Network News

X My Profile
View More Activity

About Those Democratic Polls

By Jon Cohen and Jennifer Agiesta
Yesterday's Democratic result is sure to fuel debate among poll-watchers about the accuracy of polls in contests with African American candidates. In several well-known, but long-ago examples, pre-election polls underestimated support for the white candidates in such campaigns. But in the 2006 elections, a strong showing by polls in elections with black candidates seemed to finally put that notion to rest, and there was no apparent problem with reliable pre-election polls in Iowa.

A more likely culprit than the role of race in the New Hampshire election was the "likely voter" modeling, with pollsters perhaps over-counting the boost of enthusiasm among Obama supporters following his victory in Iowa. Another possibility is that independents opted at the last minute to participate in the Republican primary, depriving Obama of crucial voters.

A further potential source of error stems from New Hampshire ballot rules. In previous contests, the state rotated candidate names from precinct to precinct, but this year the names were in alphabetical order, with Clinton near the top and Obama lower down. Stanford Professor Jon Krosnick, a survey specialist and expert witness in a lawsuit about ballot order in New Hampshire, has estimated a three percentage point or greater bounce for a big name candidate appearing high on the ballot. Therefore, if pre-election polls randomized candidate names, as most do, they would have underestimated Clinton's support by at least three points.

Regardless, there were no immediate clear answers, and lots of data analysis ahead.

By Web Politics Editor  |  January 9, 2008; 12:55 AM ET
Categories:  Barack Obama , Hillary Rodham Clinton , Primaries , The Democrats , The Pollster  
Save & Share:  Send E-mail   Facebook   Twitter   Digg   Yahoo Buzz   StumbleUpon   Technorati   Google Buzz   Previous: New Hampshire Voters Take Independent Tack
Next: Reports of Clinton's Death, Greatly Exaggerated


Hello its a very nice site! " target="_top">genuine online trading forex currency forex learn

Posted by: Braden ffawz | April 9, 2008 12:49 PM | Report abuse

it all boils down to this who cars whos ahead and what states they have vote who you like the most..everyone has the instinct to know who they can trust just by listining to them..for example if you can watch hillary cry an actually belive that its for real than u have a problem if u think that she deserves a grammy for it than u know whats goin on...presidents (the commander and chief of the united states)should not be crying about anything...and if ur trying to win an election and u fake cry a couple of times dont expect any simpathy..and yes if she does win it is because she is a women not being sexist but there are thousands of women just voting for her so they can stick it to the men and show every body that they can do anything that the guys can...on the other hand a lot of white guys are going to vote for her because they are racist against blacks...yes its not a new concept..they would rather vote for a bad presidential canidate becuse the other demicratic canidate is a black man...whitch is completly assinine...this election this country is on a knifes edge stray but a little and we are going to fall off...if u are going to be racist or sexist in this election it is better for the american people that u just dont vote at all...i am a white male voting for obama not becuse of his skin or his gender but because he is an inspirational man that will deliver when it comes time to americans need to be united hense our country's name

Posted by: infantry_killsall | February 8, 2008 4:50 PM | Report abuse

It seems like everyone is way too wrapped up in the polls and what they think we are thinking. Perhaps the citizens of the US should stop worrying what others think and vote for who they think is the best candidate. For the last two weeks the TV media has had almost nothing about the candidates and there position on the issues, just story after story about the latest poll results and who is leading. It seems the polls have become a tool to influence the outcome rather than predict an outcome. We should see more information on profiles of each candidate, how they have voted in the past, what they have supported, instead we get sound bites that are chosen to support the broadcasters political views.

