September 26, 2008

Can they do a poll to learn what kind of people hang up on pollsters?

It's kind of a problem.
Is there a certain type of person likely to refuse the probing calls? And does that affect polling numbers?

In a January op-ed in The New York Times, Andrew Kohut, president of the Pew Research Center, warned that the disparity between polls and the outcome in the New Hampshire Democratic primary--Clinton beat Obama despite polls showing him with an advantageous margin--could have been due, in part, to the fact that less affluent whites are more likely to hang up on pollsters. "These whites who do not respond to surveys tend to have more unfavorable views of blacks than respondents who do the interviews," Kohut wrote.
Oh, so if the polls show Obama winning and then he doesn't, it's because racism correlates with the tendency to hang up on pollsters?
Several pollsters I [Seyward Darby] spoke to this week said there isn't a notable disparity between the types of people who answer questions and those who do not. John Zogby, president and CEO of Zogby International, told me that over the past few decades, there has been a "democratization of refusals" and that there is a 95-percent confidence rate in polls' accuracy.

But this also means that only 25% of people go through with polls these days. So the polls only reflect what these strange people think. Who are they? Why do they lag behind the big national trend of nonresponsiveness? Who cares what they think?


John Lynch said...

(nto John Lynch)

Hmm. Busy people? People who find polls both intrusive and rude? People who have some sense of home that includes family dinners uninterrupted by phone calls?

Who are these people? And do we want their opinions?

bleeper said...

"Can the do a poll to learn what kind of people hang up on pollsters?"


ricpic said...

People resent the bias. By the third or fourth question asking them to sorta kinda admit they're racists if they won't vote Obama they get the message and hang up.

Bissage said...

Being a less affluent White, I hang up on pollsters only if they're Black.

erniecu73 said...

How about the other way around?

"- Are you a registered voter?
- No
- *click*"

Happens to me all the time.

Peter V. Bella said...

...there is a 95-percent confidence rate in polls' accuracy.

Well, he had to say that. If he did not, he may as well shut the doors and flip burgers for a living.

I always used to hang up on pollsters, just like those pesky sales calls.

MadisonMan said...

(Does the other John Lynch comment?)

Usually when the phone rings and caller ID doesn't ID, I'll answer and not say a word. If they're using technology that scans for human voice, they hang up.

X said...

i've always said polls were a great way to measure the opinions of people who have nothing better going on in their life than talking to a pollster

Ann Althouse said...

Typo in title fixed.

Pastor_Jeff said...

So if I answer a pollster's questions, does that mean I earn a "Get out of racism free" card?

Simon said...

It means that all political polls are weighted towards the minority of people who are politically engaged. People have finite time and allocate it in ways that are efficient to them, so people who have strong opinions on the issues covered by the poll have a stronger incentive to allocate time resources to answering the poll, whereas people who don't think of themselves as politically active, but who none the less vote, won't. Cf. Rational ignorance.

Whether that skews results towards the Democrats or the Republicans I don't know, but my intuition would be the former. Meanwhile, undersampling of young people because of landline biasing likely skews results towards the Republicans.

John Lynch said...


Yes, the other John Lynch comments, here on Althouse. Less frequently of late while I've commented more often of late. Still, two people, same name, both interested in this site.

The Drill SGT said...

What value are polls to me?

The Polling booth counts

I know who I'm voting for

Most polls seem to be pull polls anyway. I should spend my time answering questions either designed to change my mind or to provide info to a candidate so he can pander to me? why? Do I need to spend my time helping the MSM call the horse race? do honest reporting and then report the ballot results. who cares about the inbetween.

Let him do his own thinking, come up with his positions and tell me what he thinks.

I hate poll driven politics.

chickenlittle said...

I refuse pollsters and consider it a matter of civil disobedience.

I consider voting a civic duty.

bleeper said...

I got a call from one the other day. He refused to say who had hired him. I told him I would not answer any questions from commies. He said he was going to list me as "non-responsive". I thought "no commies" was a response.

Big Hal said...

I generally hang up on pollsters although every once in a while I take the poll and lie my butt off. It's kind of fun to choose the 'wrong' responses on a push poll. It is also kind of enlightening when I tell the pollster that I'm not voting for either McCain or Obama. They say things like "Oh you're not voting?" So I tell them about Bob Barr Libertarian candidate for President or Cynthia McKenny of the green party. which one I choose depends on my mood at the time.

wurly said...
This comment has been removed by the author.
Henry said...

Did Andrew Kohut every consider the possibility that he got the margin of error wrong? Or maybe the people who talked to his pollsters changed their minds. Or maybe the people who didn't talk to pollsters liked Hillary's more comprehensive health care plan (maybe they read Paul Krugman). Or maybe they were racists.

