Why Hillary won when the polls said she would lose

Right, no point avoiding it. Here's what I said yesterday:

Obama is 8 points clear and rising. He is going to win New Hampshire, probably with a landslide.

Doesn't look so clever now, does it?

So why did this happen? This is one of the worst polling errors I've seen. The margins of error were, after all the polls were collected together, tiny.

There are basically six theories overnight, although later there may be more.

First, the crying. The idea is that women warmed to Hillary's emotional response to falling behind. This is emerging as the standard explanation - the reverse of Neil Kinnock's Sheffield rally. The Democrats are heavily female and in this primary the vote broke 43 to 57 male to female.

Pros: The theory could explain a late shift among undecided voters, particularly undecided females. The crying made Hillary appear more likeable. Cons: The shift was pretty big, the crying not so big an incident. It seems too convenient a packaged explanation.

Second, the Reese Witherspoon effect. This is framed by Jonathan Alter of Newsweek as follows:

As in any high-school election, the studious girls who show up to vote might harbour a few resentments about the boys. It's like the movie "Election," where Reese Witherspoon's character, Tracy Flick, is an ambitious and too-perfect high school senior who has the election stolen from her after she was expected to win against a cool if inexperienced jock. By the end of the movie, she ends up on top.

Pros: Great name for a theory and it is just about tenable that Obama's bounce might have been resented. Cons: Ridiculously convoluted link to an obscure movie and no reason to explain why the resentment didn't show up in the polls before polling day. Obama's momentum in opinion polls was quickening not slowing.

Third, the First Lady issue. Both Obama and John Edwards minimised her contribution as First Lady in an ABC debate. This might have been resented by women.

Pros: Helps explain huge differential support for Hillary among women. Cons: Wasn't really a new issue that might have produced a last minute shift against Obama.

Fourth, minor candidate squeeze. There was a fall off in the John Edwards and Bill Richardson vote which helped Hillary.

Pros: There was definitely a fall in their vote. Edwards poll average was 18.3 for instance and he polled 17 per cent. Cons: The fall-off wasn't big enough to explain the whole impact and there is no reason, either, to assume that the Edwards' vote would break for Hillary.

Fifth, the Bradley Effect. Black candidates often do better in polls than in elections as voters like to be seen to be supporting them even if they don't actually do so. This is known as the Bradley Effect after Tom Bradley, the African American LA Mayor, who lost out in the California Governor's race despite leading in the polls.

Pros: Would certainly explain Obama's underperformance and the contrast with Iowa where voting wasn't secret. Cons: The exit polls, which looked wrong initially, in the end were accurate. Wouldn't the Bradley Effect have shown up in these polls?

Sixth, the Spiral of Silence. Over the last decade and a half in British elections the Tories often did better than unadjusted polls said that they would. Why? The theory is that people were embarrassed to admit that they would support them partly because their neighbours disliked the Tories so strongly. Hillary produces a similar reaction and might therefore be subject to a spiral of silence. Pollsters have now begun to adjust their polls to take account of this.

Pros: Would explain the huge gap between her poll rating and the result. This might either have been there all along or might have appeared in the last couple of days simply because people were embarrassed to admit they weren't jumping on the Obama bandwagon. People didn't shift back from Obama according to this theory, they never shifted to him. They just told pollsters that they had. Cons: Wouldn't silence have been maintained in talking to exit pollsters? In Britain it was.

These theories are not mutually exclusive, of course. They might all be true.

But my own view? It's the Bradley Effect and the Spiral of Silence that were the biggest factors.

Comments

[adam]

Maybe the pollsters overestimated a 'spiral of silence' effect for Obama. And wouldn't there be a 'Bradley' effect for Hillary?

It's difficult to explain the error in terms of gender or colour when both candidates aren't white males.

Couldn't an alternative explanation be related to the unexpected result in Iowa? Could Obama's win have stirred apathetic Clintonites into voting?

And there's always the 'Big Brother' theory, where losers are kept in the competition to make it more interesting...

[Stephen]

Are you sure the exit polls were amended using their own data Danny? I thought the recalculation came when the first poll returns were factored in. If you look at the Republican exit polls by Fox 35-30 McCain Romney the prediction was pretty good compared with Fox's initial 39-34 for Obama-Clinton.

If so your analysis is spot on and Clinton looks to be the hot favourite again

[Anne Murphy]

Isn't it just possible that a significant number of voters instinctively and emotionally liked Obama and his vision speak, and so were potential voters when polled, but when they had to make a decision they voted with their heads rather than their hearts. That they thought hard about whether there was sufficient substance in what he was saying given the inconsistancies in his track record and lack of experience, and they decided they didn't want to take the risk? When I close my eyes and listen to Obama, for me it is like Blair in 1997 all over again. And we know how empty all that rhetoric turned out to be (though, like Obama, he was awfully good at it). Perhaps Americans are just more cautious than we were/are.

