Zogby: I Did Have Hillary Rising In NH — But I Couldn't Tell You
Amidst all the hand-wringing by pollsters over how they managed to get New Hampshire so amazingly wrong, John Zogby has put out a press release containing an interesting piece of information: The last one-day sample of his three-day tracking poll showed Hillary closing the gap in a big way -- but the sample was too small to be published on its own:
My polling showed Clinton doing well on the late Sunday night and all day Monday – she was in a 2-point race in that portion of the polling. But since our methods call for a three-day rolling average, we had to legitimately factor the huge Obama numbers on Friday and Saturday – thus his 12 point average lead. Unfortunately, one day or a day–and–a–half does not make a trend and we ran out of time.
Okay, so Zogby's got his story, and he's sticking to it. But what's the excuse for all the other polls that were using one or two-day samples? That list includes American Research Group, Suffolk and Marist — and they got it wrong, too.
Comments (45)
I for the Dem wrote on January 9, 2008 1:28 PM:I'm fine with Obama and HRC, but let me tell you this...Obama lost because he got cocky and was pushing movement rather than candidacy; that's what hurt him. Period. "You're likeable enough, Hillary"...not even a smile. Nail in coffin.
Keith wrote on January 9, 2008 1:29 PM:Didn't he note that Obama was getting a bump on the Friday (even though it wasn't showing up in the data yet)? He should have noted it on Monday; it would make him and his polls more credible.
Steve in CA wrote on January 9, 2008 1:33 PM:I smell b.s. on this one. I could have sworn that previous Zogby polls that were rolling tracking polls had indications in the commentary about one-day jumps and movement.
I think this is CYA, and not particularly good CYA, at that.
Steve in SC wrote on January 9, 2008 1:39 PM:Oh dear, some more nonsense from America's fallen pollster John Zogby. Before 2004, Zogby was thought of very highly. But his forecasts have been pretty from 2004 on. I don't think he even knows how to poll people anymore.
audit the polls wrote on January 9, 2008 1:40 PM:It's simple. Polls lie. Ignore them.
Michael A wrote on January 9, 2008 1:43 PM:At this point, I would suggest that all the pollsters use their polls for the only thing that they are good for TP. Let the american people vote and have their say, making one candidate or the other "far" ahead in the polls only decreases the motivation for people to vote for the "leading" candidate. No more polls until people vote PERIOD.
drubs wrote on January 9, 2008 1:49 PM:Perhaps Iowa bounce was way overstated. All the polls I saw nightly leading up to the race had HRC up but falling with BO closing the gap. If you look at the polls (I go to angus-reid.com)up until last Wed, then last nights results aren't so shocking.
Also: Iowa is not a good predictor unless you are an incubant.
One other thought, if they are going to continue doing polls, why not do a poll of people who didn't vote and find out why they didn't. I bet a lunch alot of people didn't vote because they thought obama was running away with it, more than enough to beat clinton. What a shame.
Pepp wrote on January 9, 2008 1:51 PM:Outrageous Tripe. No difference than the Huff Post showing a voter denying Senator Clintons hand but not mentioning let alone having photos of, the abortion demonstration that forced Obama off the stage. Look the good thing is the American public can actually identify the political handlers trying to herd them and coronate their choice they were not even subtle blinded by what bias, maybe.
JO wrote on January 9, 2008 1:56 PM:How many of those "independents" voting for HRC did so in order to thwart Obama, in hopes that doing so might cause Democrats to nominate their weakest candidate to run in Nov.?
Steven T Van Haren wrote on January 9, 2008 2:02 PM:Non sequitr here:
ABC printed a opinion piece from a Stanford social scientist, Prof. Jon Krosnick, which points out that in previous New Hampshire elections, when a well known candidate's name appears on the top of the ballot, that candidate garners 3% more of the vote than the other candidates on the ballot. 3% was, of course, the amount by which Hilary Clinton won New Hampshire. Krosnick attributes Clinton's win to the fact that, for the first time in recent memory, a well known candidates name was at the top of every Democratic ballot in the state.
What seems lost in this analysis is the fact that this top of the list phenomenon would only garner Clinton 3% more votes than she was expected to get. In fact, she got about 8% more votes than she was expected to get. The 3% phenomenon may account for some of her popularity among New Hampshire voters, but it is impossible that it put her over the top, if the initial polls are to be given any credence at all.