For Kohut it's a Hobson's choice. Either he screwed up. Or people are liars. Or rubes have brains. Or they're racists.

Ah, hell, go with the "obvious."

Jim Hu said...

Do Althouse readers hang up on polls disproportionately to the general population? I do. Perhaps I should take the poll but say I'm cruelly neutral on every question.

Chip Ahoy said...

These whites who do not respond to surveys tend to have more unfavorable views of blacks than respondents who do the interviews," Kohut wrote

This tells you all you'll ever need to never believe another poll for the rest of your life. There's absolutely no way to know that. There's no way to know the race of the person who hung up.

Yesterday I got three phone calls. One was Spring making an offer. The other two I missed because they only let it ring four times and I couldn't get to it. When I returned the two I missed, my return call was sent to a messeng service which told me they too were solicitations of some sort, possibly a poll. Irritated at being interrupted fro what I was doing, dashing for the phone but still missing the call because of the caller's impatience, I kept the phone with me and resolved to answer the next call with, "WHAT? MAKE IT GOOD ASSHOLE, I'M NOT IN THE MOOD. to throw off the caller and put them off the idea of calling again, gambling it wouldn't be someone I know.

Did you know you can enter the phone number in your browser window and see who you missed? It's usually someone you don't want to talk to anyway.

Richard Dolan said...

How to measure the views of people who refuse to be interviewed is an interesting problem. All of these statistical techniques only work if the small sample being interviewed (usually between 1500 and 3000 voters for a national poll) is an absolutely randomly selected subpopulation of the larger group (here, 130 million + voters who will cast a ballot in Nov). So the polling is only accurate if the subpopultion that refuses to be interviewed matches exactly the characteristics of the larger population that the pollster is trying to evaluate. If it doesn't, then the poll won't give you an accurate snapshot of the larger population; if it does, then the poll will (provided you avoided any other problem that would impact on the randomness of the sample). Even a small self-selection bias can skew the results. (Of course, skewed results are sometimes what the pollster or the politico commissioning the poll wants.)

All of this is of interest because the conventional wisdom has settled in (at least in the usual places) that if O loses, it will be because of racial factors. People will tell pollsters that they intend to vote for him, and then won't.

In the primaries, O sometimes did better and sometimes did worse than the polling suggested he would, but the variations were usually within or close to the margins of error in the polls themselves. And the variation in his actual vote vs. polling prediction didn't account for why he lost. Going into the PA primary, for example, Clinton was polling about 50% (RCP avearge). She got 54%. O was polling 43%; he got 45%. The NH primary is usually cited as the biggest indicator of O's polling results suggesting a racial bias. But in the final RCP average, O was projected to get 38% of the vote; he actually received 36%. Clinton's final polling average was 30%; she actually pulled 39% of the vote. In Ohio, too, O's final polling number (again, RCP average) was 43%; he got 44%. Clinton's final polling average was 50%; she got 54%. In each of those examples, it just wasn't a case of people saying they were for him and then voting for her.

I don't see how you can read these poll results and conclude that people were trying to hide their actual attitudes about O. What happened was that most people who weren't for him as the election approached just never went with him. He couldn't close the sale. People have doubts about him and he wasn't able to allay them. If this pattern repeats in Nov, it suggests that if O is below 50% in the final polling, he is likely to say below 50% when the votes are counted. In all events, the primary results don't support the speculation that people either misstate their actual views to pollsters, or refuse to be interviewed altogether, because they want to hide negative views about O because of his race.

Bissage said...

Speaking of messing with pollsters, here’s a fun one (with many possible variations) to try when you’re feeling a bit too mischievous:

THEM: Would you like to answer a few questions?

US: Certainly. For how much?

THEM: Sir?

US: How much will you pay me? I’ve got something you want. Why should I give it away for free?

THEM: Sir, I’m not authorized . . .

US: But you’re getting paid aren’t you? Are you saying your time is worth more than mine? Did you call just to insult me?!

THEM: Ah . . . No, sir . . . I, ah . . .

US: How dare you!


(NB: The slam works best with a rotary phone. Maxine wouldn't get a detail like that wrong.)

Mej said...

As a person who writes, conducts and analyzes polls for a living, I feel obligated to answer whatever other polls or surveys are presented to me.

holdfast said...

Of course, if Obama is polling ahead but loses, there will likely be riots. If he loses the electoral college but wins the popular vote, there will be large riots.

On the upside, a lot of people's heads will explode and I will laugh.

former law student said...

