Hacker News Reader

Comment by blhack | Hacker News Reader

Trump isn't nearly as far behind as you may think: http://graphics.latimes.com/usc-presidential-poll-dashboard/

Yes, this poll is an outlier because their methodology is very different than most other polls.

Yes, Nate Silver has a very different outlook on the election, but Nate Silver was also not very reliable during the primary season (although that is, by his admission, because primaries don't have very good polling).

If I had to make a[n extremely uneducated guess], it's because most polling is done over the phone, and I don't think "people who answer their phone when they don't recognize the number" is a very good random sampling of voters in the US.

I don't think this race is nearly as over as most people seem to believe, which is why, if you want to beat Trump, it is REALLY IMPORTANT (all caps!) that you get out and vote!

View on HN

You've cherry-picked the one outlier.

It's, of course, very possible that "this time it's different" as they say every time. He is certainly an unconventional candidate. However, I'll throw my lot in every time with the MANY polls who are trying to account for this, and the MANY poll aggregators who are trying to make sense out of this madness.

I think 538 was actually quite reliable during the primary season. Pretty much bang on, in fact.

Everything they said anti-trump was basically opinion and they have been roundly chastised (as deserved) for treating punditry as fact.

There was a good documentary on polling on BBC Radio 4 in light of their very weak performance in IndyRef and the UK General Election. One pollster said that he actually has a high regard for outliers which statistical aggregations do not capture. In cases where the majority of pollsters got it horribly wrong, there have been outliers at some significant distance from the herd that have been on the money but ridiculed at the time.

The reasoning is that pollsters have herd behaviours - they believe publishing an outlier they will lose them credibility so they suppress inconvenient results and will rework methodologies and weightings to avoid it. For many, election polling is a loss leader - its publicity to get their name out there for commercial services. The credibility of the outlier is increased because it has escaped the herd pressure despite the incentives.

I think problem is deeper. If one predicts heads as result of coin toss and other predicts tails, than one of them will be right. It says nothing about their abilities to predict. We always can find some outlier who predicts results different from majority of pollsters. The keyword is 'always': if such a possibility always exists (probability of existence P=1), then we can't extract any information from it (-log_2 P == 0).

To extract information we need to carefully revise methodology behind outlier predictions, and such an activity is a complex task for skilled data analyst.

> I don't think this race is nearly as over as most people seem to believe

This feels a lot like the atmosphere over here (in Britain) just before the Brexit vote.

Many people thought it was a guaranteed Remain vote, and woke up to a terrible fright on that Friday morning.

The polls showed Brexit winning.

They didn't:

https://en.wikipedia.org/wiki/Opinion_polling_for_the_United...

Nate silver personally hasn't had a good record of making predictions in the primary this year, but his model had trump winning through most of the primary season. He just didn't believe his own model.

4% in national polls (that's the number averaging across the polls) is a huge difference given the strong alignment of so much of the electorate. And it isn't even the interesting way to analyze the election, winning the national popular vote isn't what wins the race.

If you look at states that are contested, things look bad for Trump, even in the ones where demographics favor him he is not getting polling results that point to a win (an example would be Pennsylvania).

Regarding that poll specifically: http://mobile.nytimes.com/2016/10/13/upshot/how-one-19-year-...

Their response: http://www.latimes.com/politics/la-na-pol-daybreak-poll-ques...

Where they eliminated that one 19 year old, and saw very little effect on the poll.

> Where they eliminated that one 19 year old, and saw very little effect on the poll.

How is that a compelling answer?

It's like if a news investigation found that a company had poor quality control and features a photo of a defective product in their writeup and the company's response was "we issued a refund for the product the news team purchased."

All I can say is that you sound an awful lot like the 2012 talking heads who were convinced Romney would win despite all the polling pointing to the contrary. For the record, Nate Silver's biggest mistake in the primaries was not listening to the polls enough. For him (and every other data-driven forecaster) to be getting the general wrong they'd have to be making the exact opposite mistake as before.