I think that people are confused about polls here. Using polls to divine the preferences of a population on a single day is reasonably straightforward. Using polls to predict how subsets of that population will behave after several days have elapsed can be more difficult. When blowhards like Chris Matthews prattle on about the polls being wrong he's being an idiot. Just because HRC won on election day doesn't mean that Obama didn't have a substantial lead only days before.
good point, rk.
anonymousXYSOB wrote on January 9, 2008 2:07 PM:If one requires a 3-day rolling average, why is one even bothering to poll on day 2 and day 1 before the vote?
And if one finds it sufficiently worthwhile to poll on day 2 and day 1 before the vote, then why not publish the polling data with the appropriate caveats?
Lame.
kjoe wrote on January 9, 2008 2:14 PM:It was the exit polls i have to wonder about. right off the bat, msnbc said their exit polling showed a five point obama victory---too close to call.ok.
The other polls---don't get me started about women. Waitresses will hear about it, and----I will be punished.
hardtimezone wrote on January 9, 2008 2:20 PM:The voters are not a collective!
This is so pathetic!
Do you think all the voters even watched the debate. Do you think they all watched the stupid moment w/ Chris Matthews?
If that was true then the campaigns would put everything on tv.
The truth is that Obama won in Iowa w/ troops on the ground. Knocking on doors, calling people, etc., etc.
Don't tell me about Hillary's "human" moment. This is bull!
If these senerios were true then cable news would have off the chart ratings. They don't compared to the likes of Desperate Housewives!
You scum!
Wake up!
Matthews, crying, Bill, etc., etc.
IT DOESN'T ADD UP!
Obama didn't lose it, the machines took it.
The handcounted counties, on average, mirrored the pre-election polls, while the machine counted counties (only for Obama and Clinton on the dem side), on average, were way off the mark.
The pre-election polls CAN'T be that off unless someone can give us proof that the residents of N.H. are a collective.
WAKE UP!!!!!!!!!!!
hardtimezone wrote on January 9, 2008 2:21 PM:The voters are not a collective!
This is so pathetic!
Do you think all the voters even watched the debate. Do you think they all watched the stupid moment w/ Chris Matthews?
If that was true then the campaigns would put everything on tv.
The truth is that Obama won in Iowa w/ troops on the ground. Knocking on doors, calling people, etc., etc.
Don't tell me about Hillary's "human" moment. This is bull!
If these senerios were true then cable news would have off the chart ratings. They don't compared to the likes of Desperate Housewives!
You scum!
Wake up!
Matthews, crying, Bill, etc., etc.
IT DOESN'T ADD UP!
Obama didn't lose it, the machines took it.
The handcounted counties, on average, mirrored the pre-election polls, while the machine counted counties (only for Obama and Clinton on the dem side), on average, were way off the mark.
The pre-election polls CAN'T be that off unless someone can give us proof that the residents of N.H. are a collective.
WAKE UP!!!!!!!!!!!
drubs wrote on January 9, 2008 2:23 PM:Oh a conspiracy, now I understand.
daw wrote on January 9, 2008 2:24 PM:Another post got it exactly right: Obama got cocky; when he said "you're likeable enough, Hillary" and couldn't muster even a smile, undecideds started tuning him out. And that trend was reinforced by Edwards' nastisness about her emotional moment as well as by the media anti-hillary pile on re: the 'misty-eyed' moment.
Peter wrote on January 9, 2008 2:24 PM:maybe i'm stupid, but i don't understand this. are all polls worthless? are they only valid up until three days before a poll? or 4.5 days? do we need to know who's rolling where and for how long before we know exactly how INvalid a poll is?
redonk.
like Francis Fukuyama - "I knew it was bunk, but, like, i needed the money!"
Jinchi wrote on January 9, 2008 2:24 PM:Just because HRC won on election day doesn't mean that Obama didn't have a substantial lead only days before.
This is circular logic. The polls got it wrong. There's no evidence Obama was ever surging, except for the very polls that also failed to spot Hillary rallying back.
Polls are being used to predict the outcomes of a political race. If they can't predict accurately more than a day in advance, what's the point in even talking about them.
I'm always amazed at how many people are willing to believe that polls are excellent measures of actual voting preferences, despite plenty of evidence to the contrary.
In addition to his arrogance coming through with the likable enough, the no comment on his crowd's boos for Hillary was again another blow at his hope and change campaign, when you're campaigning on this broad, vague platform and then your actions start to vary from your message, you too lose credibility. I'm willing to ride this Obama wave, but not so naive to consider him unique than any other pol playing politics.
