May 8, 2015

"Nate Silver fared terribly in Thursday's UK election... The fault, Silver claimed, was with the polling."

Don't blame Silver, says Silver. He only processes the data he gets from polls done by other people.
"The World May Have A Polling Problem," Silver asserted. "In fact, it’s become harder to find an election in which the polls did all that well."... "[T]here are lots of reasons to worry about the state of the polling industry," Silver concluded, citing a range of factors. "There may be more difficult times ahead for the polling industry."
Well, that's awfully bland... from Dylan Byers at Politico, who was only processing the raw material Silver gave him. Can I blame Silver? For anything, ever? He cited "a range of factors." Were they too dull and meaningless to be worth more than the repetition of the conclusion that polls just aren't that good?

ADDED: The NYT surveys some analysis of what went wrong with the polls:
“It could be simply that people lied to the pollsters, that they were shy or that they genuinely had a change of heart on polling day,” [said Alberto Nardelli, writing in The Guardian], “Or there could be more complicated underlying challenges within the polling industry, due, for example, to the fact that a diminishing number of people use landlines or that Internet polls are ultimately based on a self-selected sample.”...

“What seems to have gone wrong is that people have said one thing and they did something else in the ballot box,” [said Peter Kellner, the president of YouGov, a leading survey firm]. “We are not as far out as we were in 1992, not that that is a great commendation.”...
Rem Korteweg, a senior research fellow at the Center for European Reform in London, said... “People say who they are voting for with their heart and then vote with their wallets,”...
To tweak Korteweg's point: People say what they think will make other people like them, but they do what they think is in their interest. Re-tweak: People do what is in their interest, which is to say what they believe is socially desirable, and that won't square up with what they do when no one's looking. If this is the problem, it's a problem that will get worse as it becomes more widely believed that liberalism makes you look good. Korteweg is contributing to the contagion of this belief by saying that in their hearts people are liberal, nudging us all to say I'm a liberal, so I'll seem to be a person with a heart.



(Detail about that video here.)

52 comments:

Glenn Howes said...

It seems an impossible task to get right. Look at the results, one party gets 3.5 million votes and ends up with 1 seat, another gets 1.5 million votes and gets 57. Silver's mistake was claiming he could put error bars on such a mess.

AReasonableMan said...

Silver is only a statistician so he is obviously not responsible for the polls. He is responsible for incorrectly assessing the margin of error in his predictions and for this he can be faulted, as noted in the previous post.

Ann Althouse said...

I think that Silver has argued in the past that his method works despite the problems with the polling, because he uses so many polls (and he gives them varying weights). Isn't that why we're supposed to care about Silver? He can't turn around and blame the polls.

Phil 3:14 said...

Pay no attention to the man behind the curtain.

Bob Ellison said...

Gee. That's a stunning revelation. Who knew that people don't answer polls, or that conservatives tend to hang up, or that rightists hate pollster effrontery? Who knew I don't have a ringer on my land line (which I only keep for the alarm system)? Who knew polling is a spectacularly inexact study riddled with evidence of self-contradictory conclusions?

NotquiteunBuckley said...

"In fact, in 2012, the Daily Kos blogger Markos Moulitsas was more accurate than Silver in predicting the outcome of the 2012 electoral college. Needless to say, Moulitsas was not offered a high-paying job at ESPN.)"

Needless to say? Then why say it?

Besides that why wouldn't ESPN offer the racist Leftist Kos a job? Is there some fool who thinks ESPN isn't a prime promoter of all Leftist indoctrination?

I give ABC some credit for hiring Rush and having Ditka on along with some interesting folks who don't blindly hate anyone slightly unlike them (S. A. Smith vs. Keef Olberdouche) but ESPN is a Leftist organization advancing Leftist causes to the detriment of America.

Silver ought to predict to whom ESPN employees contribute political donations to so as to prove my instincts accurate.

Bob Ellison said...

Also: Nate Silver gets it spectacularly right in the USA 2012 election. He gets it spectacularly wrong in the 2015 UK election.

What conjectures might we entertain?

1) Silver got blind-sided by bad polls in the UK in 2015.

2) There's still some math to be done.

3) Silver's results are random.

CWJ said...

NotquiteunBuckley,

Both ABC and ESPN are owned by Disney. The mouse keeps a very very tight rein in his properties particularly their messaging. Why do you think there is such a difference between the two networks?

