Nate Silver: Dear Media, Stop Freaking Out About Donald Trump’s Polls

[Sprouts Farmers Market ✘]

No matter what happens with Trump, Nate Silver will end up writing an article at some point showing that he was "right" all along. Even if Trump wins, he'll argue along the same lines of when he says "Oh I wasn't wrong when I said this had a 1% chance of happening, because this was the 1 in the 100 times that it happened."

Nothing will beat when he was wrong about the college football playoff last year. It was just an event "impossible to predict" and the committee didn't act rationally...math is great, but it doesn't beat strong human intuition.

I'm not saying Trump is anywhere near a lock to win, but he has reasserted himself as the favorite.

So, we're going to rely on Nate Silver trolling when Paul Krugman gave a very strong explanation on the opposite position.

Paul Krugman is one of those guys who postures himself as an expert on everything. Some might call that a symptom of egomania. Soon, no doubt, he's going to start lecturing Muon2 and myself as to how to redistrict.

I almost missed this...thanks Max for putting in Irony Ore. Easily one of the best posts of the year! Only challenged by:

I'm receiving advice on Facebook which has been proven time and time again to be some of the best advice I can ever receive.

[King]

BOBBY JINDAL CAN STILL WIN IOWA

[The Other Castro]

Also, he says the true Iowa numbers are realistically like this:

CANDIDATE   SUPPORT IN IOWA
Undecided          80%
Donald Trump   5%
Ben Carson   4%
Ted Cruz           3%
Marco Rubio   2%
Jeb Bush           1%
Carly Fiorina   1%
Mike Huckabee   1%
Chris Christie   1%

He's pretty much just pulling numbers out from where the sun don't shine at this point.

[Bull Moose Base]

If no one were paying attention, the polls wouldn't be changing at all. Also, people who pay attention earlier logically seem like more likely voters. And Silver has himself in the past analyzed that early poll leaders in GOP presidential primaries have tended to win them. Of course, until 2012, that correlated with establishment support. But even in 2012, Romney maintained a stable polling lead in NH equivalent to Trump's, only losing it briefly when Gingrich surged about a month before the primary before quickly collapsing. So I buy that polls can fluctuate wildly up until voting but not that a candidate has no advantage over a rival he leads by 20 points.

[Vosem]

Of course Trump will lose, and it is in fact reflected in current polling. Head-to-heads have long shown Trump losing to everyone except Bush, who has long fallen into irrelevance anyway. Trump may very well win IA, and right now he looks like an overwhelming favorite to win NH. But eventually, the other candidates will coalesce around one person (right now, I suspect Cruz has the best shot by means of winning IA. But it may be that Trump will win IA/NH/SC and that an anti-Trump will only emerge later; it doesn't matter), and whoever that person is will smash Trump like the bug he is.

Trump may very well then run third-party; he's helping the establishment if he does. By running independent, he'll disassociate himself from the party, but he'll still bring his supporters to the polls, where they'll still vote Republican down-ballot. Hillary will be elected President, but Republican majorities in Congress won't be meaningfully dented and the Senate will be held. Then, they'll be able to stymie her at every turn, leading to another 2010/14-esque landslide in 2018, and then, after 3 Democratic terms, another one of equal intensity in 2020, leading to total Republican control of redistricting and continued dominance of state governments and the House for another decade (this way, even if a backlash comes in 2022/2024, its severity can be contained). But I'm projecting far into the unknowable future here. 2016 is still blurry in the windshield, never mind later years.

[Bacon King]

Of course Trump will lose, and it is in fact reflected in current polling. Head-to-heads have long shown Trump losing to everyone except Bush, who has long fallen into irrelevance anyway. Trump may very well win IA, and right now he looks like an overwhelming favorite to win NH. But eventually, the other candidates will coalesce around one person (right now, I suspect Cruz has the best shot by means of winning IA. But it may be that Trump will win IA/NH/SC and that an anti-Trump will only emerge later; it doesn't matter), and whoever that person is will smash Trump like the bug he is.

The problem with this analysis is that it's definitely not a given for any candidate to drop out early enough for support to coalesce around an anti-Trump, especially in the post-Citizens United era where superPACs can handle the heavy lifting with big ad buys.

As ironic as it is, Trump is in the same position that Romney held in 2012. There were quite a few candidates that would have defeated Romney in a head to head race at various points in the campaign, but the race was never just Romney vs. Anti-Romney (even as it got later in the primary season, Gingrich and Santorum were still acting as spoilers for each other) and this race will never be just Trump vs. one Anti-Trump.

You must be logged in to read this quote.

