And the Winner Isn't...

  • Playlist
  • Download
  • Embed
    <iframe src="" width="100%" height="290" frameborder="0" scrolling="no" title="NPR embedded audio player">

Listen to this 'Talk of the Nation' topic

Polls made a mess of things Tuesday night.

Polls made a mess of things Tuesday night. Source: Getty Images/Mario Tama hide caption

toggle caption Source: Getty Images/Mario Tama

OK, everyone who thought Hillary Clinton was going to win the New Hampshire primary, say "yea." In a room full of pollsters, all you'll hear is crickets. Just about every one of them called for a comfortable Obama win, and reporters anticipated the subsequent "next big story": a major shakeup in the Clinton camp. Sorry, Charlie. So what went wrong with the polls? Sure, the nice weather may have helped Hillary, whose demographic skews older than Obama's, but that's not enough to account for her surprise win. Pollmaster Andy Kohut says the answer lies in the profile of folks who respond to polls — which is not the same as who votes. It's got a lot more to do with race and class than gender and age, and it may surprise you. Do you respond to polls? Why or why not?



Please keep your community civil. All comments must follow the Community rules and terms of use, and will be moderated prior to posting. NPR reserves the right to use the comments we receive, in whole or in part, and to use the commenter's name and location, in any medium. See also the Terms of Use, Privacy Policy and Community FAQ.

What went wrong with the (pre-event) polls? Nothing. Polls are only as good as the sample size. Its tea leaves!
But, something that may have happened. When the middle-aged women heard that Hillary was falling behind (on Monday), maybe they took their soccor babies to the polls AND VOTED. Good for them.

This highlights and issue I have with NPR, ABC, CBS, NBC, AP, Gallup polls being released so close to voting day. Their trying to report on news that HASN"T HAPPENED YET! (Scooping the other networks.) And they egt it wrong sometimes. Maybe the fickle voter has finally found some backbone! And left the corrupt, network dominated, machine politics behind.

Sent by Harold | 1:55 PM | 1-10-2008

Why is Hillary's win in NH such a big deal - both she and Barack Obama were awarded the same number of delegates - 9. The news is treating her 3% point win as big news. The only reason it appears as an upset is because of the flawed predictions of the news media!

Sent by Paula Rucker | 2:11 PM | 1-10-2008

Plus, as said above, you can't categorize people so easily. A poll asks a certain group of people certain questions and then we take those averages and multiply them. At least this is my understanding. Those may work for philosophies or strategies, but people can't be categorized. Emotions, personal stories, and connections with candidates can change day to day for anyone.

Sent by Joseph | 2:13 PM | 1-10-2008

"Change" is often necessary but always frightening. I suspect that there were many people who entered the voting booth with the intention of voting for Obama and before they could vote realized that they didn't really know much about him, that the person at the wheel who wants to take a new direction had better know for certain where he was going and what he would do if he made a wrong turn. I think the car analogy is a good one -- do I know this person well enough to get into his/her car. Voters think they know Hillary Clinton much better than Barak Obama.

Sent by Kevin Gross | 2:29 PM | 1-10-2008

for the past 19 (almost 20) years, the white house has been occupied by the Bush families and Clinton's. In my opinion all 3 administrations have dabbled in/used corrupt politics as a means to an end, and Hillary has taken part, if not orchestrated some of these actions. it's time for change. it's time for a new family to take residence in the white house. although he is unseasoned, obama is by far a better choice over hillary. far as i can tell his hands are not dirty like hillary's have been in the past. i hope the voters of this country consider that
our nation shouldn't have deal with the inevitable cover-ups and questionable motives should hillary be voted in to office. i say enough of the bush's and clinton's. we need fresh new blood.

Sent by chris McCluskey | 2:31 PM | 1-10-2008

I did precinct work for Harold Washington in 1987 when he ran for re-election as Mayor of Chicago. It was clear that white people would tell me (also a white person) that they supported Washington, but when it came down to the privacy of the polling booth, they just couldn't bring themselves to vote for a Black man.

Sent by Peggy Baker | 2:46 PM | 1-10-2008

Recipy for Hillary's win in NH:
One part crocodile tears and theatrics for the"lady" vote. Add one cup of
Racism. Voila Hillary wins.
I believe Hillary's tears as much as I believed that her husband didn't have sex with "that woman."

