PDA

View Full Version : Polls - Some thoughts



Keith Wilson
12-02-2015, 08:23 AM
There's a good article in The New Yorker about the history of polling, the changes recently a nd its effects on democracy (link here (http://www.newyorker.com/magazine/2015/11/16/politics-and-the-new-machine)). It's well worth reading, although it's fairly long. Interesting discussion of feedback effects. Did you know that response rates for most polls are now in the single digits?

The beginning:


Politics and the New Machine
What the turn from polls to data science means for democracy.
By Jill Lepore

“I am who I am,” Donald J. Trump said in August, on the eve of this season’s first G.O.P. Presidential debate, and what he meant by that was this: “I don’t have a pollster.” The word “pollster,” when it was coined, was meant as a slur, like “huckster.” That’s the way Trump uses it. Other candidates have pollsters: “They pay these guys two hundred thousand dollars a month to tell them, ‘Don’t say this, don’t say that.’ ” Trump has none: “No one tells me what to say.”

Every election is a morality play. The Candidate tries to speak to the People but is thwarted by Negative Campaigning, vilified by a Biased Media, and haunted by a War Record. I am who I am, the Candidate says, and my Opponents are flunkies. Trump makes this claim with unrivalled swagger, but citing his campaign’s lack of a pollster as proof of his character, while fascinating, is utterly disingenuous. The Path to Office is long. To reach the Land of Caucuses and Primaries, the Candidate must first cross the Sea of Polls. Trump is a creature of that sea.

Lately, the Sea of Polls is deeper than ever before, and darker. From the late nineteen-nineties to 2012, twelve hundred polling organizations conducted nearly thirty-seven thousand polls by making more than three billion phone calls. Most Americans refused to speak to them. This skewed results. Mitt Romney’s pollsters believed, even on the morning of the election, that Romney would win. A 2013 study—a poll—found that three out of four Americans suspect polls of bias. Presumably, there was far greater distrust among the people who refused to take the survey.

The modern public-opinion poll has been around since the Great Depression, when the response rate—the number of people who take a survey as a percentage of those who were asked—was more than ninety. The participation rate—the number of people who take a survey as a percentage of the population—is far lower. Election pollsters sample only a minuscule portion of the electorate, not uncommonly something on the order of a couple of thousand people out of the more than two hundred million Americans who are eligible to vote. The promise of this work is that the sample is exquisitely representative. But the lower the response rate the harder and more expensive it becomes to realize that promise, which requires both calling many more people and trying to correct for “non-response bias” by giving greater weight to the answers of people from demographic groups that are less likely to respond. Pollster.com’s Mark Blumenthal has recalled how, in the nineteen-eighties, when the response rate at the firm where he was working had fallen to about sixty per cent, people in his office said, “What will happen when it’s only twenty? We won’t be able to be in business!” A typical response rate is now in the single digits.

. . . .

Lew Barrett
12-02-2015, 08:30 AM
I don't answer them unless they're posted in the Bilge.

John Smith
12-02-2015, 08:48 AM
I don't know, but I suspect many, as myself, get tired of polls. I'm even more irritated by them this year as they've been used to determine who gets in the debates. They also drive too much the candidates positions. I'd like to see less polls. Perhaps none before the people actually begin voting.

Norman Bernstein
12-02-2015, 09:32 AM
I think some polls still have value, but I get especially annoyed at polls which are blatantly ambiguous... the most common of which is the question "Is the country headed on the right track, or the wrong track?"

First, it is inevitable, regardless of WHAT the actual conditions are (economic, political, etc), that many people, if not most, will be unsatisfied with something....

...but more importantly, if someone answers 'wrong track', is it because they are conservative and think the country is getting too liberal.... or they're liberal, and the country is getting too conservative? That's an example of an arbitrary and ambiguous poll whose results mean nothing.

Another problem with polls: the tendency to magnify small effects. For example, Donald Trump is polling at around the 25% level.... 25% of what? It's of conservative voters, who constitute approximately half of the electorate.... which really puts his national popularity at around 12% or so. People often fail to understand the context of a poll, and are deceived by the numbers.

Norman Bernstein
12-02-2015, 09:41 AM
Speaking of interesting polls that probably don't mean anything:


Sen. Bernie Sanders (I-Vt.), a contender for the Democratic presidential nomination, is gaining steam against top Republican rivals, according to a national Quinnipiac University poll (http://www.quinnipiac.edu/news-and-events/quinnipiac-university-poll/national/release-detail?ReleaseID=2307) released Wednesday.

In a hypothetical matchup against the current GOP front-runner, business mogul Donald Trump, Sanders takes 49 percent of the vote to Trump's 41 percent. Against Sen. Marco Rubio (R-Fla.), Sanders leads 44 percent to 43 percent. He also beats Sen. Ted Cruz (R-Texas) by 10 percentage points and retired neurosurgeon Ben Carson by 6 points.