[James Schneider]

Stephen, but CNN and MSNBC had Hillary in front from the exit polls. I think the answer lies in the demographic break down of the votes (see the CNN or MSNBC exit poll break down).

[John Petley]

Well, the pollsters were wrong. The main thing is that this is a great pity. When Hillary Rodham Clinton finally makes her ignominous exit from the Presidential campaign, the world can heave a collective sigh of relief. The sooner the better.

[Saurabh Sircar]

Bradley Effect makes perfect sense to me.

It worked against Obama because it would be politically incorrect to express that one wasn't considering the merits of the candidate because of his race and so, the pollsters got wrong data.

Adam says "wouldn't there be Bradley Effect for Hillary". The answer is no because it doesn't make one look politically incorrect if something adverse was said about a white candidate; the opinions are expressed more freely without worrying about political correctness.

[Linda]

You can put up all sorts of arguments as to how and why, but at the end of the day the bottom line has to be that the human being is fickle, and always has the chance to change their mind at the last moment.

[Quijote]

Problem is there is a war on, and a lot of mess to clean up. Can't give the job to somebody inexperienced, who didn't understand why the mess was necessary in the first place. And then, frankly, on top of that there is the Bradley Effect.

[David Moss]

Mark Steyn has a financial Bradley effect theory* -- if NH votes the same way as Iowa, the importance of Iowa will increase, it will attract more campaign money, and NH less. You can't say that to a pollster, but that's how you cast your ballot.

----------
* http://corner.nationalreview.com/post/?q=NTFiYjFmYmY0NDc1NGU1MTdmZDE1YTRhOTBmOWVhYWE=

[Kurt]

A popular explanation on this side of the Atlantic has to do with the mechanics of the New Hampshire primary. Voters who are registered Independent may vote on either the Democrat or the Republican slate. Obama was favored in this segment, but many of them, thinking Obama had it in the bag, may have opted to vote in the Republican contest--leaving the field to the registered Democrates, among whom Clinton has more support. I don't know to what extent this can be verified.

[Pennsylvania Trail]

Perhaps it is as simple as the fact that the new electronic voting machines provided forced (unverifiable) error. Note that the exit polls suggested 39-34 or more Obama and those rural districts having ballots actually counted by hand (in other words, verifiable) showed similar results, if not a larger win for Obama.

The electronic voting machines in the cities which represented more than half of the districts were riddled with errors, contrived or inadvertent, which could help explain the results.

For example, if someone had simply made a simple software error and those votes which were actually going to Obama were tallied for Clinton and vice versa (not hard to do), then this error would explain both the failure of the total vote to reflect the poll-determined outcome and would also explain why the exit polls and the ballots counted by hand (with no chance for error) were much closer to the actual polling data.

This possibility was not lost on Obama as he obliquely referenced "Chicago-style" politics in a speech today. The only difference is that the Democratic establishment in Cook county counts the dead, whereas in New Hampshire the Democratic establishment appears to jury-rig the software.

[Jerry]

Let's be a bit thoughtful about the Bradley effect please. This is more lazy thinking, and more lazy thinking piled onto yesterday's mistakes explains why it happened in the first place. If we had a Bradley effect, that must mean that working class New Hampshire women are more bigoted than New Hampshire men. Clinton won a traditional Democratic cohort in the secular northeast: working class women, and women 45+. Alternatively, this must mean that registered Democrats harbor more bigoted options than Independents, who went strongly for Obama. It's not complicated, Clinton's ground game was relentless and got out her base who were motivated to vote for her (for reasons you have listed). Obama does not yet have a base. I live in NH, and many in her key demographic were called 6 times yesterday. So, let's be thoughtful, and not compound yesterday's error with a bogus hypothesis that is inflammatory, and ill informed.

[Jan]

re: "First, the crying"

Well, first, the fact that you use the word "crying" indicates immediately that you don't get it. "Crying" involves actually crying; Clinton never cried.

The media showed one clip, over and over and over and over, where Clinton expressed her love for America, and the media tried to use the clip to ridicule her. Not a good idea; women hate bullies.

Also, as a Clinton supporter, I've heard Clinton Haters tell me that they don't know anyone who supports Clinton. I encourage them to reveal in their next political encounter that they are "for Clinton" and see if the dialogue changes. IT ALWAYS DOES.

So, I encourage all Clinton Haters to do the same thing, and witness for yourselves "the spiral of silence" that encompasses Senator Clinton, whether YOU personally like her or not.

[JIM HEAPHY]

How about the Independent Effect? New Hampshire independents could vote in either the Democratic or the Republican contest. They tended to support either Obama or McCain. With the polls and pundits predicting an Obama landslide, maybe the independents broke for McCain, who was perceived as needing their help more.

