Primary Model Predicts Trump Victory | Hacker News Reader

Primary Model Predicts Trump Victory

primarymodel.com

26 pointscglace75 comments

My sanity check alarms started going off here:

> For the record, the Primary Model, with slight modifications, has correctly predicted the winner of the popular vote in all five presidential elections since it was introduced in 1996 ... Also note that for all elections from 1912 to 2012 the Primary Model picks the winner, albeit retroactively, every time except in 1960.

It can't be true that this model correctly predicts both the election winner and the popular vote winner, since Bush won one and lost the other in 2000.

I am extremely skeptical of this model, partly because I find the idea that voters pay attention to the primaries, but not to anything that happens later, very implausible. But much more importantly, it has a huge overfitting problem. There are tons of degrees of freedom, not just in the model weights but in the choice of which primaries to consider.

The authors picked NH in previous elections, and this year fairly arbitrarily decided to include SC as well. It's no surprise that you can make good retroactive predictions when you have many effective model parameters to play with.

> But much more importantly, it has a huge overfitting problem.

Exactly. I was thinking about how easy it would be to create a model that accurately predicted the last 100 years' election results that would likely fall flat on its face in the next.

My theory, many people are actually voting for Trump, for whatever reason it may be, but they will not tell others. Polls, friends, community, etc. It seems many communities, including SV, have tried to ostracize those with opposing viewpoints. Trump's strength is built on that, whether an individual will be public about it or not. I read somewhere (sorry no source) that this election has seen the fewest bumped stickers/yard signs in recent history.

Edit: Remember, this is HN, not Reddit, keep it civil. And also keep in mind, odds are pretty low you will change anybody's view at this point, especially over comments ;)

If that were true, then down-ticket Republicans wouldn't be seeing as much of a sag in their support. And that wouldn't account for the narrow margins in states where the Republicans usually have stronger positions. And, there's still a fairly high level of undecideds and 3rd-party voters who might switch late (as usually happens)--polls seem to be indicating that these are swinging favorably to Clinton, and not to Trump.

Look into the "shy tory factor". https://en.wikipedia.org/wiki/Shy_Tory_Factor

People are more likely to lie about their voting intentions when the media create a climate of moral stigma around their candidate. I am sure this is going to happen in the uk too. The Democrats have run the wrong campaign.

Ahh, exactly. I think this kind of thing is quite dangerous. When people become afraid to say what they think and/or get berated when they do, it's a very dangerous climate. For any side. All of the sudden people start to believe something they actually don't believe in because of this stigma. The current state of the media is the worst. Do I know how to fix it....I wish. We live in a world of headlines...And it's too easy to put almost any information you want to appeal to your viewer base, despite how inaccurate it may be. Tough problem.

I saw a great "life pro tip" recently: If the News it telling you how to feel, it's not news. Always remember that.

Sounds like you are proposing something similar to the Bradley Effect. From wiki, it was suggested that "white voters give inaccurate polling responses for fear that, by stating their true preference, they will open themselves to criticism of racial motivation. Members of the public may feel under pressure to provide an answer that is deemed to be more publicly acceptable, or 'politically correct'. The reluctance to give accurate polling answers has sometimes extended to post-election exit polls as well."

It's not a white/black choice this time around but the social exclusion in some circles of people who support Trump might see something similar in play. That said, it would need to be a pretty big effect to make a difference if the gap is really 6 or 7 points.

[1] https://en.wikipedia.org/wiki/Bradley_effect

> My theory, many people are actually voting for Trump, for whatever reason it may be, but they will not tell others. Polls, friends, community, etc.

I know quite a few in this camp.

> It seems many communities, including SV, have tried to ostracize those with opposing viewpoints. Trump's strength is built on that, whether an individual will be public about it or not. I read somewhere (sorry no source) that this election has seen the fewest bumped stickers/yard signs in recent history.

I'd believe that. It'd be a combination of people not actually liking Clinton enough to want to show support and people being scared of actively showing support for Trump. The Clinton camp has done a decent job of bucketing all Trump supporters as racists assholes so it naturally makes it uncomfortable for people to voice their support. I'm pretty sure that's working as intended.

Wouldn't this be the sort of indirect, gut level factor that betting markets would capture better than anything else? And they have Trump winning at around 15-20%.

The exact same thing has been said about 2008 in regards to Obama's popularity in popular media and the idea that it was socially unacceptable to oppose him. It was wrong then and the final outcome closely resembled the polls leading up to the election [0]. I suspect this theory is wrong in this case as well.