TomF
12-02-2015, 10:53 AM
And of course, most polls still rely on land-line telephone contact. The data's skewed against certain demographics, in which many don't have and never have had a land-line. Some solutions have been mooted, but they're far from perfect.

Norman Bernstein
12-02-2015, 10:56 AM
And of course, most polls still rely on land-line telephone contact.

Actually, I believe that is no longer true.... pollsters and demographers have figured this out quite a while ago. Admittedly, they don't have perfect solutions; if they bias the polls towards cellphone users, they miss the older demographic who hasn't adopted that technology.

My own opinion is that 'early' polls, like we're seeing now, are pretty meaningless, since they represent fickle tendencies of people willing to answer the poll.... but the ones just prior to an event like a party convention, or just before an election, are probably a great deal better... albeit still flawed.

TomF
12-02-2015, 11:04 AM
Subject to any number of different kinds of skewing. You can get decent samples of cell users, landline users, Facebook or other users, but are they representative? How much overlap, confirmation bias, etc.? The serious folks are using big data to extrapolate and target, because data collected about your personality from mining a person's activities which actually express it (like shopping, social media commenting etc.) tends to be pretty reliable.

But you still need to find a representative sample somehow of those niches to project the larger group's support of this or that candidate.

SullivanB
12-02-2015, 11:07 AM
I saw the end of a piece on MSNBC in which they discussed that particular Quinnipiac poll and I've been thinking about some of the conflicting results they mentioned. For example, Clinton's still thought to be untrustworthy by 60% of the respondents, yet she's gained ground in some important categories. Especially difficult to understand is the roughly 10% gain between she and Sanders since the last poll among those thinking of themselves as "very liberal". It just doesn't make sense. Why would the most liberal of us be dropping Bernie for a corporate shill like Clinton?

Then it dawned on me. Of course! Liberals are always reluctant to back a socialist with a hernia. Happily, Bernie has just recently had his hernia repaired and I'm confident we'll see a post-hernia-repair bump for Bernie among the most liberal of us in the next Quinnipiac poll.

Norman Bernstein
12-02-2015, 11:10 AM
Subject to any number of different kinds of skewing. You can get decent samples of cell users, landline users, Facebook or other users, but are they representative? How much overlap, confirmation bias, etc.? The serious folks are using big data to extrapolate and target, because data collected about your personality from mining a person's activities which actually express it (like shopping, social media commenting etc.) tends to be pretty reliable.

But you still need to find a representative sample somehow of those niches to project the larger group's support of this or that candidate.

Yeah, you point out the key problem: insuring that the sample is reasonably representative. I'm sure that the polling firms stay awake at night worrying about that... and exert a lot of energy and time during the day, testing to see whether they do indeed get representative samples.

My guess is that the problems caused by 1) the transition to cell phones, and 2) the increasing unwillingness of the public to respond to polls and surveys, is at least somewhat compensated for by the big data analysis.

My biggest thing about polls is the disconnect between how people respond, versus how they act. Right now, something like 27% of the GOP electorate is perfectly willing to ignore the outrageous lies and insufferable ego of Donald Trump... but I simply refuse to believe that this represents votes. I think it's just an angry backlash which, a year away from the election, means very little.

Norman Bernstein
12-02-2015, 11:13 AM
I saw the end of a piece on MSNBC in which they discussed that particular Quinnipiac poll and I've been thinking about some of the conflicting results they mentioned. For example, Clinton's still thought to be untrustworthy by 60% of the respondents, yet she's gained ground in some important categories.

It's no different than Donald Trump getting 27% of the support in GOP polls... the mistake would be in presuming that 27% of the GOP electorate are going to vote for the guy. Hillary's 60% 'untrustworthy' response does NOT mean that 60% will not vote for her.

If you're interested in a more surprising poll, take a look at the disconnect in the percentage of blacks in the south responding to Hillary, versus Bernie.... the gap is positively enormous. Does Bernie really have a problem connecting to black voters? Or is the poll meaningless?

My guess: it's the latter.

RonW
12-02-2015, 11:49 AM
Norman Bernstein says -
Hillary's 60% 'untrustworthy' response does NOT mean that 60% will not vote for her.

Oh really, then the next question on the poll needs to be, would you vote for an incompetent known liar and cheat for public office. I think we all know what that answer would be. Hillary will be lucky to hit 40% in a general election due to the 60% untrustworthy response. After all what would be more important for the biggest political office of all.

But polls no longer are used to relay the pubic's opinions, but more to influence and push the public to go along with the majority.
They generally only call between 500 and 1,000 of whom they already know their political stance and then phrase the questions to get the desired answer.