David Lasagna wrote on January 9, 2008 2:32 PM:Please look over at Bradblog and BlackBoxVoting for discussions of the discrepancies. They are examining aspects you are ignoring. Please open the discussion up to the possiblity of election fraud. ALL possiblities need to be examined. Fraud is always given short shrift despite the overwhelming evidence of it the last six years. Please check out the new documentary-Uncounted--The New Math of American Elections.
sincerely,
David Lasagna
As one NH resident pointed out elsewhere, the pre-election polls are looking at _likely_ voters. When you have a lot of _new and unlikely_ voters show up, you can - and should - expect your polls to be inaccurate.
Mary wrote on January 9, 2008 3:12 PM:John Edwards cooked his own goose.
His willful, calculated choice to "go after" Hillary at the Saturday debate, thereby eliminating her and then hoping Obama would trip up down the road, leaving him the lead, was a DISASTER.
Besides, the very women who might have supported him---low income, needing government help----switched en masse to Hillary at the last minute. Most probably because of his trash talk at the debate and afterwards.
He only got 17% of the Democratic vote.
Time for him to give it up and throw to one or the other candidates left.
He's toast.
gcs wrote on January 9, 2008 4:00 PM:I, for one, am goddamned sick and tired of listening to media douchebags tell us what's going to happen instead of reporting what did happen.
In fact, Josh Marshall happens to be one of the worst offenders in my opinion. For days all he did was report stories about Hillary breaking down, running out of money, basically helping to anoint Obama while bitch-slapping Chris Matthews (petty and weak Josh, petty and weak).
Then this morning Josh says he'd rather not live in a world without the fake foreknowledge that exit polls offer, so the networks can project a winner before half the damned country has voted?
The alternative? How about an election where the voters make up their minds by listening to the candidates? Huh? How about that? How about the media get the fuck out of the way and stop telling us who's "inevitable" and who's "done" huh, how about that? How about each and every one of us decide for ourselves who the best candidate rather than the teevee bobbleheads try and make up our minds for us by telling us we're wasting our vote, or who is going to win.
Because let's face it folks, the media bears a lot of the blame for Bush winning in 2000 (remember the free ride he got, while they obsessed over Gore's sighs?). They're also to blame for Iraq (thanks Judith Miller) and Bush's so-called reelection.
Do us all a favor, wouldja? Go back to reporting WHAT HAPPENED, you know the facts. I'm sure it's all right there in your journalism textbook from freshman year. And let the American people decide what's going to happen.
We'll all be a whole lot better off.
Char wrote on January 9, 2008 4:09 PM:Edwards comments not only hurt him, they also hurt Obama. His comments gave the impression that the two men had an agreement to gang up on Hillary. I hope Edwards didn't plan for that outcome.
Bill Collins wrote on January 9, 2008 4:24 PM:Hello folks:
Hillary's victory last night in New Hampshire leaves her in a very strong position going forward.
I'm not a fan of hers, but will of course vote for her in the fall if she wins the Dems' nod. As it stands now, we have a one-on-one race between a white woman and a Black man for the Dems' nomination. Just as was the case in 1988 when the nomination settled into a Dukakis-vs.-Jackson contest, Hillary is now in a strong position to win as the "white hope."
Hillary is now the beneficiary of both the latent and blatant racial fears of white Democrats. Today, post-New Hampshire, that makes her a sure winner. Hopefully hubby Bill understands this (in 1992, he went South after losing New Hampshire, but did not face a strong Black candidate in the race, and that's one of the best reasons that he won the nomination), so hopefully the Clintons won't inflame racial fears by baiting Obama unnecessarily.
If the Clintons fail to understand this, anti-Obama barbs coming from their camp have the potential to turn the Clinton-Obama race during the next few weeks into a 2008 version of the white/Black conflict of the mid-1990s that emerge around the O.J. Simpson case.
Let's hope that doesn't happen. Please, Bill. Don't go there. Working people need for the GOP to lose in November. It will be difficult enough for the unpopular Hillary (unpopular in a way similar to the way that many Dems didn't like the patrician Kerry) to win in November. Let's not make it worse by needlessly antagonizing the Dems' base in the Black community across this country.
frankly0 wrote on January 9, 2008 4:27 PM:I was likewise really struck by Obama's "you're likable enough Hillary". The affect was just terribly cold, smug, and patronizing. It's was really jarring to watch. It didn't suggest that there was a lot of genuine feeling in the man under the charismatic facade.
patboy wrote on January 9, 2008 4:31 PM:Why can't HRC victory be attributed to the fact that Iowa Caucus allow do over when your candidate is not viable; while the NH Primary allows only one vote per person.