[0] http://www.realclearpolitics.com/epolls/2008/president/natio...

I don't think the media was colluding against McCain the way they are against Trump. There's a real stigma being created about supporting Trump.

> There's a real stigma being created about supporting Trump.

Would you consider this unwarranted? Not necessarily in a political sense, but by way of the man's earned & self-professed reputation.

Don't get me wrong - Trump has certainly earned negative attention on his own. But the DNC has worked with the supposedly "unbiased" media to focus a disproportionate amount of negative energy on him, and they've hired people to harass and assault his supporters. These are dirty tactics they've employed to try to win the election. So yes, the incredible amount of stigma against him is unwarranted.

At what point should the media be biased against a person?

How bad do they have to be before the media is given a free pass on being against that person? Is it never acceptable to you?

If it's never acceptable for the media to be against a candidate, they're going to be completely worthless in a variety of situations.

I think the media is bias in the credibility it gives to many of Trump's claims and proposals. For instance, Facebook CEO stated that articles regarding Trump's proposal for a Muslim ban technically qualify as hate speech and would normally be banned. However, due to the circumstances of the proposal coming from a presidential candidate from a major party, it is allowed.

Care to supply examples?

But yeah, ideally the media should as unbiased as it is realitistically possible. If things are truly bad, then they'll be recognized as such without the need to distort the truth. The media can only hurt itself by willfully manipulating the message, since that undermines the public trust on journalists, whose impact on the public is crucially dependent on that trust.

It's a "boy who cried wolf" kinda scenario, or sorta like how everyone rolls their eyes when an "X causes cancer" headline pops up. If journalists choose to trade their credibility for anything else (eye-popping headlines, ads revenue, or subserviency to whomever is bank rolling them) they're paying it with their relevancy in the minds of the public. And if journalists really conceive themselves as on a moral mission, that's the last thing they should do away with.

Alright, basically I didn't want to jump straight to Hitler, because it's so overused, overdramatic and distracting. But do they actually gain credibility by being "fair" to the campaign of Hitler back from the dead? They don't.

When they call a spade a spade, the media gains credibility in the next election between two normal candidates.

I'm sure someone will hate this comparison. But really, the republican candidate is a fucking monster this time. It's never been this bad. It's not even a bad comparison anymore. I've always hated when people use it, but this guy wants to do some scary shit.

Why would people need to lie to the professional pollsters, though? If the election were held today and Trump won, it would be way outside of the margin of error on the polls.

Garbage in, garbage out. No need for anyone to lie.

Take for example the most recent Rasmussen showing Trump winning by 2%. To quote rasmussenreports.com:

"Typically, calls are placed from 5 pm to 9 pm local time during the week."

"After the surveys are completed, the raw data is processed through a weighting program to ensure that the sample reflects the overall population in terms of age, race, gender, political party, and other factors. The processing step is required because different segments of the population answer the phone in different ways. For example, women answer the phone more than men, older people are home more and answer more than younger people, and rural residents typically answer the phone more frequently than urban residents."

Seriously, who answers spam calls at home between 5pm and 9pm? Anyone here? At all? rasmussen and all other poll operators keep their transfer function secret. Its entirely possible their transfer function bears no relationship with who it actually going to show up and vote in November.

Lets pick on a semi-public transfer function from IDB. Their poll shows a dead tie Trump and Clinton both exactly 41%. However check the fine print about their transfer function "Party identification breakdown: (Unweighted) 261 Democrats/261 Republicans/275 Independents; (Weighted) 291/235/271."

Seriously, they believe 23% of republicans are going to stay home on voting day? Seriously? I mean when I was a kid, even Pravda would have been embarrassed to publish something like that. Obviously the editorial board saw a Trump win in the results and ordered the Republican weighting decreased until the printable result was a tie at worst.

Pew claimed in 2015 "When the partisan leanings of independents are taken into account, 48% either identify as Democrats or lean Democratic; 39% identify as Republicans or lean Republican. The gap in leaned party affiliation has held fairly steady since 2009, when Democrats held a 13-point advantage (50% to 37%)." So Pew implies the IDB transfer function should be wildly different, something like 48%/39%/13% than what they actually used to "surprisingly" provide a predetermined result.

Note that for the actual vote in November, plus or minus the usual dead people always voting democratic, the transfer function is very close to 1:1. My vote is counted as one vote, not some weird fractional part of a vote depending on my race or whatever the polling companies are doing.

