January 9, 2008

Why were the polls so wrong? Did people lie about voting for Obama?

Was it "the Bradley effect"?

ADDED: Mickey Kaus considers the Bradley effect + 3 other theories:
2. Lazio Effect. No ganging up on the girl! First, Edwards turns on her in the debate. Then Obama says she's merely "likeable enough." Then the press disparages her anger, mocks her campaign and gloats over its troubles. They made her cry! And then that mean macho John Edwards goes and says the crying makes her unfit to be president...
And she was also mocked — I heard it on Rush Limbaugh — for trying to get the Lazio effect going. Which would be a veritable Lazio vortex.
3. Feiler/Skurnik Effect: ... The familiar Feiler Faster Thesis holds that voters are comfortable processing information at the vastly increased speed it can come at them. Jerry Skurnik's "Two Electorate" theory holds that voters who don't follow politics are much less informed than they used to be, which causes polls to shift rapidly when they do inform themselves. Put these two together...

The Feiler/Skurnik Effect magnifies the significance of any events that occur in the final day or two of the campaign.
If this theory is true — or even if the campaigns think it's true — look out! (Good for blogging, though.)
4) The Congestion Alert Effect: I remember when the Southern California transportation authorities installed a state-of-the-art series of electronic signs alongside the freeways to give motorists instantaneous warnings of traffic delays. The signs don't do that any more. Why? It turned out that when you warned drivers of congestion on Route A, they all took Route B, leading the latter to become congested instead of the former. Similarly, independent voters in N.H. were told by the press that the Democratic race was a done deal--so they voted in the closer, more exciting Republican race....
A primary is not an election, so this kind of funny business is always part of the mix. I remember when my father voted for McGovern. He was damned pleased about it and assured me that's what everyone was doing. And here's the final electoral map for 1972. I was crushed. How dishonest!


Ben (The Tiger in Exile) said...


It was the women, not the whites. (Well, white women, I guess.) Gender, not race.

Peter V. Bella said...

No one took into account the heigher turnout. According to the pundits, the polling was based on the 2004 turnout.

MDIJim said...

Indeed. Mickey Kaus argues that Obama benefited from a reverse Bradley Effect in Iowa in that when people in a caucus had to publicly declare their vote in front of neighbors, all of them Democrats, many of them decided to show off their enlightenment by declaring for the African-American candidate.

A more prosaic explanation for NH is that the polls came before Hillary's public display of emotion. It turned me (a male) off, but it seems to have swung women from Obama to Hillary. You could call it the Gloria Steinem effect.

Phobos5 said...

This probably can never be proven empirically but the history of this phenomenon suggests that it played a role. While there was certainly a late shift in sentiment, perhaps largely among women, New Hampshire is one of the most conservative Eastern states. Obama did well in places like Portsmouth, which has a significant number of new age type voters. Much of the rest of the state, however, has what is pejoratively characterized as a "red neck" flavor. If the "Bradley" or "Wilder" factor played a role, that is distressing and, I'm afraid, likely to continue to do so.

Mortimer Brezny said...

You could call it the Gloria Steinem effect.

Exactly. I think Maureen Dowd answers this question in today's New York Times.

KCFleming said...

The answer to "Why were the polls so wrong?" is all of the above.

Polls are modern-day tarot cards, trying to divine the future. Merely because there are numbers involved, especially decimals, it wears the vestments of science.

Polls are palm readings for the political class, seances for the sceptered ...What are they thinking? What will they do?

Roger J. said...

"Polls are palm readings for the political class, seances for the sceptered ..." Wow, Pogo--very alliterative! you are on a rhetorical roll today!

George M. Spencer said...

It was the Billagain Syndrome.

The more Democratic primary voters see Bill Clinton in person and on TV the more they are going to get nostalgic for the good ol' days.

He has been away for a while, and we must admit that the wily ol' rascal is a charmer.

They gonna eat him up like hot buttered biscuits down in South Carolina.