None wrote on January 9, 2008 4:33 PM:Sometimes it puzzles me why there is so much bafflement about polling results and why pollsters don't even TRY to use the wonkish explanation. Here: took the poll results from Pollster.com for the last 5 surveys. Then, saw what the sampling error and confidence level for each was. This gives a range where we can expect the result to be. "O" = Obama's result, "C" = Clinton's result
ARG: 36% Suffk:34.7% Zogby:38.6% Rasmn:35% CNN: 35%
Statistically, ANY result within the range is equally probable. ANY result. It is totally legitimate to take the lowest result for O and compare it to the highest result for C. This is NOT "pick and choose". It's how the probability works. So, what did the polls say with 95% probability?
ARG: Obama 36 vs Clinton 35
Sffk: Obama 34.7 vs. Clinton 38.3
Zogby: Obama 38.6 vs. Clinton 32.4
Rasmn: Obama 35 vs. Clinton 32
CNN: Obama 35 vs. Clinton 34
Final numbers, 36 vs. 39
A 3-4% change for both over the weekend based on all the factors you've already listed? Not difficult to believe at all. "Stunning" and "surprising" it's really not, either. The polls are all lies and propaganda and useless? Don't seem that way. But of course, it's not exciting to report "Obama could get somewhere between 36 and 44% and Clinton somwhere between 27 and 35%, which means that basically they could be only 1% apart in the end." You don't even have to consider the 95% confidence to explain this one.
frankly0 wrote on January 9, 2008 4:36 PM:Just a point about how big the swing must have been between Friday and Saturday to Sunday/Monday.
Doing the math, if on Sunday/Monday Hillary came in at 2 points down, and the average over the three days was 12 points down, then over Friday and Saturday, she must have been 17 points down.
Now that's a swing.
frankly0 wrote on January 9, 2008 4:42 PM:None,
Your argument is confused.
You're right that for a given individual poll, you can say that the margin of errors work out the way you suggest.
But what's certainly wrong is to suggest that it makes sense to imagine that ALL of the polls are wrong in precisely the same way. Randomly, you'd expect that variations from the polled result could go in EITHER direction, with equal probability.
Obviously, under the scenario you're describing, they are not randomly going in either direction, but in each case in only one specific direction -- to the high extreme for Clinton, and the low extreme for Obama.
Your "wonkish" explanation is just dead wrong.
markg8 wrote on January 9, 2008 4:43 PM:Heard on NPR today anecdotal testimony from a NH voter that he and many friends he's talked to since were set to vote for Obama but looking at his certain landslide decided to take Republicans ballots to vote for McCain for some odd reason.
If Zogby or any of the others had broken out their numbers there might have been a different result.
Sarah wrote on January 9, 2008 4:50 PM:I'm sick of hearing about Obama's "likeable enough" line at the NH debate and how mean it was. At the time, Hillary had spent the previous week attacking Obama's record and lack of experience. She was the one who took the gloves off, basically calling him a puffball that's all talk and no experience.
I agree that Obama could've responded better. But it was a 2-second sound bite, not a prepared statement. It strikes me how the public has reacted to this event with such sympathy for Clinton, even saying that Obama and Edwards teamed up against her unfairly. As one of the toughest people in Washington, she can take it. Even if she is a girl.
frankly0 wrote on January 9, 2008 4:55 PM:But it was a 2-second sound bite, not a prepared statement.
The exact point was that it was not prepared, but spontaneous.
That's exactly what people didn't like. It's when the mask falls that people pay their most careful attention in highly scripted events like the debate.
None wrote on January 9, 2008 5:29 PM:frankly0:
Sorry, but they're all indpendent, within their range they have equal probability of happening. They're not all "wrong" in the same way. Reference to normal distributions and normality tests is relevant:
http://en.wikipedia.org/wiki/Normal_distributions
None,
I repeat, you are simply confused.
Peter wrote on January 9, 2008 6:48 PM:as for the 'let the voters make up their minds' meme - i guess it's safe to say it now that Tom Brokaw has - let's put things in perspective.
of the corruptions of (parliamentary) democracy we see before us during this election cycle, these just happen to be some of the more superficial ones - polls and dropping candidates from tv debates, etc. really, you have little to no choice of who you get to vote for - now or in the general.
the whole point of these debates is to entertain us all - as they do - and prevent any real discussion issues most Americans care about.
now stop meddling in the affairs of learned men. your job is to ratify our decisions (when we allow it), and you should be happy we give you that much power.