I'd be happy to give you 10:1 odds on Trump losing. And I'm biased here: a few months ago, I was convinced Trump would win.

I haven't seen the crosstabs on the USC polls, but my gut feeling is that you're right; the kind of very fine demographic reweighting they're doing isn't common, because (A) it's a pain in the ass and of dubious real-world value; and (B) it's vulnerable to magnifying sample-size and selection problems, of which the skewed Trump support amongst young African-Americans is one example. So while I'll be interested to do a retrospective on the poll's accuracy and precision after the election, I'm personally not putting much faith in it for now.

That's not true, they saw the same affect as upshot, which is to say nearly 1 percentage point movement by removing him.

1 responder should not move a 2500 person poll by 1%

> Of the poll’s 14 weeks so far, there were three times when removing that weight switched the result from a Trump lead to a Clinton lead. Generally, the shift was between one and two percentage points...Some people have ignored the margin of error in order to make a political point, but there’s not much we can do about that.

That seems highly disingenuous a response.

You've just selected the national poll that is the most favorable to Trump. An average over polls >> any individual poll. Note that I am not saying that poll should be thrown out. But picking one poll in isolation is almost useless.

Looking at this specific poll in isolation is fine as long as you understand that you are looking at it in isolation.

I say so because of its polling methodology, which is unique.

If you're a politics/polling nerd, this poll is interesting. If you're not, then yes, you are probably better off either just reading 538, or looking at the RCP polling average.

https://news.usc.edu/109339/why-the-usc-dornsifel-a-times-pr...

Most politics/polling nerds look at aggregations of polls. Not sure where you get this idea that they are interested in a poll like USC which is such an outlier.

The fact that it's an outlier doesn't interest you? Maybe we have different definitions of "nerd".

The fact that it's an outlier means, statistically, that it is more likely to be wrong.

Yes, it's more likely to be wrong.

But this poll performed well in 2012[1] with the same methodology. As a polling nerd, wouldn't you be interested in why it is producing unusual results this year and what implications that might have for polling in general?

[1] http://www.latimes.com/politics/la-na-pol-poll-faq-20161006-...

It is the height of stupidity to believe in any one poll. Especially that one which has had major issues with sampling:

http://www.nytimes.com/2016/10/13/upshot/how-one-19-year-old...

There are plenty of aggregation sites which are far better at smoothing out any issues:

http://www.nytimes.com/interactive/2016/upshot/presidential-...

They all show Trump losing in a landslide with Clinton getting around 330 EC votes.

USC's response to this: http://www.latimes.com/politics/la-na-pol-daybreak-poll-ques...

This is exactly why you should never blindly trust a single poll.

We don't know who is right/wrong until after the election so it is better to assume that a range of different polls taken at different times with different methodologies is likely to be more accurate.

I heard about this poll, but I haven't paid attention to it. You got me reading up on it.

Now I can't sleep...

If you look at the pictures of the number of people in attendance at their respective events one would conclude Trump is ahead.

Do we have any evidence that number of people at campaigning events correlates to more votes? If not, why would we conclude that?

Is it not (in absence of any evidence) equally plausible that the type of voter that will vote for Trump is more predisposed to attend a campaign event than other voters?

Which is why thats an awful gauge of the electorate. The vast majority of voters do not attend political events, and "attendees at events" is really just an uncorrelated measure of "enthusiasm for a candidate amongst their base". If 40% of the country loved him so much that they all showed up at every single rally, yet the other 60% hate him, and will vote for HRC despite not really loving her, then you see a rally filled to the brim with Trump supporters, and nobody at hers, yet his chances are zip. Rally attendance is a meaningless metric.

It might not be great, but I trust it more than these dishonest polls

Bernie is a good counter-example. Huge rallies, way behind in votes.

have you not read wikileaks? the DNC/Hillary rigged it against him from day one, he never stood a chance

Clearly you haven't read those emails: they show nothing of the sort, which is why you couldn't provide any evidence to back that claim.