Sloanasaurus said...

It always seems that the politicians who "energize" the young voters seem to fall on their face in the end. Perhaps its because only a fraction of the young voters are actually energized enough to vote.

In Iowa, it takes a lot of commitment to attend a caucus and vote in the caucus. If you are busy and politics is less important than something else you will not attend the caucus... however, you will take 15 min to vote in a primary.... Maybe Hillary is getting all the votes from the people who view politics as down on the ladder and perhaps don't want change.

Mortimer Brezny said...

“When we’ve been told we’re not ready or we shouldn’t try or we can’t, generations of Americans have responded with a simple creed that sums up the spirit of a people,” Mr. Obama said. “Yes, we can. Yes, we can.” Throughout the evening, the confidence of Mr. Obama’s campaign gradually fell as returns poured in from across the state, which never put him over Mrs. Clinton. Aides said they believe that women rallied behind Mrs. Clinton in the final hours of the race, when news coverage was dominated by accounts of her nearly breaking into tears as she answered a voter’s question.

Roger J. said...

Re Ann's question, two words: Rove Diebold

Peter V. Bella said...

Pogo said
Polls are palm readings for the political class, seances for the sceptered ...What are they thinking? What will they do?


Tank said...

One reason, easily forgotten, is that they let independents vote. I still can't figure that out.

If the purpose of the primaries is for the parties to each pick their own candidates, shouldn't you have to be registered in one part or the other to vote in the primary? Why do Non-Repub's or Non-Dem's get to vote for the parties' candidates in the primaries?

I think that adds greatly to the unpredictability factor. It also helps to explain why McCain is so strong in NH.

KCFleming said...

It's her Party and she'll cry if she wants to.

You would cry too if it happened to you.

KCFleming said...

Since we're replaying the 90s, 1992's "The Crying Game" seems appropriate. Except, one hopes, the surprise.

I know all there is to know about the crying game
Ive had my share of the crying game

First there are kisses, then there are sighs
And then before you know where you are
Youre sayin goodbye

Mortimer Brezny said...

Perfect analysis.

Fritz said...

It's obvious, the Clintons stole the election. If this were Bush, the polling wouldn't be suspect, the voting would.

Hillary's tears were the difference. Liberalism comes down to emotion. Even my wife who doesn't like Hillary felt sorry for her.

Bissage said...

Mort's 9:02 has made me feel the prick of conscience and I need to confess.

I detested Reagan until he got shot.

[Wipes sad little tear from eye.]

* sniff *

Richard Dolan said...

Jay Cost points out, based on the exit polls, that Hillary's core support was more downscale, union household type Dems. There is always the possibility that the polls were correct in measuring general support, but were badly skewed on who was a "likely voter." The basic story might be Hillary's superior ability to get more of her core supporters to the polls than the other guy. The unions and "machine Dems" were once very good at that.

Original Mike said...

If this were the first time polls were wrong, it would be an interesting question. Given the historical record, I think a better questions is, "Why are people fixated on polls?"

PatCA said...

I think Pogo has it--it's a lot of things, all of them beyond the grasp of the puny pollster.

Even the Lutz guy said there are so many independents and new voters that you might as well throw out the old formulas.

Also, they know everybody is watching them, so people sort of enjoyed confounding the pollsters.

Smilin' Jack said...

It's a Pyrrhic victory for Hill. People thought she was a soulless zombie who wanted to eat their brains, but when she cried, they thought she might be human after all, so they voted for her out of pity. So then in NH she came back to life. Just like a zombie, people are thinking now, she's not going to fool me again. She won't win another primary. People want to keep their brains.

PatCA said...

I think Pogo has it--it's a lot of things, all of them beyond the grasp of the puny pollster.

Even the Lutz guy said there are so many independents and new voters that you might as well throw out the old formulas.

Also, they know everybody is watching them, so people sort of enjoyed confounding the pollsters.

An Edjamikated Redneck said...