Posted by: john316xyz | January 12, 2008 8:30 AM | Report abuse

The truth is that the media bias against Hillary and their effort to fix the primary votes in favor of Obama created a backlash. Attempts to demonise one candidate to prop up another is bound to offend fairminded voters. After all, most voters were behind Hillary before the Obama mania started just before the Iowa elections. I have watched all the debates; not one of the questioners asked Obama any searching question, but to Hillary it was all about her past and her unelectability. Come on, the fact is that Obama has no real past, and the hype around him is hollow except for his oratory and youthhood. If you want to elect a black president, there are many people who have stood in the line of fire for civil rights, Obama is not one of them. If he were, he would not be the favorite of the media. Hillary and her husband have a much better record on supporting black causes than Obama. The media attempt to make a fairy tale about Obama succeeded initially in Iowa, but generated a backlash against the venom shown towards Hillary. What did she do to desrve this witchhunt against her in the media? The fact is that Obama is a very good future candidate, but is not ready yet to lead. The media attempt to back him recklessly against Hillary may create an ironic general election Republican win that even the blacks don't want. Wait until the Republicans start withering attacks against him using all dirty tricks (including I should regretfully say the race card), and New Hampshire result will look like a minor bump. I think that a healthy debate, where the media plays fair and square, will see the best candidate to emerge with no surprises. It will serve the democrats and the country well. It is time that there is a reality check!

Posted by: vaidyatk | January 11, 2008 8:09 AM | Report abuse

I find all these comments very interesting...Here is a case of reality...
My daughter is in Concord, NH. Last night she called around to about 20 of her women friends and colleagues (mostly "undeclared" voters) who as late as Sunday had either been still undecided, leaning toward Obama or going to vote for Obama. She got the same story from all of them...they were irrate over their perception that Hillary was being held to a different standard (by the mostly male pundits). They mentioned the way she was treated after the debate and the "over the top" focus on Hillary's "emotional moment". It seems that many women just got fed up with the white, male commentators and decided, "You go girl!"

Sue in VT

Posted by: smahoney1950 | January 9, 2008 1:48 PM | Report abuse

yes.. the NH primary voters get quite a lot of credit for being "strategic" thinkers (obama upporters vote for mccain; edwards supporters vote for clinton).. I suppose this could be true.. but more likely.. this is just another example of the media trying to make a HORSERACE. I can't wait to see how they will spin the general election to keep it close!!

Posted by: tdloendo | January 9, 2008 1:19 PM | Report abuse

Why would the Edwards supporters go to Clinton? That is a strange one, considering his rhetoric. Of course the results you get with a caucus (i.e. 2nd choice, viability, all that) are much different than the results you get with a primary, so people are thinking strategy to some degree. McCain would be out of the race without a win here, and Obama could stay in either way, that may have stuck with some of the independents. Plus with the media reporting a double digit lead people might have assumed it was no big deal.

Posted by: grimmix | January 9, 2008 12:30 PM | Report abuse

Hey...has everybody forgotten that we're at war in Afhganistan and Iraq?

Posted by: dfc102 | January 9, 2008 11:29 AM | Report abuse

Short answer. No one lied. The polls weren't wrong except Zogby. The numbers weren't analysed for trends and the Saturday top of the bounce wasn't eliminated. Why??? When everyone knows the Iowa bounce comes down. Always has. Always will.

The numbers were not wrong. No one lied. It was misreported in the press and by the polsters.

Posted by: slbk | January 9, 2008 11:24 AM | Report abuse

Re: College Students

In Iowa, it wasn't out of state students who could vote, but rather residents of another state who are going to college or university who could vote. I went to college in Minnesota and grad school in Iowa and caucused in the states where I was attending school, rather than voted in the primary where my parents lived. It's a matter of where you decide to have your legal residence.

Posted by: FairlingtonBlade | January 9, 2008 11:23 AM | Report abuse

This comment was posted to the fix but is equally relevent here:

Clinton did not defy the polls. As I tried to explain yesterday on your blog, the numbers clearly showed Obama trending downwards and Clinton upwards. This happened over three days, starting Saturday night after the debates.

This is not remarkable. In fact, it is completely standard for a bounce to come down. It always has. It always will.

I predicted a 2% win by Obama yesterday based on hype carrying the day, but qualified it and said Clinton could win if youth and independents dont turn out in record numbers, which they didn't.

At least two polls only showed a 1% obama lead on the eve of the race. Look at Rasmussen
Sat 14% lead Obama
Sun 6% lead obama
Mon 1% lead. obama rolling average 7% lead

Extrapolate Tuesday 4% Clinton win, which is exactly what happened. The question is why didn't reputable polsters throw out the data from Saturday and just do a reliable one day poll?