Simple on line right wing polls such as drudge and breitbart who have over 1/3 million responses have constantly showed jed bush at the bottom of the heap, but the national liberal media polls up till now have tried to convince everyone that bush was 2nd or 3rd when it was anything but that. Once again a effort by the media to influence.

Norman Bernstein
12-02-2015, 11:53 AM
Simple on line right wing polls such as drudge and breitbart.....

*roflmao*

'Polls'? :):):)

RonW
12-02-2015, 11:56 AM
*roflmao*

'Polls'? :):):)

What is the definition of a poll ? Do inform us...

Norman Bernstein
12-02-2015, 12:00 PM
What is the definition of a poll ? Do inform us...

According to you, it's when some blatantly partisan right wing website conducts a popularity contest, exclusively among readers of the blatantly partisan right wing website. The 1/3 million responses constitute something like 0.6% of the electorate.....

....and YOU actually think they mean something. See my other thread entitled 'Priceless :)' to understand what is going on here. It's called 'stupidity', and there's really no other way to say it.

RonW
12-02-2015, 12:05 PM
According to you, it's when some blatantly partisan right wing website conducts a popularity contest, exclusively among readers of the blatantly partisan right wing website. The 1/3 million responses constitute something like 0.6% of the electorate.....

....and YOU actually think they mean something. See my other thread entitled 'Priceless :)' to understand what is going on here. It's called 'stupidity', and there's really no other way to say it.

Well I hate to break the news to you norm, but if you want to know how the right is going to vote in the republican primary, then you need to poll the right and not the left.

And if you want to know how the left is going to vote in their primary then ask the left..

And that would lead us to think that if we want to know how the general public is going to vote in the general election, then we need to ask the general public with a proper representation of the left, right and center.........

Or do you continue to believe and try to make others believe that only what the left thinks is what matters?

Now I am ROFLMAO.

Keith Wilson
12-02-2015, 12:17 PM
. . . if you want to know how the right is going to vote in the republican primary, then you need to poll the right . . .
. . if you want to know how the left is going to vote in their primary then ask the left . . .
. . . if we want to know how the general public is going to vote in the general election, then we need to ask the general public with a proper representation of the left, right and center . . . .
Very true. And that, Ron, is precisely why online surveys like Breitbart and Drudge don't tell us much. They are accurate representations of the opinions of those who answered the questions, with no way at all to connect those opinions to the population of likely Republican primary voters, or any other population, for that matter.

RonW
12-02-2015, 12:31 PM
Very true. And that, Ron, is precisely why online surveys like Breitbart and Drudge don't tell us much. They are accurate representations of the opinions of those who answered the questions, with no way at all to connect those opinions to the population of likely Republican primary voters, or any other population, for that matter.

Oh really, so these right wing wacko sites can't give us a basic idea as to how the right wing wackos are going to vote in the republican primaries. is that what you are saying.....? Well maybe we should go to huffington post, mother jones, or think progressive to tell us how the right is going to vote in their primary for their candidate.

Get real keith. The righty polls show trump and cruz bumping heads and all the rest have fallen down.....

TomF
12-02-2015, 12:42 PM
Oh really, so these right wing wacko sites can't give us a basic idea as to how the right wing wackos are going to vote in the republican primaries. is that what you are saying.....? Well maybe we should go to huffington post, mother jones, or think progressive to tell us how the right is going to vote in their primary for their candidate.

Get real keith. The righty polls show trump and cruz bumping heads and all the rest have fallen down.....Statisticians call it "selection bias," Ron.

Nobody doubts that the particular niche of conservatives who inhabit Breitbart or Drudge's domains are going to behave like they say they will. The question is whether the Breitbarters' and Drudges' preferences are good predictors of the voting behaviour of conservatives like, say, Cris Ross, or BrianW. And what proportion of conservative voters fall into the Breitbarter/non-Breitbarter camp.

I'm not convinced that you speak for most registered Republicans, for instance. Let alone for most folks who'll end up voting Republican in a year's time.

RonW
12-02-2015, 12:48 PM
TomF --
I'm not convinced that you speak for most registered Republicans, for instance. Let alone for most folks who'll end up voting Republican in a year's time.

I never said or insinuated that I do represent the majority of the right, and neither does the trump supporters.

The right goes from mild to the hard cores like me, while the left goes from mild left of center to the radical far left socialists like you..

But left wing polls don't tell you where the right wing is at..

TomF
12-02-2015, 12:56 PM
"radical far left socialist." Tee hee.

I agree, though, that polls of people who spoon around on Lefty websites won't tell you about the behaviour of Righties. Polls published on either Left or Right websites, OTOH, are a different thing.

No reason that Drudge or Huff couldn't report the results of something using a statistically valid sampling methodology, and show what's actually out there beyond their own little echo groups. If the folks conducting the poll found good ways to avoid the skewing issues we've been trying to discuss.