I only answer the phone if I'm expecting a call. The rest of the time I let it go to the answering machine to see who's calling.

If they said they were a poll I would pick up.

Regarding who's behind the poll. Knowing ahead of time might bias your responses. But there should be no harm in telling you afterwords, so they should agree to it.

There's no way to know the race of the person who hung up.

Even if they know they called Zbigniew Brzezinski?

Conversely, the chance that an American named Washington is white is infinitesimal.

ABP said...

I've never hung up on a pollster. I've never been polled. That may be because I don't pick up the phone if I don't recognize the number (especially if it's an 800-style number). Do pollsters rely entirely on the non-caller-id-using population, or am I an outlier in this regard?

Arturius said...

There's no way to know the race of the person who hung up.

Even if they know they called Zbigniew Brzezinski?

Conversely, the chance that an American named Washington is white is infinitesimal.

That's a fascinating observation because Ryan McNeil certainly doesn't look Scots/Irish to me.

michaele said...

Caller ID has had a huge influence on my phone answering behavior. If I don't recognize a number, I let it go to the answering machine. Unknown callers almost never leave a message. When I stayed at my 89 year old mom's for a couple of weeks who doesn't have caller ID, there were frequent poll calls. I suspected she was on a list of dependable answerers. We are pretty much opposite politically so if I took the call, I had fun giving whacko anwers.

Windbag said...

What about the pollsters who hang up on me, when I'm clearly not giving the responses they wanted to prop up their "data?"

The last pollster I agreed to talk to didn't have a clue what she was talking about. She was reading from a script. Quote: " that like a liberal?" At least she was willing to admit her ignorance on the issues she was polling me for. How many other pollsters just fill in boxes and keep going.

bagoh20 said...

Zogby: "there is a 95-percent confidence rate in polls"

While Zogby himself is pretty accurate, I remember the LA Times polling Schwarzenegger down by 20 points 2 weeks before he won handily.
Margin of error +-30

former law student said...

The last pollster I agreed to talk to didn't have a clue what she was talking about. She was reading from a script.

To be able to compare the responses, they have to standardize the questions. If the pollsters start winging it, they introduce a variable which might bias your answers in an unpredictable way. Therefore everyone calling for a particular poll must use the same script.

Harwood said...

I regret that I have but one phone to hang up when pollsters call.

Harwood said...

A thought experiment: What if, overnight, all polls ceased to operate. No polls, zero. What difference would it make to the democratic process?

rhhardin said...

One unavoidable problem with polls is that public opinion does not exist.

It's not that it's hard to measure, but that there is no such thing.

Public opinion never speaks for itself. It's only cited. It requires newspapers.

A vote is another matter. People decide, and take responsibility for their decision, if only silently.

A decision comes out of a vote, not an opinion.

bagoh20 said...

"A thought experiment: What if, overnight, all polls ceased to operate. No polls, zero. What difference would it make to the democratic process?"

I would expect higher participation at the pols at least since nobody would either give up or feel safe ahead of time.

We may get the same result if they just become unreliable to people. Unfortunately, I think people like polls when they agree with them regardless and therefore they will live on.

madawaskan said...

OK let's walk everyone through this-particularly the-

Zogby believers! {Honestly}

California Democratic primary-

last polls before the actual results-

Zogby had Obama winning by 16%

Rasmussen had Obam winning by 1%

Suffolk had Obama winning by 1%

Actuall bloody results-the one that counts on voting day-

Hillary Clinton beat Obama-

51.9 to 42.3 a difference of 9.6% in elections that is huge.

But hey what's even bigger than that?

Zogby's margin of error.

Because not only is he of by the 13% you have to add that to the 9.6% victory that Hillary actually achieved...

That makes Zogby's margin of error-


The Zogster...

Richard Dolan said...

Mad: I agree that Zogby is not very reliable. But even in the CA example you cite, the more interesting fact is that O has consistently shown that he has not been able to sway voters at the end of the race. His final polling result in CA before the primary was 44% (RCP average); he got 42%. Clinton's final polling result was 43% (same average); she got 53%. The polling on O's support was within the margin of error. But everyone who wasn't already with him broke for Clinton. Same pattern as in PA, NH and OH: he hasn't been able to move voters who weren't already with him. He has a problem if, come Nov 4, he is polling under 50%. His track record has not been good in moving voters from 'undecided' to the O column.

Metanis said...

I just explain politely that I don't have time for their propaganda and then I hang up.

rhhardin said...

I refuse to answer their questions by misunderstanding and evasion. Real pollsters are not allowed to hang up the phone themselves, and the idea is to keep them on the line as long as possible.

Assuming you welcome the break from whatever you're doing.