Ther question you need to ask is why didn't the MSM see the clear trend (with the exception of Zogby)

In fact its even clearer in South Carolina had anyone bothered to look (and even I didn't, but I could have used these numbers yesterday to demonstrate the downward trend)

South CArolina's numbers from Real Clear Politics.

04-06 obama 20% lead
06-06 obama 12% lead
07-07 obama 7% lead

NOw you do the math. If the vote had been in SC yesterday, would Obama have won by 13% the average or would the trend prevail with a close race and possibly a Clinton victory, with standard turnout demographics and a more solid base of support?

The number were clear. The trend was clear. MSM chose to ignore it and the polsters decided to include Saturday in rolling averages enthough a bounce coming down is standard not rocket science. Zogby probably modified its sample to reflect Iowa turnout of youth and independents, bad idea.

I think there was a masssive bias against Clinton by you ,and basically everyone else, who didn't take the time to look at the trends, which are always more important than the rolling numbers themselves.

It's not American Idol, its the Presidency
of the United States of America. Try to be a little more neutral, add critical tough questioning of all candidates, and leave the high school idolization at the door in future please.

Posted by: slbk | January 9, 2008 11:20 AM | Report abuse

The idea that the error of the polls was the fault of lying low-life racist New Hampshire voters is ludicrous. The accusation says more about the media than it does about the voters.

Posted by: glclark4750 | January 9, 2008 10:57 AM | Report abuse

An earlier poster mentioned a possible defection of Edwards supporters to Clinton. I'd like to hear more about that.. The bottom line for me is I am so happy that the DEMS finally have several GREAT candidates to choose from.. any one of them will trounce the GOP candidate come next Nov!! Also.. I wish Joe Biden would get off of the fence and endorse on the candidates. I love his common sense approach and respect his opinion.

Posted by: tdloendo | January 9, 2008 10:43 AM | Report abuse

ps re: absentee ballots. Clinton's margin seems to have greatest among senior citizens, precisely those most likely to have cast pre-Iowa absentee ballots. (Seniors wintering out of state would also have been missed by pollsters, of course). Again, the key question is what percentage of votes were absentee, and whether early or late deciders made the difference.

Posted by: whunt | January 9, 2008 10:32 AM | Report abuse

Two weeks ago, Hillary Clinton was considered the prohibitive favorite in New Hampshire. Instead, Hillary won NH by only a razor-thin margin. The fact that Barack Obama won 37 percent of the vote to her 39 percent is a tremendous accomplishment for Obama. If he had not beaten her in Iowa, his surprising triumph in the NH primary last night would be the clear headline this morning.

The pollsters proved that they are faillable and the NH voters proved that they are willing to lie to pollsters to avoid being perceived as racists.

Posted by: dee5 | January 9, 2008 10:27 AM | Report abuse

Most of this discussion is beside the point if coumaris is even half-right about a large proportion of absentee ballots being cast before the Iowa primary and its resulant "bounce." The facts can't be hard to establish-- so what are they? The issue is whether Clinton's victory was due to "late deciders" or "early deciders". If a significant fraction of the ballots were cast before Iowa, than the polls may have been fairly accurate all along, since Clinton was leading at that time.

Posted by: whunt | January 9, 2008 10:05 AM | Report abuse

I am so sick of Hill and Bill talking about how experience is the most important quality in a candidate. They are making the case for electing Bill Richardson, not Hillary Clinton.

Posted by: NMModerate | January 9, 2008 10:03 AM | Report abuse


I don't know if out of state students voted in the Iowa caucuses.

If that is true, then it would mean that Obama flooded the Iowa caucuses with college students from Illinois.

There is a name for that: WHOLESALE FRAUD.

Posted by: tropicalfolk | January 9, 2008 9:50 AM | Report abuse

How rediculous can you get? No one is stupid enough to vote based on whose name appears first on the ballot! You can't claim name recognition either. No one who hadn't heard of both Obama and Clinton would have bothered to go to the polls! Up until Iowa Obama didn't have that great lead in the polls and after Iowa, the polls were taken with way too small a sampling size. A couple of them had less than 400 people in their sample with 5% error. Based on a 5% error the final results of the election were not actually outside the predicted spread by more than a point. And as someone else pointed out, they sampled who they thought might vote, not who actually voted.