A Trump third-party candidacy is pretty much logistically impossible at this point, for the record, due to ballot access requirements, filing deadlines, and sore losers laws. I guess it would still be possible if he wanted to flush tens of millions of dollars into a nationwide operation dedicated entirely to just getting his name on the ballot every where, but the point of no return for even that option is rapidly approaching.

Also if he were to run as an independent I would not be surprised at all if Trump also got a bunch of his prominent supporters to run for Congress across the country as some sort of "MAKE AMERICA GREAT AGAIN PARTY" and it'd be comparatively easy to do if he already had ballot access operations funded in all fifty states anyway. This would maximize Trump's spoiler effect, as a "[Inks] you" to Republicans all across the board (just imagine Mike Ditka running against Kirk in Illinois, Hershel Walker challenging Isakson in Georgia, Hulk Hogan running in Florida, etc etc)

You must be logged in to read this quote.

"losing this election is good because it means we win the next one" is never a solid philosophy - a bird in the hand and all that

[○∙◄☻¥tπ[╪AV┼cVê└]

Webb was supposed to be the anti Hillary. And less than 2 months ago, 538 said Jindal could win Iowa. Joke site.

[Vosem]

Of course Trump will lose, and it is in fact reflected in current polling. Head-to-heads have long shown Trump losing to everyone except Bush, who has long fallen into irrelevance anyway. Trump may very well win IA, and right now he looks like an overwhelming favorite to win NH. But eventually, the other candidates will coalesce around one person (right now, I suspect Cruz has the best shot by means of winning IA. But it may be that Trump will win IA/NH/SC and that an anti-Trump will only emerge later; it doesn't matter), and whoever that person is will smash Trump like the bug he is.

The problem with this analysis is that it's definitely not a given for any candidate to drop out early enough for support to coalesce around an anti-Trump, especially in the post-Citizens United era where superPACs can handle the heavy lifting with big ad buys.

As ironic as it is, Trump is in the same position that Romney held in 2012. There were quite a few candidates that would have defeated Romney in a head to head race at various points in the campaign, but the race was never just Romney vs. Anti-Romney (even as it got later in the primary season, Gingrich and Santorum were still acting as spoilers for each other) and this race will never be just Trump vs. one Anti-Trump.

This race will never be Trump vs. one Anti-Trump, but it doesn't need to be. Romney won a string of very narrow victories over Santorum by utilizing his superior fundraising and ground game/support for the establishment, which is something Trump will be unable to do against anyone who emerges as the Anti-Trump. Also, I question that Gingrich and Santorum were acting as spoilers; after the February 7 primaries (Colorado/Minnesota/Missouri), Santorum emerged as the challenger to Romney and Gingrich afterwards rarely hit double-digits outside the South (where Santorum largely won anyway). Of 16 Romney victories over the period between February 7 and Santorum's drop-out on April 10, Gingrich+Santorum>Romney was only true in four states (Wyoming, Michigan, Ohio, and Wisconsin). Also, while both were "anti-Romney", Santorum ran a strongly socially conservative campaign while Gingrich's was more fiscally conservative/personalist, and I don't know how many of Gingrich's voters would've sat out or had someone other than Santorum as a second choice. The point being: saying Gingrich spoiled the nomination for Santorum isn't really backed up by the data.

What you're saying about the superPACs is true, but this is not a field of lots of strong candidates with well-funded superPACs trampling on each other's toes: the field is overwhelmingly under 5%, both nationally and in the early states. Also, if someone begins to emerge as the Anti-Trump, the pressure to back off from donors, for an establishment politician (say, someone like Bush or Rubio; this may not apply to, say, Cruz, and certainly not to Carson) will be much greater than the pressure on Gingrich to step aside for Santorum ever was.

You must be logged in to read this quote.

A Trump third-party candidacy is pretty much logistically impossible at this point, for the record, due to ballot access requirements, filing deadlines, and sore losers laws. I guess it would still be possible if he wanted to flush tens of millions of dollars into a nationwide operation dedicated entirely to just getting his name on the ballot every where, but the point of no return for even that option is rapidly approaching.

I know this, but Trump's begun floating the possibility again and it's been brought up on other threads.