You should go through the exercise of posting a mea culpa after the election (assuming the IBD poll is in line with the election results). Or just write the mea culpa on a piece of paper that you burn.

What you are doing here is taking your facile analysis of what their skew correction might be and using it to speculate that a relatively conservative, information focused publication is trying to influence the vote by lying in a way that won't really do much to influence the vote.

>My theory, many people are actually voting for Trump, for whatever reason it may be, but they will not tell others.

http://fivethirtyeight.com/features/trump-supporters-probabl...

http://fivethirtyeight.com/features/what-could-the-polls-be-...

Probably not.

http://www.newyorker.com/magazine/2016/10/10/in-the-heart-of...

"So when they see that the media élite is driven out of their mind at the success of Donald Trump it makes them want to root for him. It’s like giving the middle finger to the rest of the country."

I've heard this from many sources but do you have anything to back up the assertion?

That's the Brexit Theory. We'll see if it applies to Trump.

Prior to the vote, the polls were at a dead-heat meaning the outcome was within margin of error.

http://www.economist.com/blogs/graphicdetail/2016/06/britain...

I don't think any model has accounted for how personally distasteful Trump is. Regardless of his politics (which I find abhorrent as well, but that's another matter), there are a lot of Republicans who refuse to support him simply because he has shown himself to be unqualified to lead ANYTHING.

Even if you're an ardent Trump supporter, you have to admit that Hillary showed how easily manipulated he is in the 3 debates. She just poked at his soft spots and tried to get a reaction, and it worked.

None of this is about electing Hillary; it's entirely about keeping a petulant child out of the most powerful job in the country. Is she in bed with special interests? Definitely. Was she trying to avoid FOIA documentation requirements with a private e-mail server? Of course she was. Are the American people sick of elitist political dynasties like the Clintons and Bushes? Absolutely. These things are all true of most politicians.

But in contrast to Trump, Hillary is an adult. I trust Hillary to be able to put aside her ego and do what's right for the country. I don't feel the same about Trump; and I don't want to get involved in World War 3 because some world leader made a comment about Trump's hair.

The GOP should have won this election handily -- but they've painted themselves into an untenable platform that simultaneously is built on racism and diversity. No candidate could form a cohesive vision out of the nonsensical platform they put forward (limited government + legislating morality; business-friendly + anti-immigration; guns should be legal but not drugs; etc.) so they ended up with a nonsensical candidate and a disaster of a party. The party is so fractured over Trump that even if the GOP can retain a majority in the house and senate, the Democrats will likely be able to find enough Republicans to break ranks to get legislation passed.

But this election isn't about politics. It's about whether we are angry enough at the whole system to burn it all down. This should serve as a wake-up call that over 40% of the country thinks things are so bad that they're willing to elect a would-be dictator.

"I trust Hillary to be able to put aside her ego and do what's right for the country."

s/the country/herself/g

You are right. http://i.imgur.com/SKBitNQ.jpg

> a would-be dictator.

Why that choice of word?

Because he has shown time and again that he thinks the rules do not -- and should not -- apply to him. A candidate running on a law-and-order platform who thinks he is above the law is essentially the textbook version of a dictator.

"he has shown time and again"

It should be very easy to list examples, then. I found the most stunning thing about the debate topics was everything Trump was being accused of was at most shady but legal, and everything Hillary was being accused of would send anyone but a Clinton to prison.

The IRS is happy with his taxes therefore he must be a crook. His lawyers followed every bankruptcy law and the judge agreed he did nothing wrong therefore he must be a crook. The amount of money he paid in taxes is legal but Hillary thinks any tax rate below 100% is morally wrong therefore he's a crook. He should have known that in a decade he'd be running for president and made sure that a subcontractor of a subcontractor of a general contractor bought USA steel instead of perfectly legal Chinese steel therefore he's a crook. A billionaire supposedly molests a dozen women for as long as Bill Clinton has been in the news for doing worse, but they all forgot about the tremendous billionaire payday for decades until three weeks before the election, all at the same time, huh, what are the odds of that, I'm sure it must all be true? Oh and he might have used a naughty (but legal!) word a couple times in his life.

Luckily we have an incredibly sanctimonious holier than thou candidate who despises half the country she supposedly wants to lead, is above campaign finance law, above confidential documents law, is a child rapist enabler both in home life and professionally as an attorney, her death count in the middle east must be a million or so, but "Hillary" so that makes it acceptable and magically somehow she's not above the law nor a dictator nor a madwoman because she's a globalist elite pawn. Clearly she is on the opposite of a law and order platform, we agree there. If she were a law and order candidate, to be self consistent she'd have to put herself and her husband in prison...