Christine wrote on January 9, 2008 10:24 PM:Is it beneath everyone's dignity to even consider that the vote count could be wrong? From what I have heard from NH voting specialists, over 80% of the votes in the state are counted by a private company using Diebold optical scanners which like DREs have been proven to mistabulate both by design and by accident.
Sean wrote on January 9, 2008 11:56 PM:http://www.reuters.com/article/newsOne/idUSN0433304720080108
Obama continued to feed on the momentum from his performance in Iowa last week, when he and Republican Mike Huckabee scored breakthrough wins that left Clinton and Romney reeling.
Obama, a senator from Illinois bidding to make history as the first black U.S. president, led the one-time Democratic front-runner 42 percent to 29 percent. Former North Carolina Sen. John Edwards was third with 17 percent.
"Obama is still on a roll and not slowing down. He had another big day," pollster John Zogby said.
Obama led Clinton in all categories of voters except women and voters over the age of 65, and was pulling away from the New York senator among base Democratic voters.
"This is the same thing he did in the closing hours in Iowa," Zogby said of Obama.
The rolling poll of 862 likely Democratic voters and 859 likely Republican voters was taken Saturday through Monday. It has a margin of error of 3.4 percentage points.
-----
To the claim that late deciders broke overwhelmingly for Clinton: ABC's (outstanding) polling director, Gary Langer, has this response: "...Wednesday afternoon p.s.: Some folks are suggesting that "late deciders" made the difference - a common explanation for poor estimates. But the exit poll doesn't support the notion. Remove voters who decided on Tuesday and the New Hampshire exit poll result is Clinton +2 – exactly her actual margin. (Among those who decided "just today" it was Clinton +3.) Next theory." http://blogs.abcnews.com/thenumbers/2008/01/new-hampshires.html
Exit poll: http://www.cnn.com/ELECTION/2008/primaries/results/epolls/index.html
Neil wrote on January 10, 2008 12:07 AM:Jon Stewart just had John Zogby on his show and completely destroyed him. Stewart was honestly surprised by how badly he made Zogby look, which is funny because this is Stewart's job.
gtash wrote on January 10, 2008 8:19 AM:Instead of laughing at Zogby and others, some serious looks into voting machine irregularities should be reviewed. Why was every pollster so wrong?
There is a chance that they were all coincidentally miscalculating.
There is a chance that people had a last-second conversion to Hillary (allegedly a moment a heart-felt sympathy or feminist outrage--take your pick) behind the curtains of the voting booth.
There is a chance everybody actually thought the elections were fairly counted.
I want to know how McCain can end up splitting the independent vote with Obama, yet only the race between Hillary and Obama was affected. I know there are voting blocs in theory, but I don't believe people vote monolithically nor divide so cleanly. It's not in the nature of people nor of statistics.
Jim H wrote on January 10, 2008 5:55 PM:The pollsters didn't screw up. The reporters and pundits screwed up by only ripping the headline off, and not noticing, or publicizing the fact that as much as 40% of the Democrats hadn't made up their minds or weren't committed. So the results were in line with the polls, if you know how to read polls. Instead, we have our Diebold wackos, and Chris Matthews and Jesse, Jr. start talking about racism, with no evidence at all. People broke the other way this time. It's not as settled as activists on either side would have it. Apparently, there were even people deciding not to vote for Obama and vote for McCain, because, of course, these candidates have so much in common. Oh, well, what do I know?
Erik Larson wrote on January 10, 2008 6:20 PM:Perhaps we just need to make sure the vote-counting is done in the open, by hand in the local precincts, instead of on proven unreliable and easily hacakble voting machines with secret software. Hand counting can be done in several hours, and with all the indications of widespread electoral fraud, starting with the 2004 elections which people in Ohio were convicted of manipulating, there should be no shortage of citizens willing to turn out and help count paper ballots. At a minimum, anyone that wants to vote on a paper ballot should have the right to do so, and the Op Scan machines need to have open source code.
Who's afraid of having elections that are verifiably honest? Let's put an end to the "conspiracy theories". Ensuring that elections reflect the People's will is crucial, if government is to have the appearance of legitimacy.
