a significant chance Trump could win.

The worst part with Nate Silver and the 2016 primary is that he 1) continued to construct a "with endorsements" model for R's months after it had been proved useless, and 2) didn't integrate any sort of fundamentals into his models - there was no boost to Cruz in evangelical states, no boost to Sanders in white-dominated states, etc. It was literally just RCP with a boost to whoever was leading in national polls and/or endorsements. A five year old could have recreated his model.

Point 1 is obvious, but point 2 is patently false. Silver created a demographic model that gave Sandersan edge in solidly white states and Cruz a boost in states with more evangelicals. His model projected a close race in Michigan, yet the polls showed otherwise. I think the demographics mode put Michigan at either +2 Clinton or +2 Sanders.

Yeah, but those were buried in various articles. What he actually considered his official prediction was the polls plus and polls only models which had zero demographic considerations.

[Devout Centrist]

Didn't Nate Silver say Hillary had a >99% chance of winning Michigan in the primaries?

The model he built did. Rare events happen in state polling, especially in the primaries. Out of 58 primary contests, his model called 52. The other 6 usually didn't have solid polling or, like Democrats Abroad, were simply averaged out because of the lack of information on the electorate.

Basically, you're using one wrong answer to call Nate an idiot. Should we say he's an idiot for calling Indiana wrong in 2008?

Finally, it should be noted, as Nate noted himself, that 80% is not a sure thing. There is still a significant chance Trump could win.

The worst part with Nate Silver and the 2016 primary is that he 1) continued to construct a "with endorsements" model for R's months after it had been proved useless, and 2) didn't integrate any sort of fundamentals into his models - there was no boost to Cruz in evangelical states, no boost to Sanders in white-dominated states, etc. It was literally just RCP with a boost to whoever was leading in national polls and/or endorsements. A five year old could have recreated his model.

Point 1 is obvious, but point 2 is patently false. Silver created a demographic model that gave Sandersan edge in solidly white states and Cruz a boost in states with more evangelicals. His model projected a close race in Michigan, yet the polls showed otherwise. I think the demographics mode put Michigan at either +2 Clinton or +2 Sanders.

Yeah, but those were buried in various articles. What he actually considered his official prediction was the polls plus and polls only models which had zero demographic considerations.

It should have been included in the model, I agree.