So where's this "time and again"? An unpaid parking ticket, perhaps? An overdue library book? Anything at all?

> thou candidate who despises half the country she supposedly wants to lead,

Show me a candidate on either side who hasn't made these kinds of statements. This is more a symptom of our partisan divide than any one candidate.

> is above campaign finance law

You can't run for president without bending the campaign finance rules. Again, not ideal, but if we want to fix campaign finance let's fix it legislatively. Neither side wants to deal with it, so it never happened.

> above confidential documents law,

Again, she did no worse than any of the previous GOP administrations. Not saying we shouldn't hold her to a higher standard, but nothing she did here is truly shocking.

> is a child rapist enabler both in home life and professionally as an attorney

Ok, WHOA. I don't think either of these happened. You can't count professional work as a defense attorney because the lawyer is there to ensure that her client receives a fair trial. It's in everyone's best interests for her to mount a competent defense because it should clarify reasonable doubt. Just because she may have defended a child rapist doesn't mean she supports him -- she was an agent of the court playing a role in a system that ensures everyone has the right to a fair trial.

Bill may be a horndog, but that's beside the point. It's not his wife's fault; nor is it any of your business how they handle their marital problems. And I don't recall any allegations of him ever assaulting anyone underage...

> her death count in the middle east must be a million or so

Ok, just stop. Even if you were to blame Hillary for every combat-related death in the middle east and north Africa during and after her tenure as Secretary of State, they don't even add up to more than 500,000. Exclude Syria (you can only criticize Hillary for inaction here - and what exactly do we have to gain getting involved in a proxy war between Saudi Arabia, Iran and Russia?) and Libya (not gonna touch it with all the hate from the Benghazi witch hunt where no wrongdoing was found after a few rounds of congressional investigations, but needless to say you're giving Hillary way too much credit if you blame the entire situation in Libya on her) you don't even get to 10,000. Yes, the refugee crisis is real, but there's not much we can do about it from the US. The EU has to figure that one out on their own, unfortunately.

Honestly, the world is in an unprecedented period of peace right now. Major global conflicts are relatively low-burn affairs where the death count is measured in thousands per year rather than millions. Our goal should be to keep it that way, not go and start the next big war or escalate a smaller one.

>And I don't recall any allegations of him ever assaulting anyone underage...

Please google "Lolita Express" (one link here [1]). His friend, Jeff Epstein, who organized these trips is a convicted pedophile.

In 2014, another young woman filed a lawsuit claiming that Epstein used her as a sex slave for his powerful friends—and that she’d been at parties on his private island with former President Clinton.

http://www.washingtontimes.com/news/2016/may/14/bill-clinton...

Epstein, Trump's-formerly-acknowledged longtime friend who he now denies knowing, is also the alleged facilitator in the case in which Trump is accused of repeatedly and forcible raping a young teen.

So, I'm not sure reports that Bill Clinton may have been present at an Epstein party is exactly a damning indictment of how Hillary Clinton is somehow getting away with things worse than what Trump is accused of.

It's all unsubstantiated, honestly. I don't believe any of the rumors on either side, because both sides have a huge incentive to lie and insinuate bad things about the other. It's dirty politics 101, and the Clintons invented it (say what you want about them; but they are masters of dirty politics -- which you have to be to become president).

I judge them based on what they've said; and Hillary has offered up some good ideas and a whole lot of rhetoric while Trump has spouted inconsistent positions, many of which are frightening to me. Furthermore, when you do even a cursory bit of research on their statements, it becomes clear that Trump either doesn't know the truth or simply doesn't care. Add to that the ease with which Clinton manipulated him in the debates (she goaded him into a fight which plays well to his 30% base, but makes him look very unpresidential to the rest of the voter base). If Clinton could manipulate him so skillfully, how would he react to provocation elsewhere on the international stage?

Regardless of Trump's politics (which I happen to disagree with), he as a person has shown that he is unfit to hold public office at any level. He seems to have little regard for ethics and fair play -- his idea of "fair" is when everything goes in his favor, and everyone is conspiring against him when it doesn't. He is little more than a con-man reality TV star, while Clinton has lived a life of public service.