Also if he were to run as an independent I would not be surprised at all if Trump also got a bunch of his prominent supporters to run for Congress across the country as some sort of "MAKE AMERICA GREAT AGAIN PARTY" and it'd be comparatively easy to do if he already had ballot access operations funded in all fifty states anyway. This would maximize Trump's spoiler effect, as a "[Inks] you" to Republicans all across the board (just imagine Mike Ditka running against Kirk in Illinois, Hershel Walker challenging Isakson in Georgia, Hulk Hogan running in Florida, etc etc)

This I doubt, on the other hand. Trump's always been a solo act and tries to avoid fraternizing with other politicians (even if they would be "outsider"-like politicians). I don't think he would recruit others to run for down-ballot offices (certainly not for House seats), and if he did it would be an irregular patchwork, probably of Trumpian yes-men who would run far behind Trump. The Trump does not like to be upstaged.

You must be logged in to read this quote.

"losing this election is good because it means we win the next one" is never a solid philosophy - a bird in the hand and all that

You're right, but I do think keeping control of Congress in this election for Republicans is more significant than winning the Presidency -- ultimately foreign policy is very much predicated on the individual holding the Presidency and the circumstances of the era as opposed to domestic politics, and the President takes a backseat to Congress on domestic policy. I do think Trump running for President as an independent, unlikely though that scenario is, would help Republicans in congressional races, unless he deliberately recruited spoiler candidates (which seems out of character and a tactic of dubious success anyway).

[Orthogonian Society Treasurer]

Silver is still in the denial stage of grief.

[mencken]

Let's try an actual statistical analysis here. RCP has been generous enough to provide comparable data from the 2008 and 2012 cycles (which both appeared comparably chaotic at the time). The data indicates that roughly half the time, the candidate ends up receiving as his final caucus total within ~5% of his polling average 70 days prior. The only trouble is that this distribution is decidedly skewed right, so you also have extreme examples like the collapse of Cain and Giuliani and the meteorotic rises of Huckabee and Santorum. In light of this information, it might make more sense to survey the field for frontrunners showing signs of wear from media scrutiny (possibly Carson?) and for ascendant social conservatives (possibly Cruz?). Other than that, I see little reason to believe that polling 70 days out will not be reasonably good estimators of final performance. The polls are going to be a good indicator of the final performance of most candidates, and an atrociously poor indicator for two or three of them.

EDIT: I did 30 dumb random simulations (i.e. not taking into account the plausibility of the outcomes) based on the above described asymmetric distribution of past Iowa caucus performances. Trump won a bit less than half the time, Carson a quarter of the time, Cruz a tenth of the time, and the rest scattering. However, even this is almost certainly too bearish for these candidates, since I guarantee that the chance of Bush, Kasich, Fiorina, or Christie winning Iowa is zero.

[dudeabides]

John Kerry was at 4% in the polls 27 days before the 2004 Iowa Caucus. Anything can happen between now and Iowa. I think Nate Silver is correct and I thank Torrie for posting this. Marco Rubio is rising in the polls. Someone like Jeb Bush, John Kasich, or even Rand Paul could emerge as the winner in Iowa, New Hampshire, or South Carolina. My personal guess is Marco Rubio will compete in Iowa and New Hampshire and that he has a good chance at winning in South Carolina. Governor Bush has been damaged by Donald Trump and his debate performances, Governor Kasich is investing too much into one state, and Senator Paul is now trying to get his father's supporters after spending two years reaching out to the establishment. But who knows; Governor Bush still has a lot of money and appeal to the political center, Governor Kasich comes across as the everyman, and Senator Paul could tap into the libertarian vote in New Hampshire. It's early still.

[DrScholl]

Numbers do matter, particularly the top line. Silver's logic here can apply to any candidate, not just Trump and the idea that Trump is guaranteed to drop at some point has proven faulty for months now. At this point, Trump is a real contender for the nomination.

[TheDeadFlagBlues]

Now, children, calm down. Does anybody think Trump does as well with Likely Voters as he does with Registered Voters? Sure Trump may be high energy, but his base is low information - just the type to be more interested in the theatrics as opposed to the substance, and then fail to do something as boring as going to the polls. Polling stations are not entertainment venues.

I'm looking forward to the next chapter of this drama, when Cruz and Trump claw at each other. That should be happening soon.

The thing is Torie, Trump's ultimate performance is besides the point. What's really in doubt is the "conventional wisdom" that the establishment's preferences are meaningful. At this point, it's increasingly clear that they're not all that meaningful.

Trump, Carson and Cruz are pulling a combined 58% of vote in the polling aggregators. Rubio, Christie, Bush and Kasich are pulling a combined 22% of the vote in the polling aggregators. It may very well be that Trump's support erodes but, if it does, it's going to flow to Cruz and not to Rubio.