Bob Ellison said...

I just threw a dart at a dartboard, and it says the Hamburglar is going to be elected POTUS in 2016. If my dart is correct, you're all going to worship me.

MayBee said...

It seems the polls generally under predict conservative wins.

Anonymous said...

Dear Leader's Merdes Touch struck again: Axelrod advised Milliband to oblivion.

PuertoRicoSpaceport.com said...

It's all about the advertising dollars and campaign contributions.

In the same way that nobody would go to see a football game in which the smart money predicted a 30-0 blowout, people won't get excited about an election if the expect results like yesterday or like 2016.

Gotta keep it close so the marks keep giving money to pols. The pols keep giving money to advertising agencies, PR firms, consultants, newspapers, TV and so on.

When people talk about $1bn being spent in the 2016 presidential election cycle, think of where that $1bn is being spent.

If Hillary is 5 points behind Paul (or 5 ahead of Walker or whatever) both sides will bang that drum mercilessly. "Ya gotta give us money. Just a bit more will put us ahead (or keep the other from getting ahead)"

Or, "It looks like we might win. We won't forget who our friends were."

If Paul is 30 points ahead of Hilary, if Hilary is likely to do a Mondale and pull only 1 state, who is going to donate? Everyone will see the money as wasted either way.

Give it to Hilary? Why? Maybe give some to Paul so we have some access but we won't need to give as much.

Everyone has to keep the polls tight all the way through. Their rice bowls depend on it.

John Henry

Anonymous said...

Next: vindictive Dear Leader will side with Argentina and Putin in the brewing Falkland War.

PuertoRicoSpaceport.com said...

re Nate Silver failing horribly: So what?

He'll have some plausible excuses and nobody will remember his failures.

In 20 years everyone will be talking about how Awesome Silver is for getting one poll right in 2008.

Nobody will remember how many he failed.

John Henry

Carter Wood said...

Pollsters got things right in Alberta, where the NDP just swept out the Progressive Conservatives. From ThreeHundredEight.com (Canadian stats/news site), "Redemption for the pollsters, revolution for Albertans." The NDP, for goodness sake. Not quite the green-socialist NDP you find in BC, but still.

clint said...

"Glenn Howes said...
It seems an impossible task to get right. Look at the results, one party gets 3.5 million votes and ends up with 1 seat, another gets 1.5 million votes and gets 57."

Nonsense.

The British election is 650 individual elections for Parliament. This was known in advance by all polling companies and Nate Silver.

It's just like trying to predict a U.S. Congressional election -- throwing up your hands because a national poll isn't very predictive is silly. That's why we do polls of likely voters in competitive districts.

"Silver's mistake was claiming he could put error bars on such a mess."

Exactly (except for the "mess" bit). His problem isn't that he picked the wrong answer, it's the level of certainty with which he picked a completely wrong answer.

PuertoRicoSpaceport.com said...

Mike Royko was saying this as far back as the 60's but it always bears repeating:

Never, ever, tell a pollster the truth. We have an obligation to lie to them.

What possible benefit do we, the individual voter, get out of polls? Politicians, of all ideologies use polls to try to discern what we want to hear. They then tell us that, even though they don't believe it and have no intention of doing it when elected.

If we can get a general movement going we can make polls even more untrustworthy than they already are. Then Heavens Forbid!!! Politicians might have to tell us what they actually think. We can't have that. There's no money in it.

John Henry

themightypuck said...

Silver didn't have a model for the UK General Election. He was parroting http://www.electionforecast.co.uk/2015/index.html. He is like a weatherman on the local news getting all his information from the NWS.

Original Mike said...

I've got an idea. Why don't we set one day when all the voters gather together (at, say, a polling place) and register their preference. Would remove a lot of the guesswork.

tim maguire said...

Silver would be a good politician. When he's right, he's a genius, when he's wrong, well, it's all probabilities and bad polling.

Bill said...

I'm inclined to lie to a pollsters as a matter of principle.

tim maguire said...

I don't think he's entirely wrong to say that in their hearts, people are liberal. But following your heart usually leads to disaster. It has to be tempered by your brain, which, if it's more or less fully functional, is conservative.

Steve said...