You can find signs that people in the establishment personally favored the establishment candidate – the least surprising revelation in political history – but there's no evidence that lead to any concrete action. That's why the only claims of rigging have been intentional misrepresentation based on the knowledge that some people like you would repeat those claims without checking the sources.

When you have Donna Brazile feeding the Clinton campaign townhall questions so that she can prepare the answers. Threats for super delegates switching to Bernie and Debbie Wasserman Shultz stepping down. It's clear evidence of a rigged election. Who knows what was happening that is not in the emails.

Try citing specific ptimsry sources for events which actually happened. There's plenty of hyperventilating and outright propaganda on right-wing blogs but there's a reason why nobody with credibility is claiming the primary was rigged.

Hint: it's the same reason why the alleged victim is going around telling his supporters to vote for Hillary. If you trusted his judgement enough to think he should be president, why not trust his analysis now?

There's a reason why nobody with credibility is claiming the primary was rigged

And that reason is elementary game theory. The only people with incentives to claim that the Democratic primaries were rigged are those who will be left with no political influence if the Democrats fare poorly.

When the Republicans do well, it's because they value party unity over literally everything else. One of HRC's strengths is that she brings the same thinking to the Democrats. For anyone within the party, working against her carries no conceivable upside. And as we've seen in DWS's case, working for HRC means you'll be well taken care of, no matter what.

Basically, if Sanders or anyone else on the left has beef with the DNC, they will be much better off if they wait to bring it up until after the election. And maybe not even then.

Except Bernie's rallies were full of young people, who are always up for a rally but who can seldom be bothered to vote. Meanwhile, Trump's rallies are full of old people who have nothing else to do but vote.

I'm sorry, but offhandedly dismissing "these dishonest polls", when, frankly,

1. They're the best data we have

2. We have no reason to believe they're incorrect

3. There is REALLY no reason to believe they're dishonest

Is 100% intellectually dishonest.

This is exactly what happened 4 years ago, and apparently some people have not learned from their tremendous mistake.

If you want to know why "turnout at a campaign event" is a crappier metric than "real actual scientific data", I don't know how else to help, other than to point to other instances in which it has failed:

"Mitt drawing larger crowds" http://www.politico.com/story/2012/10/mitt-drawing-larger-cr...

"Donald Trump continues to draw YUGE crowds. That matters less than he thinks." https://www.washingtonpost.com/news/the-fix/wp/2016/09/28/do...

"Trump brags about crowd size but will it turn into votes?" http://bigstory.ap.org/article/d74fbd472f7a420a90737e55cfb20...

I do not know or care who you support in the Presidential election. However, I suggest you get realistic about your metrics, as you're way off in fantasyland at the moment, as far as I can tell.

If you turn out to be incorrect, I hope you learn from this experience. I know I will, if the data turns out to be wrong.

I'm in fantasy land? If you think the polls aren't manipulated for political advantage you are in fantasy land. You're appeal to 'real scientific data' is childish, since the data can be easy manipulated.

And yet, Trump supporters never had any trouble believing the polls when they showed their candidate leading during the primaries. Every single time, Trump would open his rallies with "have you seen the latest polls?".

When the polls show him to be ahead, it's evidence that the voters love him. When he's behind, it's evidence that the polling is rigged.

Hillary Lead Over Trump Surges After Reuters "Tweaks" Poll http://www.zerohedge.com/news/2016-07-30/clinton-lead-over-t...

All of my comments stand.

Campaigns also conduct their own internal polling which may or may not have biases, but they're not released; they're used for the campaign. So it only benefits them to know what's actually going on. Trump's is one of the few campaigns that seems only interested in good news, and in not learning from what their internal polling tells them.

That said, of course all campaigns will selectively brag about the polls that make them look good, and quietly ignore those that don't.

Again, the polls are extremely reliable in this country. If you can't see that, you're unlikely to be swayed by yet another data point when this election is complete.

By that logic, Bernie Sanders would be the Democratic nominee.

Reply on news.ycombinator.com