It's rather pathetic that so-called "moderate" Republicans are grasping at straws like this. Look, we learned in 2012 that an experienced frontrunner with an ample warchest and perfect hair could nearly be bested by a low-energy loser wearing a sweater-vest and a profound bestiality-type promising to build moonbases. In 2016, the contest will likely be between a low-IQ Cuban, who supports "amnesty", has connections to the drug trade and uses the word "lavenous" in his tweets, a charismatic media professional and an intellectual/forensic powerhouse who singlehandedly shut-down the government + wrote a undergraduate thesis that could be published in a law review publication. Do you really think that Marco "enchilada amnesty boy" Rubio is going to beat Ted "stentorian" Cruz or The Donald, King of the Media? Good luck Torie, it's not going to happen!

Note: I think Trump is going to collapse and that Cruz is going to be the nominee. Of course, I also think that the GOP is more likely to nominate Trump than Rubio.

[I spent the winter writing songs about getting better]

Some of Silver's points are valid, but this idea that Trump's support is a giant bubble that just has to burst at some point is quite discredited. Remember this is a candidate who is mocked by the media and talking heads all the time, people should be if anything embarrassed to support him, he's not the source of bandwagon support to be cool. And yes, his "real" numbers in Iowa are clearly just something he pulled out of his ass. Also he makes the argument that no candidate like Trump has ever won a major nomination before (true) and then says in the footnote that there's even stronger precedent for Cruz or Sanders, comparing them to Barry Goldwater and George McGovern? WTF? Cruz as Goldwater I can see but anyone who knows anything about McGovern knows that he was nothing like Sanders...Howard Dean 2004 might work as a comparison, but Sanders?

[RI]

Also he makes the argument that no candidate like Trump has ever won a major nomination before (true)

That's the thing about politics and other low-observation phenomena: trends are true until they suddenly aren't.

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

Some of Silver's points are valid, but this idea that Trump's support is a giant bubble that just has to burst at some point is quite discredited. Remember this is a candidate who is mocked by the media and talking heads all the time, people should be if anything embarrassed to support him, he's not the source of bandwagon support to be cool. And yes, his "real" numbers in Iowa are clearly just something he pulled out of his ass. Also he makes the argument that no candidate like Trump has ever won a major nomination before (true) and then says in the footnote that there's even stronger precedent for Cruz or Sanders, comparing them to Barry Goldwater and George McGovern? WTF? Cruz as Goldwater I can see but anyone who knows anything about McGovern knows that he was nothing like Sanders...Howard Dean 2004 might work as a comparison, but Sanders?

Sanders and McGovern have one thing in common - they're both completely unelectable. Granted, with today's polarization Sanders would do much, much better than McGovern, but neither can reach 270 barring a perfect storm.

[○∙◄☻¥tπ[╪AV┼cVê└]

Also he makes the argument that no candidate like Trump has ever won a major nomination before (true)

That's the thing about politics and other low-observation phenomena: trends are true until they suddenly aren't.

A woman and a non Christian have never won a major party nomination before. Congrats, Democratic nominee Martin O'Malley.

[Asian Nazi]

Also he makes the argument that no candidate like Trump has ever won a major nomination before (true)

That's the thing about politics and other low-observation phenomena: trends are true until they suddenly aren't.

A woman and a non Christian have never won a major party nomination before. Congrats, Democratic nominee Martin O'Malley.

Don't be an idiot

[eric82oslo]

As ironic as it is, Trump is in the same position that Romney held in 2012. There were quite a few candidates that would have defeated Romney in a head to head race at various points in the campaign, but the race was never just Romney vs. Anti-Romney (even as it got later in the primary season, Gingrich and Santorum were still acting as spoilers for each other) and this race will never be just Trump vs. one Anti-Trump.

The difference is that Romney was consistently the best debater in every debate in 2012, while Rubio has held that same position in every debate so far this season.

[SUSAN CRUSHBONE]

Also he makes the argument that no candidate like Trump has ever won a major nomination before (true)

That's the thing about politics and other low-observation phenomena: trends are true until they suddenly aren't.

A woman and a non Christian have never won a major party nomination before. Congrats, Democratic nominee Martin O'Malley.

a non-christian won the nomination in '08

[Pandaguineapig]

As ironic as it is, Trump is in the same position that Romney held in 2012. There were quite a few candidates that would have defeated Romney in a head to head race at various points in the campaign, but the race was never just Romney vs. Anti-Romney (even as it got later in the primary season, Gingrich and Santorum were still acting as spoilers for each other) and this race will never be just Trump vs. one Anti-Trump.

The difference is that Romney was consistently the best debater in every debate in 2012, while Rubio has held that same position in every debate so far this season.