Posted by: jennythacker304 | January 9, 2008 9:38 AM | Report abuse

Give coumaris a medal. You have nailed it, and this issue, the issue of the absentee voter, needs to be analyzed carefully. Personally, I have always voted in person at my polling place since retiring from the military. There is just something special to me about physically being first in line and physically placing my ballot in the box. I do know that many times, my absentee, filled out in a hole somewhere far, far away, never got to the county registrar, and when it did, my votes were based on very limited and very dated information about candidates and issues. I think we are making a mistake on the progressive side by pushing absentee registration. Progressive candidates tend not to have the same advantages of constant and continuous media attention, financial support, and organization. In both Iowa and New Hampshire, politics is "retail' in nature, small groups and events, individual and chance meetings with candidates, and organizations staffed with your neighbors, not professionals from Wash DC. Voting on the day of the election, in person, allows one to absorb all of those messages. The absentee voter might regret having sent in that ballot 4 weeks before the candidates showed up on his or her doorstep. Semper Fi

Posted by: nrringlee | January 9, 2008 9:37 AM | Report abuse

The polls were wrong because polls are sometimes wrong. Usually they are only designed to be accurate 95% of the time. Sometimes the outlier result gets published. It's also very possible that many voters changed their minds or made up their minds on the last day.

Hopefully Democrats will be mature enough to pull together to support whoever ends up being the party candidate in the general election. Last thing we need are some neo-Naderites whining "So-and-so isn't good enough, so might as well let the Republicans win." That has not helped the country this decade.

Posted by: lartfromabove | January 9, 2008 9:33 AM | Report abuse

It is simple to explain what happend. Clintons managed to put just enough doubt on Obama's 'change' and 'hope', in the minds of voters to force them rethink.
Clintons used Karl Rove's book (He was able to put just enough doubts about John Kerry in 2004, to scare voters).
Clintons used the words like 'talker' vs 'doers', 'no beef in Obama's rhetoric' and 'fairy tale' etc.
It all started on Saturday debate when she was shrewed enough to grab the chance and gave a lecture on change. I think Obama is finished if he opts not to attacke Clintons.

Posted by: zic786 | January 9, 2008 9:20 AM | Report abuse

Silly is how I would describe basing an election on Iowa and 12 voters from a small New Hampshire town. But that's just me! More and a linkback to your WaPo article:

Posted by: davelucas.notes | January 9, 2008 9:01 AM | Report abuse

The spin put on this so-called win for Hillary is rediculous. It was a tie Obama received 9 delegates and Hillary received 9.

Posted by: patriceaustin | January 9, 2008 8:48 AM | Report abuse

1. Independent voters supported McCain to keep him in the race
2. Thinking Obama would win easily, other independents and D's assumed their votes weren't needed and they stayed home
3. NH HATES to be told their vote doesn't matter, that the results are a foregone conclusion
4. Half of the votes were cast before Iowa (= no effect from the Iowa bounce-- Huckabee also didn't get a bounce)
5. One of Clinton's flaws is that she is incredibly calculating-- she finally seemed human when she cried

Posted by: kmcnyasha | January 9, 2008 8:38 AM | Report abuse

Did emotional tears win New Hampshire for Hillary Clinton?


Posted by: usadblake | January 9, 2008 8:19 AM | Report abuse

Oh, stop being dopes! The polls were wrong because people lied. Period. People want to be perceived as being someone who will vote for change, for a fresh face, etc. When it comes to flipping the lever, they went for experience.

Posted by: RoboFlop | January 9, 2008 8:00 AM | Report abuse

The Iowa results don't refute the black-candidates-poll-lower thesis--they confirm it. Remember, there was no secret ballot in Iowa. It's certainly conceivable that lots of Iowans didn't want to appear racist in their public meeting places among their neighbors. And since, in NH, the Republican results tracked pretty well with polls, it seems unlikely that there was a switch from Obama to McCain. If so, it didn't show up in the numbers. I'm leaning toward the idea that voters in NH, at least to some extent, wanted the pollsters to believe they were willing to vote for a black man when they really weren't.