This could be an interesting General Election in November if the polling/results issue holds through all of the primaries for the Dems.

Does it mean that although lip service is given to the female or black candidate in public, when th time comes to pull the lever will voters default to the white male?

Exit polls in the general could indicate Hillary or Obama, but the votes may go to the white male.

This will not be proof of the racist or sexist tendancies of the voters, but proof the nasty Republicans have stolen another election.

Unknown said...

Ann, I'm missing some context here. How were your father's assurances "dishonest"? Wrong, yes. Are you saying he knew he was wrong but was saying otherwise to get you to vote for him?

Anonymous said...

I agree with phobo5. New Hampshire was never onboard the Obama peace train to begin with, and not before, during, or after the Iowa caucuses. Hillary was always going to win there.

What may never be known or proven is the extent to which the Clintons knew full well that she always had it in the bag, but still somehow managed to juke the pollsters into dancing with Obama, thereby allowing Hillary to do the comeback kid thing, the specter of 1992 that she's really been itching to recreate.

Randy said...

1. At least one, perhaps two respectable polling firms had very close numbers right up to the end.

2. Everyone seems to be ignoring the margin of error for polling, which in the case of NH polls was about 4%. Even in the 10-point blow-out scenarios, this means that it was potentially a tight race.

3. 40% of the votes came from independents and Obama failed to carry them as heavily as he did in Iowa. Maybe more independents chose the Dem primary at the last minute.

4. I said this before and I'll say it again: his performance in the Saturday night debate was underwhelming.

5. As middle class guy has pointed out, if the polling model was flawed, the polling results would look askew. I'm not so sure about this as Clinton cleaned Obama's clock among registered Dems.

6. If you look at the election results, Clinton won most of the bigger counties by good margins. Obama took the college towns by good margins. A lot of smaller counties went heavily for Obama. So much for the "redneck" idea.

and finally:

7. The Bradley effect is utter BS. Bradley won among those actually voting on election day, just as the polling data showed. What none of the polls took into account was that, for the first time in history, anyone who wanted to could vote absentee just by asking, so the state GOP mailed absentee ballot applications to every registered Republican in the state. More than 500,000 new absentee voters voted for Deukmejian so when ALL of the votes were counted, Bradley lost the election. After that election, both parties began actively soliciting absentees (they now make up about 20% of all votes cast) and the difference is negligible. But at that time the difference was decisive. (NB: one of my near & dear designed that particular campaign program for Deukmejian.)

former law student said...

The late Chicago columnist Mike Royko advised voters to lie to exit pollsters, to prevent networks from projecting a winner before the polls close -- why would be believe exit polls?

Jim Hu said...

Any explanation has to account for the observation made by one of Yglesias' commenters that Obama actually got the percentage of the vote expected from the pre-primary polls. I ponder that here. Key point: this is inconsistent with either the Bradley effect or the hypothesis that Obama independents got overconfident and crossed over for McCain.

MadisonMan said...

"Why are people fixated on polls?"

A better question: Why are News Organizations fixated on polls? They report the polling results as if they are facts -- I suppose in some way they are -- when in reality polls are in large part fiction. Then the echo chamber on either side picks up the poll "News" -- if they find the result amenable -- and runs with it, trying to turn a poll result into a wishcast.

If I were a politician, I would have a list of noted polling failures, and ask any interrogator who framed a question with poll results why they considered the poll results to be a good predictor when they've been so notably wrong in the very near past. It strikes me that any journalist who uses poll results to frame a question is being a little lazy -- too lazy to go out and dig for facts themselves to support their arguments.

Along the same lines, why should anyone now believe the commentariat members who yesterday were chanting about the demise of Hillary!?

John Stodder said...

(Crossposted from a later thread, but seems more relevant here)

I don't know if Tom Bradley was actually a victim of the "Bradley effect."

It's based on the California 1982 gubernatorial race pitting the LA Mayor (and my old boss) Tom Bradley against Attorney General George Deukmejian.