In the UK you can be criminally charged as a racist for saying something not Politically Correct. So many people parrot the correct lines while adhering to the non-liberal opinion.

Sebastian said...

"it's a problem that will get worse as it becomes more widely believed that liberalism makes you look good. Korteweg is contributing to the contagion of this belief by saying that in their hearts people are liberal, nudging us all to say I'm a liberal, so I'll seem to be a person with a heart."

Which gives callous conservatives an excellent opportunity to f*** with their minds a bit.

Still, the more-widely-believed hypothesis needs testing. Better data on trends in poll unreliability and alternative explanations would be good.

Of course, liberals also vote wallets rather than hearts -- other people's wallets.

Eric said...

So Nate says it's "garbage in, garbage out." Where exactly is his value added?

tim in vermont said...

Nobody at the BBC voted for the Conservatives, so WTF?

Michael K said...

"“It could be simply that people lied to the pollsters,"

Of course they did. Our problem is that the LIV here actually vote for these assholes.

Anonymous said...

Back when Silver was being hailed as the Warren Buffett of political forecasting, I privately suspected that he might turn out to be the Long-Term Capital Management of political forecasting. So it has turned out to be.

YoungHegelian said...

"The fault, dear Nate, lies not in the polls but in ourselves."

Birches said...

People aren't telling pollsters how they are really going to vote because the Media Elite are constantly telling everyone that Conservatives (or UKIP or the Republicans) are evil.

Birches said...

I live in one of the most conservative counties in the country and I still won't discuss politics with my neighbors. This is what the Coastal Media Elite has wrought.

n.n said...

In their hearts, people are conservative. In the open, people are liberal when they hope to avoid confrontation, capable of exerting leverage, or are provided opiates to suppress their dignity or response.

Rick said...

I'm reminded of the 2012 election aftermath with many left leaning critics insulting the intelligence of everyone who didn't take the polls as the final word. But I don't want to focus on the fact polls are more subject to interpretation than they held at the time. Rather, focus on how the left immediately identified the many confounding factors and compare to their 2012 reaction.

"Not believing the polls" was their criticism at the time, claiming those who didn't were science or reality deniers, or deluded wishful thinkers. But in fact the interesting thing about polls is how they change rather than their snapshot position. The analogy to a race would be one runner leading but the other catching up. The question is not whether the person is leading (which left criticism of the right in 2012 was based on) but rather how fast the other is catching up relative to the finish line. Today's commentary shows most of the left understands this.

Which means in 2012 these leftist critics of the right were pretending to be stupider than they actually are because it allowed them to paint the conservatives as stupider than they knew them to be. And this includes Silver.

Human psyche is amazing.

Alex said...

So considering the success of conservatives over in jolly old England, can we defeat the Hildabeast next year?

Conservatives, YES WE CAN!

Brent said...

Hugh Hewitt one minute ago:

"Possibility that pollsters aren't way off but that voters obtain/process the information they provide and react much more quickly".

Skipper said...

All these rationalizations would be asserted quite differently had the outcomes been reversed, that is if liberals won after the polls leaned conservative.

Foobarista said...

The real problem is that non-lefties know that pollsters are leftist media types who will disapprove of an "incorrect" answer (and in the UK, Tory="incorrect").

So, they'll fib (who cares?) or not answer the pollster's questions.

Left Bank of the Charles said...

Polling works best in a 2-way race. With more candidates, voters have to consider whether it is best to vote for their first choice or second choice, if the third choice is unpalatable.

The theorists call this the voting paradox, but in practice the public uses polls to help them make that decision.

In other words, I might prefer a Labour government but abhor a Labour/SNP coalition. So Conservative might be my second choice, but how I vote on election day if I don't think the Labour win will be big enough. Every day before that I might still be hoping for my first choice.

Now, you could poll for my preferences, but as long as I have hope I have every incentive to lie. I want enough other people to vote Labour that I don't have to vote Conservative. I want to encourage them to vote Labour. I don't want to discourage potential Labour voters by admitting I plan to vote Conservative once I run out of hope.

Sammy Finkelman said...

I would suspect a lot of this is simply respoonse bias: Not people lying to pollsters, but people not answering polls.

In the United states this skews drastically conservative, and it is an at to correct for it.