Also Trump will never have the fundraising or organization that Romney had in the primaries

[BlueSwan]

Also, he says the true Iowa numbers are realistically like this:

CANDIDATE   SUPPORT IN IOWA
Undecided          80%
Donald Trump   5%
Ben Carson   4%
Ted Cruz           3%
Marco Rubio   2%
Jeb Bush           1%
Carly Fiorina   1%
Mike Huckabee   1%
Chris Christie   1%

He's pretty much just pulling numbers out from where the sun don't shine at this point.

Silver just went full Dick Morris with this nonsense.

He fails to distinguish betweens true "undecideds" and the fact that some people change their minds during a campaign. The fact that of the 30% that currently support Trump, a fair bunch of them might very well end up supporting someone else does not mean that they are "undecided". They have decided for now, but they might chance their mind. The difference is that they have to be swayed to change their mind. Up until now, increased exposure hasn't hurt Trump one iota.

Anyway, Silver seems to have forgotten what got him his glory in the first place. It was ignoring gut feelings and the usual punditry and sticking to crunching numbers. Silver always argued that polls matter and now he seems to be arguing the opposite. He seems to be operating from his initial assumption that "Trump can't win" - like most other pundits - and desperately looks for any indication or argument that can support this gut feeling. That is just bad science and the exact opposite of what Silver is known for.

Now, I think there are still plenty of good reasons why Trump might not win the nomination, but saying that his chances are way below 20% at this point rings untrue when you look at his constant domination of all relevant polls since july.

[heatmaster]

Now, children, calm down. Does anybody think Trump does as well with Likely Voters as he does with Registered Voters? Sure Trump may be high energy, but his base is low information - just the type to be more interested in the theatrics as opposed to the substance, and then fail to do something as boring as going to the polls. Polling stations are not entertainment venues.

I'm looking forward to the next chapter of this drama, when Cruz and Trump claw at each other. That should be happening soon.

An awful lot of pollsters are gonna be major league putz's, ah sure won't they be the entertainment fodder that Trump is now? Trump actually buys into the polls himself, he's tight fisted and lazy to commission a poll of his own and find out how hard or soft his support is. The polling data is skewered and his support is over inflated. Answer me this question who buy into the polling showing Trump in the lead.  How come the likes of Bush, Fiorina and Rubio are hanging around? Why are they not panicking?  Why is Carson beginning to fade? Why is Trump's support not increasing?  I think his support was like a tsunami, it has a certain shelf life and he like Carson is looking at a situation of diminished returns. The public will grow bored of this Trump fad that is. He served a purpose and the likes of Bush, Rubio and Fiorina will begin to flourish and there rise in the polls will be as inevitable as Trump's fall.

[Mister Mets]

Also, he says the true Iowa numbers are realistically like this:

CANDIDATE   SUPPORT IN IOWA
Undecided          80%
Donald Trump   5%
Ben Carson   4%
Ted Cruz           3%
Marco Rubio   2%
Jeb Bush           1%
Carly Fiorina   1%
Mike Huckabee   1%
Chris Christie   1%

He's pretty much just pulling numbers out from where the sun don't shine at this point.

Silver just went full Dick Morris with this nonsense.

He fails to distinguish betweens true "undecideds" and the fact that some people change their minds during a campaign. The fact that of the 30% that currently support Trump, a fair bunch of them might very well end up supporting someone else does not mean that they are "undecided". They have decided for now, but they might chance their mind. The difference is that they have to be swayed to change their mind. Up until now, increased exposure hasn't hurt Trump one iota.

Anyway, Silver seems to have forgotten what got him his glory in the first place. It was ignoring gut feelings and the usual punditry and sticking to crunching numbers. Silver always argued that polls matter and now he seems to be arguing the opposite. He seems to be operating from his initial assumption that "Trump can't win" - like most other pundits - and desperately looks for any indication or argument that can support this gut feeling. That is just bad science and the exact opposite of what Silver is known for.

Now, I think there are still plenty of good reasons why Trump might not win the nomination, but saying that his chances are way below 20% at this point rings untrue when you look at his constant domination of all relevant polls since july.

There is a major difference between primary polls and General Election polls.

The difference between the Republican and Democrat is greater than the difference between two Republicans (or two Democrats) which means that primary voters are more likely to change their mind. As a result, primary polls don't matter as much.

[Absentee Voting Ghost of Ruin]

Also he makes the argument that no candidate like Trump has ever won a major nomination before (true)

That's the thing about politics and other low-observation phenomena: trends are true until they suddenly aren't.

Yeah, I was briefly shocked to see Nate Silver pulling out such an... innumerate argument, then remembered that he jumped the shark some time ago.