"The Primary Model relies on presidential primaries as a predictor of the vote in the general election. For the record, the Primary Model, with slight modifications, has correctly predicted the winner of the popular vote in all five presidential elections since it was introduced in 1996. In recent elections the forecast has been issued as early as January of the election year."

If I'm reading it correctly this isn't a new prediction; this is a new article about the Primary Model's existing prediction dating from March 7 of this year.

If Trump had been a normal candidate, he would have beat Clinton. No tawdry sex scandals, crazy public fights with Ms Universe, gold star parents, leading members of his own party. If he had raised money on the same scale as Romney or established a ground game and analytics like Clinton. Based solely on primary data, yes, he probably would win.

Wait wait wait... "for elections since 1952 only the New Hampshire Primary has been used to measure primary performance."

How is that legit at all? I could probably pick another state and show no correlation, or even show that primary winners are general election losers. Especially if i can change which states are used half way through...

Its a swing state, chock full of independents, so you get the pulse of the average independent, assuming they're politically relevant this time around. Its mostly a white state with whatever implications you have for race in the election (hillary won in '08 for racial reasons but 99.9% of black people everywhere else voted for Obama with a predictable result ... and likewise this time around Hillary isn't going to get the black turnout that Obama got from that racial block of 99.9% D voters both in the primaries and in November). Its very small and proportionately politically active so the residents tend to get involved more than a large empty state so at least in theory it predicts how politically active people across the country will tend to vote. Wikipedia claims a win results in a historical average boost of 27% of total primary votes. Every other election comes after the news media pounds who is a winner and who is a loser into the brains of votes which surely has some predictive capability looking at an un-influenced voter. Barring special racial situations or whatever (see '08) it historically has strong predictive capabilities.

Note that you can "win" the primaries like Bernie did and still have your nomination stolen from you. I'm sure the Bern victims will be just as motivated to vote for Hillary as they were to vote for Bernie, I mean the Bern victims sure are a big bunch of globalist corrupt banker supporters, LOL, so they'll fit in perfectly with Hillaries supporters. Data from NH does predict excitement and likelihood of voter turnout and so forth which has certain implications in November. All 30% and then some who voted Trump are almost certain to vote in November, the 30% Bern victims well maybe not so much.

Good predictive capability often has nothing to do with how much the results or methods are liked.

I'm guessing it has something to do with it being the second primary (and thus heavily watched as well as very influential)

It's the first primary. It's in our state constitution.

Whoops, misread Google. :) I bet that's why then; it's "untarnished" by previous primaries.

Well it's still influenced by the Iowa caucus, which they ignore.

So he took a whole bunch of data, and found coefficients which correctly determine 5 data points?

Um, okay.

Andrew Tanenbaum (creator of MINIX) and historian Christopher Bates comment on this:

http://www.electoral-vote.com/evp2016/Pres/Maps/Oct23.html#i...

Would love to see Nate Silver discuss this.

Not specifically about this prediction, but he touched on the uncertainty factor in the latest Model Talk podcast episode: http://fivethirtyeight.com/features/how-much-uncertainty-is-...

Reading those predictions with a history of getting it right, always reminds me of the Simpsons episode with Professor Pigskin: http://simpsons.wikia.com/wiki/Professor_Pigskin

Honestly, I want to see calamity. I want a tied electoral college. I want the election to be thrown into utter chaos. Why? Because it will actually throw a monkey wrench into what has been the worst election, with the worst candidates, in the history of this nation.

A tied electoral college wouldn't "throw the election into utter chaos" at all. It just means that the President would be picked by the House of Representatives.

This has actually happened twice before.

How would you see it happening in modern times? Peace and calm reflection?

Which would - barring a spectacular twist - mean a Trump presidency and disaster for millions.

The decision would be made by the new House, whose composition is anyone's guess.

Even assuming the balance doesn't change, the Republicans have a majority of only about 30 seats. Are you willing to bet that 30 representatives can't be flipped, given the...lukewarm support Trump has received from the Republican establishment? I wouldn't be. I mean, not even the Speaker of the House endorses Trump.

In any case, the point is that an Electoral College tie wouldn't mean "chaos". There's a clear procedure for handling it. There are even further backup procedures in case the House is unable to reach a decision.

Ryan hasn't as yet specifically unendorsed Trump.

And 20-30 GOPs flipping to vote against the Republican nominee for President would count as a "spectacular twist', I think.

> There's a clear procedure for handling it.

Indeed there is!

Reply on news.ycombinator.com