Norman Bernstein
12-02-2015, 01:01 PM
Statisticians call it "selection bias," Ron.

Forget it, Tom... he either doesn't understand that, or doesn't WANT to understand that. He desperately hopes that the entirety of the right is extremist, like him... and any information to the contrary doesn't make it through his filter.

Gerarddm
12-02-2015, 01:04 PM
I am fond of the old joke that goes ' 9 out of ten doctors surveyed said they didn't want to be surveyed any more '.

Canoeyawl
12-02-2015, 01:09 PM
That was a good article and it convinced me that "polls" are pretty much useless. Unless you want to spread disinformation that might sway public opinion.
They poll a couple of hundred people, get about 10 answers and that is science?
No wonder Trump and Carson are in the lead.

Keith Wilson
12-02-2015, 01:21 PM
No, they poll more than that, and that keep going until they get enough people and what they think is a representative sample. The trouble is that with low response rates, you never know if there's a significant difference between those who did answer and those who didn't.

I think Trump's lead is real, among that subset of the population - FWIW, which is not much.

RonW
12-02-2015, 01:24 PM
Forget it, Tom... he either doesn't understand that, or doesn't WANT to understand that. He desperately hopes that the entirety of the right is extremist, like him... and any information to the contrary doesn't make it through his filter.

Oh really, well then maybe you failed to read and comprehend my post # 12 where I plainly stated......


But polls no longer are used to relay the pubic's opinions, but more to influence and push the public to go along with the majority.
They generally only call between 500 and 1,000 of whom they already know their political stance and then phrase the questions to get the desired answer.

And for your information I am so far to the right that I think rush limbaugh is a centerist, But you set a lot further to the left then I do to the right.......

Gerarddm
12-02-2015, 01:30 PM
Much depends on who will actually exercise their franchise, and not just talk about it. Thus the concern always for ' mobilizing the base '.

Canoeyawl
12-02-2015, 01:32 PM
"The modern public-opinion poll has been around since the Great Depression, when the response rate—the number of people who take a survey as a percentage of those who were asked—was more than ninety. The participation rate—the number of people who take a survey as a percentage of the population—is far lower. Election pollsters sample only a minuscule portion of the electorate, not uncommonly something on the order of a couple of thousand people out of the more than two hundred million Americans who are eligible to vote.... A typical response rate is now in the single digits."

edit to add;
"Rivers then started a company called Polimetrix, which he sold to YouGov for an estimated thirty-five million dollars. There he developed a method called “matched sampling”: he uses the U.S. Census Bureau’s American Community Survey, which surveys a million people a year, to generate a random sample according to “fifteen variables of representativeness” and to determine who will participate in polls. “You get a million people to take the poll, but you only need a thousand, so you pick the thousand that match your target population,” he explained to me."

TomF
12-02-2015, 01:39 PM
No, they poll more than that, and that keep going until they get enough people and what they think is a representative sample. ....In Canada, a representative sample of a bit under 1000 will yield nationwide trends +/- 3%, 19 times out of 20. Of course you need to enlarge the sample size to get meaningful regional or local information, but a well constructed sample can be smaller than one would think.

oznabrag
12-02-2015, 01:57 PM
"radical far left socialist." Tee hee.

. . .

I think that if certain of our RWWs actually MET a radical, far-left socialist, and heard what they had to say, they might begin to understand just how afraid the Kochs are that these ideas might catch on.

S.V. Airlie
12-02-2015, 02:16 PM
Oh really, then the next question on the poll needs to be, would you vote for an incompetent known liar and cheat for public office.

You mean a Republican RonW? NO! In fact, most Republicans can't be called incompetent because they have never had to prove they are competent except by what they say. If what they SAY seems that they would be incompetent,I can't and won't vote for them.

Trump says he's a genius and made millions. He also went through more than a few bankruptcies. Implies to me at least, he's not as competent as he thinks he is.

TomF
12-02-2015, 02:21 PM
I think that if certain of our RWWs actually MET a radical, far-left socialist, and heard what they had to say, they might begin to understand just how afraid the Kochs are that these ideas might catch on.It's worth remembering that Marx thought that communism would start in Germany or England, because the pain and structural disenfranchisement of the workers was biggest there.

In Germany, it was Fascism that came out on top, but it could very easily have gone the other way. The same disenfranchisement was at the base of each extremism.

oznabrag
12-02-2015, 02:41 PM
It's worth remembering that Marx thought that communism would start in Germany or England, because the pain and structural disenfranchisement of the workers was biggest there.

In Germany, it was Fascism that came out on top, but it could very easily have gone the other way. The same disenfranchisement was at the base of each extremism.

We live in dangerous times, my friend.