Posted by: Tewkesbury | January 9, 2008 7:52 AM | Report abuse

NH Voters were not fooled like Iowans and looked past obamas empty preaching and found NO SUBSTANCE! WE ALL LOVED HOW FOX AND CNN "the Best Political teams on TV" who had constantly promoted obama would win double digits, didn't have a clue what voters are thinking... Fire them all! The media has failed miserably by NOT doing their job reporting the facts instead reporting their personal biased opinions. They need to report the FACTS so that more educated voters like NH can make the educated decision and vote what best for the country and not what's best for shows ratings. They ALL have egg on their faces today! And the entire crew at FOX and CNN has lost credibility...GO HILLARY! Educated voters do not let bogus news tabloid TV make our decisions! CLINTON 2008

Posted by: dyck21005 | January 9, 2008 7:42 AM | Report abuse

The two poll questions I would like to but never will see are;How many probable dem voters will vote GOP if Hillary wins the primarys?How many blacks found it unforgivable for Bill and Hillary to use the gender card and slurs on a fellow Democrat who also happens to be the best black candidate in the history of our country.Donna B went OFF on CNN yesterday about the fairytale slur

Posted by: Playthetri | January 9, 2008 7:03 AM | Report abuse

With the greatest respect to Tassie Tiger and other posts, the proposition that being on top of the ballot paper in a voluntary ballot is worth 2-3% is a nonsense. In Australia voting is compulsory with fines for those who don't vote. Australia has exhaustive preferential voting i.e. you must place a number against each candidate in order of preference for the vote to be valid. This invites reluctant voters who are compelled to vote to simply vote down the ballot (or sometimes up the ballot paper) and is a so-called "donkey vote". It's hard to imagine this applying in a ballot where voters make a conscious decision to cast a ballot on a Winter's day in January as is the case in both Iowa and New Hampshire. It's also a feeble effort to explain the massive failure of the polls to accurately pick the outcome of the New Hampshire primary.

Posted by: ozrod2001 | January 9, 2008 6:56 AM | Report abuse

I'm more than a little surprised that pollsters and commentators didn't take age as a factor in understanding poll disparities. New Hampshire is a state whose 18 to 30 demographics has been dropping over the years. Obama beat poll expectations in Iowa do to the huge turnout of young, first time voters. Those voters just weren't there in New Hampshire in the same numbers. I think Obama holding Clinton's lead to 2 percentage points does show a surge for him, just no miracle.

With the proliferation of cell phones, and the energized youth vote, polls will continue to be wrong throughout the primaries. Maybe someone could suggest that talking heads spend more time on discussing candidates platforms and stop channeling Jimmy the Greek. Especially twitty bird.

Posted by: DonnaMariaInChicago | January 9, 2008 6:51 AM | Report abuse

Perhaps voters recalled the last time an inexperienced, but seemingly idealistic candidate - who promised change - actually won the nomination. That was Jimmy Carter - who turned out to be the second worst President ever (the incumbant takes the gold!). Hillary Clinton's experience and consistency in fighting for what she believes in, and her demonstrated ability as a Senator to work with a broad range of people to actually accomplish our goals is being recognized and appreciated.

Posted by: wenfu | January 9, 2008 6:26 AM | Report abuse

In Australia it has always been thought that top of a House of Representatives ballot paper is worth about 1-3 per cent to the lucky candidate, so I find the 3 per cent proposed in regard to the Clinton vote to be quite convincing. Australian ballot positions are now drawn by lot to lessen the importance of alphabetical order.

Tassie Tiger

Posted by: bbandsb | January 9, 2008 5:50 AM | Report abuse

If this was Russia, and Putin had been carried nonstop the day before crying to show his humanity, we'd be complaining the election was fixed. We have less democracy than we think we have. It was a good night for Republicans to see dynasty politics revived.

Posted by: paulnolan97 | January 9, 2008 5:35 AM | Report abuse

Coumaris wrote: "No one is mentioning that half the ballots in NH were cast before the Iowa results."

-- No one is mentioning this because it is not true.

Posted by: bongura | January 9, 2008 4:37 AM | Report abuse

Live free or Diebold.

That is the only comment this article needs.

Comma delimited database: NH municipalities hand count vs use Diebold machines:

Posted by: Rubiconski | January 9, 2008 4:37 AM | Report abuse

A young woman on PBS or C-Span noticed from the exit data that quite a few supporters of John Edwards abandoned him last night and voted for Clinton rather than Obama. Now that's a surprise.