[HoosierDaddy]

It seems like we are talking about Obama staying right on or only slightly below his poll numbers and Hillary picking up several percentage points. Could it be that Hillary benefitted from a massive shift of the undecideds toward her? It might be explained by the undecided people tipping toward a known quantity as opposed to a newcomer. Has crossed the numbers to see if she pulled heavily from Edwards and Richardson at the last minute? Those people had to be losing enthusiasm in their original candidates as well.

[Andrew Milner]

Diebold saved her ass?

[Tim]

I don't think it's the "crying" itself (actually just a slight waver of the voice) but how the media played it up and tryed to make a mountain out of less than a molehill. Women didn't like that.

I think the polls were right, but at the moment in time (actually range of time) they were taken, and it changed in the 36 hours or so after that. Some pollsters said there was a shift from the early collecting period toward the end.

[Porco Rosso]

The even more bizarre thing about the polls is that they were virtually spot on for all other candidates making the Hilary / Obama turn around even more puzzling.

Hillary somehow managed to instigate a last minute near 20 point swing to defeat Obama by three per cent.

Of course the proven 'easily hackable' Diebold voting machines were used in these elections and their continued use seriously compromises any results.

Expect more of the inexplicable and bizarre.

[Cllr Ken Tiwari]

Why should media be correct at all time, media hype is acceptable @time
but not all the, because the opinion
media collect are home made, medias
own product, most of the time poll
may be right, but always,
Cllr Ken Tiwari(oxford UK)

[Cllr Ken Tiwari]

Once again, all pollsters are the
guesser, why the media or any one in
the world, will think just because,
you wone one state, you could win the rest, if you understand UK's
electorates, you will understand the
USA i.e. in UK-south vote different-
to that north of UK!Cllr Ken Tiwari
(Oxford UK)

[Aaron]

As much as I wish that it were the Bradley effect -- indicating racism among Democrats -- the polls predicted that Obama would take ~37%. He ended up taking ~37%. Edwards got 2% less than the polls said he would, and I haven't seen polling data for the other candidates (~14% total), but Hillary took all of her new support, an increase of 9% from them and none from Obama.

This also dispenses with the "Independent Effect" that Jan suggested (to which I was initially friendly) that all the independent voters who would have voted for Obama assumed he was going to win anyway so switched to McCain. Obama's numbers were spot on. He didn't lose support. Hillary gained it from Edwards and the also-rans.

I'm personally sympathetic to the theory that women were suckered in by Hillary's pathetic whining, and that people were generally resentful of the media telling them that the race was already decided before any votes were cast.

[Tony]

WAKE UP.

http://www.youtube.com/watch?v=gocC_456PzI&feature=related

NH was stolen.

How soon we forget Ohio and Florida.

"It would flip the vote"

Look at the hand counted votes vs. the Diebold opti-scan results. Curious, huh?

"They'd never see it"

last note: why does these voting irregularites ALWAYS benefit the status quo candidate???

STOP BLAMING THE POLLS! Does some investigative work, so called journalists. Follow this lead:

LHS Associates
John Silvestro

Follow the connections, see what comes up...

[hillary clinton monster]

Hillary’s negative attacks on Obama are OUTRAGOUS and just goes to show how she will do anything and say anything to get elected no matter who she tramples in the process. I have never seen attacks so vicious against a fellow Democrat. She is an Obnoxious Braggart and the sound of her voice is irritating. If you listen carefully to her speeches you will find that she is constantly saying “I” did this and “I” did that. Everything she says sounds rehearsed. She sounds like an old school politician and like someone who is part of the past.

Hillary talks about healthcare like it’s the NUMBER ONE ISSUE of importance to most Americans. IT ISN"T and she doesn’t GET IT. There are far more important issues which Barack Obama has so eloquently articulated during the debates and Hillary hasn’t.

We need radical change after having an incompetent idiot for president for the past 8 years. As Barack Obama has stated in his speeches, we need to take our government back from special interest groups and lobbyist. He wants to make our government for the people, of the people and by the people as our founding fathers intended.

Hillary supporters are quick to point out how she won Ohio and Texas. The fact is… she used scare tactics, negative ads and dirty Republican style politics to help obtain those votes. Let us not forget that the majority of voters in Ohio and Texas voted for BUSH so apparently a lot of them are easy to fool.

More than 20 people who were associated with the Clintons, including Vincent Foster died mysteriously. Do you think it's just a coincidence? Clintonmemoriallibrary.com has lots of info about the deaths, crime connections and lies to the American people.

I TOTALLY agree with Samantha Power… Hillary Clinton is a Monster.

[malshi]

I took a big look for the things which are happening between Obama and Hillary.This shows that the people are always waiting for a new unique play of Obama.
...........
Malshi
http://www.SelectWealthSystem.com/?t=wc