After 1936, through about 1960, polls used to have a conservative bias, with liberals not trusting them. Then it got more even till about 1988. Now conservatives refuse, and only a minority of people answer polls.

Sammy Finkelman said...

In Great Britain, the exit polls, although more accurate than the pre-election polls, still undeestimated the Conservative vote.

They won an absolute majority. 331of the 650 seats. Both Britain’s Labour Leader Ed Miliband and Liberal Democrat leader Nick Clegg have resigned as party leader. Clegg did keep his own seat.

http://www.nytimes.com/2015/05/09/world/europe/david-cameron-and-conservatives-emerge-victorious-in-british-election.html

The Scottish National Party won 56of 59 seats in Scotland. They were only running in Scotland. Labor has only one remaining seat in Scotland.

Nigel Farage, leader of the U.K. Independence Party did not win a seat, and his party won only 1 seat.

Anonymous said...

If you aren't a liberal when you're twenty, you have no heart; but if you're still a liberal when you're forty, you have no brain. Of course, I wasn't a liberal when I was twenty. . . .

tim in vermont said...

"Possibility that pollsters aren't way off but that voters obtain/process the information they provide and react much more quickly"

So Hugh Hewitt is using the Feller Faster principle now? Mickey Kaus would be so proud.

Sammy Finkelman said...

Article about betting in the Uited Kingdom about he election (from last month)

http://uk.reuters.com/article/2015/04/08/uk-britain-election-gambling-idUKKBN0MZ0ZM20150408

tim in vermont said...

Advice on how to leave the United Kingdom

Ha ha ha ha ha!

Just don't come here! I am sure you can all move the Venezuela, where your approved form of government is now in power. But bring our own toilet paper,... and food ... and blankets... and clothes...

It's a SOCIALIST PARADISE!

retired said...

The polls have been useless for decades.

Gabriel said...

Remember how the Right was taunted in 2012 about the skewed polls, and how Nate Silver was the gold standard?

Now he's complaining about skewed polls.

Anonymous said...

Nothing could be better for a Democracy than to lose all trust in polling.

For some odd reason, it's human nature to want to vote for a winner.

Wouldn't it be nice if we could go into an election having no idea where our candidate is at with the polls?

But polls cause people to drop out of the race, or say stupid things they wouldn't normally say, or vote for someone else because they are trying to make sure the person they don't want is beaten.

Get rid of polls, at least for the Primaries.

Big Mike said...

[Phone ringing]

Citizen: Hello?

Pollster: I'm calling from the XYZ polling firm and we were wondering if we could ask you a few questions about the upcoming election?

Citizen: Yes, please proceed.

Pollster: Thank you. Do you plan to vote in the coming election?

Citizen: Yes

Pollster: May I ask whether you plan to vote for those filthy, scum-eating Tories or are you voting for those fine, upstanding people in Labour?

Citizen: I am voting Conservative

Pollster: Really, do we need to send some yobs around to help you make a better choice of party?

Citizen: Er no, just joking, Labour it is.

Pollster: Thank you for your time.

[Hangs up]
_____________

Note: "Yobs" are young thugs prone to reshaping skulls with wooden staves or cricket bats, plus a tincture of violence.

richardsson said...

Yes, Silver had Cameron losing seats and instead he gained seats. Ha Ha Ha. I used to advocate lying to pollsters in my Political Sciences classes. After all, it's none of their business who you vote for even if they think otherwise. In the early days, the students were skeptical. They would laugh nervously. But with caller ID, the rise of cell phones, and the decline of land lines, and the Patriot Act, the students started taking me seriously. For the pollsters, I know how they validated the randomness of their samples in the old days but those methods are no longer possible.

Rusty said...

I couldn't watch the video. Some asshat was screaming about something. Nice architecture though.

averagejoe said...

Watching that video I am reminded why I could never be a police officer, and how amazing it is that so few citizens are shot, beat and curb-stomped by cops every year. Really, there should be a lot more cops smashing progressive democrats over the head with billy clubs. The world would be a kinder, gentler, and more quiet place. I'm also old enough to remember when American civil unrest featured construction workers getting together and kicking the shit out of hippie protesters. Where have you gone Joe DiMaggio?

Peaceful warrior said...

At least in 2012, Silver used more accurate state polls in lieu of gravely flawed national polls. I sense that wouldn't have been possible here.