The polls going into the election did show Bradley with a small lead, and in the end he lost narrowly. How often has that happened when the candidates were of the same race? Many times of course.

In '82, there were two other critical factors that get forgotten.

1) This election was the first time where a major effort to get voters to vote absentee was tried by either party. Before '82, the only people who asked for absentee ballots were people who really couldn't get to the polls due to disability or traveling out of state. It was a small factor. But that year, the GOP realized it could increase its base by making sure its voters got absentee ballots and were encouraged to use them. The polling models didn't account for this.

2) A liberal group put a gun control measure on the ballot that fall. It was a stupid move, obviously. It brought out a lot of rural voters who normally didn't vote -- the kind of voters not generally counted as "likely voters" by pollsters -- to vote against the measure. These were not voters who were going to support Bradley.

I'm not denying the effect pollsters talk about might exist. But I don't think historians believe it's responsible for all the variance between the polls and results. The absentee and gun control factors really threw off the models.

Absentees were partly responsible for Obama's loss, too. What percentage of NH voters had already voted before Obama's inspiring Iowa victory? How many of them might've changed their votes, caught up in the Obama momentum?

Original Mike said...

"A better question: Why are News Organizations fixated on polls?"

You're right. And the answer probably is that it requires no effort to report and discuss the latest polling results. It's even easier than reporting on what Brittany did today.

KCFleming said...

The disturbing thing to me is that polls and the race itself has long superseded what the race was about.

Rather than some sober discussion of the issues at hand, the elections have, by way of the infotainment imperative, devolved into a mere high school student council election.

Yes, I know elections in the US have always involved the theatrical and emotive, but in the last 30 years it has become only that, and nothing else. No longer does it matter that serious questions go unanswered.

Now, they go unasked altogether.

Soon we will vote based on the best posters and music.

Revenant said...

Both Obama and Edwards received the same percentage of the vote (within the margin of error) that polls had indicated they would. Hillary, on the other hand, did much better than her poll numbers. What this suggests is that undecideds went overwhelmingly for Hillary.

So the question, it seems to me, isn't "why were the polls so wrong" but "why did Hillary pick up most of the undecided voters".

MadisonMan said...

Revenant, you see the polling data and think most were right. I see them and think some were wrong, and therefore the correct polls were coincidence.

I wonder which is a more valid viewpoint.

Revenant said...


My gut feeling is that the undecideds really DID go overwhelmingly for Hillary, whether because of her debate performance (she was the only one who seemed ready for prime time) or as a reaction to her defeat in Iowa. But who knows; that's just idle speculation on my part.

IgnatzEsq said...

For the more geeky among us (I have a degree in mathematics) this site is pure poll analysis gold. And it's worth a read before you jump to any conclusions of what went wrong.

IgnatzEsq said...


Mortimer Brezny said...

So the question, it seems to me, isn't "why were the polls so wrong" but "why did Hillary pick up most of the undecided voters".

Because Biden and Dodd didn't endorse anybody.

Trooper York said...

I never listen to the poles. Now the Romanians usually have a handle on the situation. And they know which candidate is a werewolf. (My money is on Mitt. As a werewolf of course).

Pastafarian said...

Maybe people lied just to lie. The late columnist Mike Royko always said you should lie to pollsters. I lie to 'em every time. Just 'cuz.

Pastafarian said...

Sorry "former law student". I didn't see your comment.

DADvocate said...

New Hampshire is only 1.1% black. With such a small black vote to counter act it I'd say the Bradley effect, i.e. racism is a strong possibility.

Joe said...

Why not go with the simplest explanation; independent voters were trying to game the system. Individual voters thought Obama had it made, so decided to use their vote to more effect by voting against Mitt Romney and for McCain.

Trooper York said...

The only person who ever got screwed by the Bradley effect was
Cazzie Russell.

NotWhoIUsedtoBe said...

Higher turnout with more lower class and less educated voters equals more Clinton votes.

Higher turnout helps Clinton.