Sent by MO | 2:47 PM | 1-10-2008

The more interesting thing on the republican side is that aparently people can't agree on how to distribute delegates. By some Huckabee is in the lead by 12 delegates in others Romney is ahead by 6 or more. Which is even more suprising since a lot of the media is saying Romney is a dead act. If the media only reports the fact then we have multiple facts.

Sent by Tony Bartholomew | 2:47 PM | 1-10-2008

Can the polling problem be just mechanics? - 1) polling stopped on Sunday instead of Monday and 2) the Hillary Clinton campaign did a great job of getting out absentee votes in the weeks before Iowa.

Sent by Jack Johnson | 2:48 PM | 1-10-2008

I wonder how many independents who told pollsters they planned to vote for Obama decided at the last moment to cast their votes for John McCain, since they found the other Republican candidates so unsavory.

Sent by Amy Buzby | 2:49 PM | 1-10-2008

When will we get rid of these primarys that allow a few people in a few states determine who the rest of the country gets to vote for? I am for Hillary Clinton and almost did not get a chance to vote for her.

Sent by James | 2:52 PM | 1-10-2008

I, personnaly, resent the opening representation that Clinton's supporters are poor and uneducated while Obama's supporters are well educated. I'm neither poor nor uneducated. I'm a professional woman from Illinois with a higher degree. If you follow the issues and the candidates' records, it is much more likely that intelligent minds may differ regarding the best candidate based upon the subjective issues important to the voter.

Sent by PJ | 2:53 PM | 1-10-2008

What about vote tampering as a possible cause? Blogger Bill Noxid compared hand counted ballots to computer tallied votes and found Obama had 7.5% lead in hand counted ballots and Hillary ahead by 5.5% in computer Accuvote tallies.

Sent by Erika | 2:54 PM | 1-10-2008

I have always wondered HOW does one GET polled? I would be a very active participant, yet never once have I been polled. I think this question is pretty important especially in light of the comments about poor or otherwise "inaccessible" voters not being involved.

Sent by Jenn Flanagan | 2:56 PM | 1-10-2008

I don't think I buy the low income, less educated who don't answer polls factor to account for the reason for Clinton to get more votes than predicted because of the race factor. It could be reasoned that the same type of person who wouldn't vote for a black man for president also wouldn't vote for a woman for president. I think the errors occured because people changed their minds rather quickly when the events surrounding Mrs. Clinton unfolded.

Sent by Kat Morgan | 2:57 PM | 1-10-2008

I'm wondering if there are data re: the hypothesis that some independents who were inclining to vote for Obama went to McCain instead given the forecast of a big Obama win on the Democratic side.

Sent by Richard Holub | 2:58 PM | 1-10-2008

I am wondering when polls become detrimental to the race's outcome. I am thinking specifically of a woman I heard on NPR yesterday who, when interviewed, explained that the reason she voted for McCain was that she understood from the polls that Obama was far enough ahead so that he didn't need her independent vote. How common is this, and when do public polls start to affect how people actually vote?

Sent by Becky | 2:59 PM | 1-10-2008

I don't see this as involving race. The theory as I understand it is that whites say they will vote for the black so as to appear to be politically correct and then vote for the white guy. Here you had two politically correct one would have worried that people would think they are not PC if they said they were going to vote to put a woman in the oval office. I am sure polls involving a race between a black and white can be distorted by the race issue but I don't see it here.The distortion requires not only that you have a politically correct candidate but also need to have a politically incorrect candidate.

Sent by Joe Sewell | 3:00 PM | 1-10-2008

Perhaps since Hillary Clinton's constituency is supposedly so poorly educated and unsophisticated they could not understand the questions posed by the pollsters....give me a break!

Sent by Kristy Richardson | 3:00 PM | 1-10-2008

It seems quite bizare that very few people including NPR are not asking the more important question: What if the polls got it right? Which is to say, I looked VERY closely at the results of the voting in each district, Hillary fared VERY very well in all the districts that had electronic Diebold voting machines and Barack did much better in places where it was paper ballot. So it becomes more obvious that the polls were correct and it is the very easily hackable Diebold voting machines that are helping Hillary create a comeback that does not exist in reality.