Posted by: Daedalus | January 9, 2008 4:29 AM | Report abuse

They threw the Kitchen sink, then for good measure they piled the Bathtub and the Urinal on top of her back.

Then to top it off they went and wrote her Obit; there's just one tiny little problem here, here name isn't Terri Shiavo, this woman ain't brain dead, weary yea! who wouldn't be with what the MSM has put her through, and the grueling pace that has been this primary in Iowa and N.H.

The bottom line here is that the stacking by the so called Independents did'nt work nor did the so called young vote.

The Obit writers guild has a few new members that need to attend refresher courses on how to write Obituaries, the first lesson in writing them is to have a dead body that you can count on staying dead,


Posted by: nightslider | January 9, 2008 4:16 AM | Report abuse

Baloney. People will lie to pollsters when a black is a candidate. Many will say they support a black candidate, but will vote against the candidate in the booth. Same thing happened to Kennedy in 1960. Polls predicted a big win for him over Nixon, but when push came to shove, anti Catholic bias came out when the curtain was pulled. Same thing will happen to Obama.

Posted by: RFN8143 | January 9, 2008 3:55 AM | Report abuse

It is a simple matter of statistics that the smaller the group, the harder to sample: especially, diverse, ill defined groups where participants can cross party lines. But weirdest of all, in a nation of 250 million, what does it matter how 500,000 vote in a popularity contest? Talk about seeing the profound where there is none.

Posted by: rjduwors | January 9, 2008 3:08 AM | Report abuse

It could be that the Obama college kids that swarmed all over Iowa during winter break had gone back to class. Unlike the Iowa caucuses, out of state students can't vote in New Hampshire. That may explain why there were more than double the participants in the Iowa Democratic caucuses than in the Republican caucuses, in a state where registered Republicans outnumber registered Democrats 2 to 1.

Posted by: afgailafgail | January 9, 2008 2:47 AM | Report abuse

It could be that the Obama college kids that swarmed all over Iowa during winter break had gone back to class. Unlike the Iowa caucuses, out of state students can't vote in New Hampshire. That may explain why there were so many more participants in the Iowa Democratic caucuses than in the Republican causes in a state where registered Republicans outnumber registered Democrats 2 to 1.

Posted by: afgailafgail | January 9, 2008 2:39 AM | Report abuse

So if the alphabetical order accounts for 3% of the vote. Still, Obama was ahead double digits. If 50% of the NH votes were cast before Iowa results, the percentage each candidate was the projected winner, before or after Iowa, was a wash.

It is curious to note that the winning percentage is within the margin of error.

But heh yeah, good news for the rest of the 48.

Posted by: jr1123 | January 9, 2008 2:20 AM | Report abuse

This seems to be a polling debacle on par with the exit polling from 2000. It wasn't just one or two polls that were wrong. I think a previous commenter brings up a real interesting point, that of the overall percentage of absentee votes.

Posted by: armoura | January 9, 2008 2:15 AM | Report abuse

Why is it so difficult to simply say: The polls were erroneous?
One of the significant stories about the New Hampshire primary is that the pollsters were WRONG.

Posted by: josephakocon | January 9, 2008 2:07 AM | Report abuse

The so called "Bradley effect", named after the African American candidate for California governor has been pretty well studied.

Posted by: Trumbull | January 9, 2008 1:54 AM | Report abuse

No one is mentioning that half the ballots in NH were cast before the Iowa results.

Posted by: coumaris | January 9, 2008 1:47 AM | Report abuse

I think the exit polls will show that most of the 'live free or die' independents in New Hampshire like Obama, but they rose up and voted Republican to derail Huckabee's Evangelical Express before it really got rolling.

Posted by: rossasmith | January 9, 2008 1:30 AM | Report abuse

How solid is that data on the impact of ballot order? If what is written here is true, it's stunning! Three percent or more?!

In general elections does ballot order typically vary by precinct? Does order matter less if the candidates are from different parties?

Posted by: illinois2 | January 9, 2008 1:15 AM | Report abuse

The comments to this entry are closed.

RSS Feed
Subscribe to The Post

© 2010 The Washington Post Company