Sent by Andrew | 3:00 PM | 1-10-2008

1. From a technical perspective how does the analysis models account for uncertainty? I believe there is an uncertainty principle at work. you know the candidates, the parties, but you can't know what all the people think at any one time. Things like unconventional or new types of candidates (Paul, Obama, etc) can have big impact on how people respond, are we accounting for this type of uncertainty, Can we?

2. What if reporting poll results were banned until the elections are completed. If the media wants them thats fine, they can help drive the media but they shouldn't drive democracy. Previous questions about polls driving the media or media driving polls and less popular candidates not getting enough media are certainly impacted by the assumption that one candidate or another is a cinch to win. As in the case of New Hampshire.. What recourse does the public have when polls are wrong? We know that people criticize politicians for being poll driven, but there is an aspect of polls like these driving peoples decision on voting.

Sent by Dave | 3:03 PM | 1-10-2008

Is it true that the actual vote count for Obama and Edwards fell within the margin of error of most polls? That only the vote count for Clinton was underestimated?
If that's the case, the experts need only explain the surge for Clinton. It rather than other factors would explain her unexpected win.

Sent by Ron | 3:06 PM | 1-10-2008

In response to a caller on Talk of the Nation who asked, "Are the polls skewing the media or is the media skewing the polls [sic]."

It's not a linear, deterministic situation: they influence each other. A sort of feedback loop can develop which is why news analysis shows like "Talk of the Nation" are important.

The reality is that polls are media. "The media" just reiterates, amplifies itself when it reports on polls, which is also an important function.

Sent by Kaleb Coberly | 3:06 PM | 1-10-2008

In my opinion, two things happened after pollsters stopped polling. The iron and Iran. The sign that said "iron my shirt" ticked off a lot of people, who then turned out to vote or changed their vote; and secondly the showdown in the straits of Hormuz, bringing fear back into the equation - and who was experienced enough to handle it. The two that came in first decided that answer.

Sent by Carol | 3:17 PM | 1-10-2008

Why is it that the GOP polls were correct and the Democratic polls not? Were the same pollsters doing both? Was it just margin of error? At the close of voting the "pundits " declared Obama would win with 39% to Clinton's 34% based on exit polls not on prepolls. Why would folks lie after they have just voted? This is not logical..

Sent by Mary | 3:37 PM | 1-10-2008

One reason the polls ended up being wrong relates to the quirky nature of the NH Independents. I worked the local voting poll all day on election day. On Election Day, at least a half dozen people said to me "I'm a huge Obama supporter and will vote for him in the general election but today I voted for x Republican because Obama's got a lock on the win (because all the polls said that he was up by 10% or so) - and I want to make sure x Republican does/doesn't get the Republican nomination. The day after the election, two more friends told me that'd they'd done the same thing and felt terrible that Obama didn't win after all. You can say that these folks display a lack of understanding of how electoral politics work - and how important each delegate is - but that doesn't change the fact. Many Obama supporters in this state thought they were safely gaming the system - and we all saw the cumulative result!!

Sent by Kate Reed | 3:55 PM | 1-10-2008

Are the pollsters now trying to effect future caucuses and primaries? In all the stories explaining their failures, the pollsters have characterized Obama supporters as intelligent, wealthy, sophisticated and all-around cool. Clinton supporters are old, ignorant, poor racists. Gee, what group would you want to join?

Sent by Susan H. | 4:17 PM | 1-10-2008

On nearly every NPR program that interviews potential Democratic voters an interviewee makes the statement that they are voting for a candidate based on the fact that the candidate is a minority.
Whether it is because they feel that a black man or a woman of any color could bring something needed to the presidency, or that it is time to give someone other than a white man a chance, it is nothing better than hypocritical. To begin with, shouldn't we be voting for the person who will be the best leader of our country, regardless of sex or race? And perhaps more importantly, the assertion that a person could be better qualified for office because of his or her race or sex lends legitimacy to those who would say that a woman or black person by their very nature can't run the country as well as a white man. Isn't this what most supporters of the Democratic party have been fighting against for the last 50+ years?
Where do I, as a white male, cast my vote? Is it OK for me to vote for Obama, he's a man, and his mom was white, so can I vote for him? May I vote for Hillary? (It appears that both of her parents were white.) Or do I need to vote Republican regardless of the candidate cause they're all white guys like me? Perhaps I'm stuck with Edwards, he's white and from the South, after all I have little in common with an Italian American, there's nothing similar between me and a cor-pirate Mormon, and I think we've had all the radical religious leadership that our country can take. Should I still cast my vote for a white guy?
Personally I feel, have felt for nearly ?? century, that it is my duty to vote for the person who will be the best leader of our country. It might mean that I have to pay more taxes, it might mean that someone gets more from the government than I do, I might even have to shift over a bit and give someone with different beliefs equal ground.

Sent by David | 4:24 PM | 1-10-2008

The final comment and Andy Kohut's response triggered a slightly off topic thought. When you favor, or approve of, two or more candidates over the entire field, you probably needs to check the polls to see who to vote for, given the flawed voting system we are stuck with. Let's say you favor Richardson and Obama over everyone else because you consider them the most positive. The only way to pick who you vote for is to check the polls, so you know to vote for Obama.

The fix is a switch to approval voting: you vote for everyone you approve of, and don't vote for those you don't approve of. The winner will be the closest to the concensus candidate because more people approve of the winner, even if the winner isn't the first choice of a plurality.

However, this system is so foreign, introducing it will require a long period of education.

I would like Andy Kohut to comment on what he thinks about Pew adopting such a system in their polling. I've been polled on products and concerns, and one pattern is to ask the following:
1) for X what name comes to mind?
2) which of the following names do you recognize?
- a
- b
- c
3) which name do you most commonly chose

If this type of poll had been done last weekend and only question two reported, the debate on Monday would have been along the lines of: "All the democrats have very high approval ratings, and Hillary and Obama have the support of two-thirds the voters. So, which of their favorites will voters pick to cast their vote? The Dems and undeclareds really like them all!"

Sent by mulp | 4:36 PM | 1-10-2008

For Andy Kohut to not even acknowledge, let alone address, that the internet is abuzz with conspiracy theories regarding electronic vote fixing can only contribute to their proliferation. If there is some logical refutation of those who think that secret vote tabulations by companies controlled by Republican partisans is a concern, I for one, would dearly like to hear it articulated.

Sent by Doug Clayton | 8:29 PM | 1-10-2008

Could your polls be showing another factor that works along with this!. Women are in lower income brackets ( when income is alone) Women have less of a voice when women and men are talking in the same place. Women will give their true opinion when men are not present, ie in private. All of these factors do not suggest that poorer women (than rich women) are more biased agaist a dark skin candidate. Women may be more biased toward women when giving their opinion in private. I do see more higer educated votes going for Obama, but not all. I have worked on opinion polling a lot in the past as a graduate student in Sociology.

Sent by Gail Woodbury | 10:57 PM | 1-10-2008

The data is from:

20% of the Votes were hand counted.
80% of the Votes were Diebold counted.

The hand counted vote shows a 4% win for Obama.
The Diebold count shows a 5% win for Clinton.
If you counted only the hand counted ballots, Barack won.

Fortunately, as we all know, computers NEVER makes errors and are NEVER hacked. So this is not even worth exploring or considering, right?

Sent by Lori L. | 10:17 AM | 1-11-2008

Polls were wrong because there was a media coup against Hillary which did not work; because people of NH new better.

Sent by Nancy Sabet | 1:49 PM | 1-11-2008

As a progressive from New Hampshire, I think Edwards stole votes from Obama. I was teetering on the Obama/Edwards fence and finally chose Obama. No one in NH viewed Clinton as progressive especially with the heavy-weight establishment democrats in her court.
If he is serious about supporting solutions, Edwards should throw in the towel and get behind Obama.

Sent by David Stamps | 2:03 PM | 1-11-2008

I think the public opinion polls are great tools for candidates to use, but I don't think it's something that should be relied upon as truth. There are plenty of surveys taking place all over the country on a number of topics; but we don't hear about them. Why? Because we're not the ones concerned with what percentage of middle income people choose Kellogg's over General Mills cereals. Climates change, weather and political alike. Nothing went wrong.

Sent by David | 2:10 PM | 1-11-2008

NPR thanks our sponsors

Become an NPR sponsor

Support comes from