Missing the outcome in California
Super Tuesday may have lacked a runaway winner in either party, but when it came to anticipating the outcome of both primaries in California, there was one clear loser -- the Reuters/C-Span/Zogby poll.
The man in charge of the survey, John Zogby, didn't equivocate or obfuscate today when assessing the bum results his firm came up with in the Golden State. "We blew it," he told us.
He also pointed out that the polls he supervises got the victors right in six other races on Tuesday (perhaps most impressively, his had Barack Obama winning narrowly in Missouri, unlike any other of the last-minute surveys for that state).
Still, the final figures his company posted for California spoiled the day for him.
In the Democratic face-off, based on interviews conducted Sunday and Monday, the Zogby poll gave Obama 49% of the vote, Hillary Clinton 36%.
The 13-point margin was close to ...
the mark, but there was that little problem of who ended up on top. Although the vote tabulation isn't complete in California (and won't be for days), the count as of now gives Clinton 52%, Obama 42%.
In a release Zogby International sent out today, the pollster tried to explain what happened: "It appears that we underestimated Hispanic turnout and overestimated the importance of younger Hispanic voters" (i.e. far more Latinos, especially older ones, cast ballots than anticipated, and those voters went heavily for Clinton). Also, the release added: "We also overestimated turnout among African American voters."
That addressed the Democratic contest, but what about the Republican race?
The findings in the final Zogby poll: Mitt Romney 40%, John McCain 33%.
The almost-complete actual results: McCain 42%, Romney 34%.
Zogby told us that his staff still is vetting what went wrong in this case, but the problem likely is that those voting in the GOP primary were less solidly Republican than the firm's model anticipated. Although only registered Republicans could cast a ballot in it, exit polling showed that roughly 20% of those who showed up for it viewed themselves as independents.
Summing up his thoughts on the two California polls, Zogby said: "This is not one of our happier moments."
-- Don Frederick



The polls are probably correct; BUT the California voting system was wrong. The election snafus are inexcusable and resulted in the poorest of planning on the part of state officials. Americans thought that California was better than this. Too bad you let your country down...
Posted by: Brenda | February 06, 2008 at 04:08 PM
Polls are hardly ever accurate if RACE is involved.
Posted by: smokey | February 06, 2008 at 04:35 PM
Is it possible that, since so many people in California voted by mail (about half, I believe), 50% of the votes were cast two weeks ago when Clinton was leading by a wide margin, and as the election drew nearer, those voters changed their minds? This would explain why the day-before polls showed Obama tied or leading: those voters who now favor Obama had already cast their votes for Clinton.
Posted by: Matt | February 06, 2008 at 04:37 PM
Off 23%? How can anybody miss that much? I bet the "Bradley Effect" was also in play athough nobody in the media wants to talk about it.
Posted by: Ralph | February 06, 2008 at 04:45 PM
there is only a 3 to 5 point margin of error stated for most studies, these ones show that error margin is often much wider than the surveyors imagine. goo case study in not believing polls very much.. just a little..
Posted by: Joseph Cecil Smith | February 06, 2008 at 05:03 PM
Most people think that they control things. There is the unseen factor whom you may call whatever It is. And this is the one that decides for the we are reaping now.
Destiny prevails!
Posted by: miriam biteng | February 06, 2008 at 05:40 PM
Zogby didn't get it wrong, Obama won among CA voters who actually voted yesterday but up to 50% of CA voters did not go to the polls, they voted early when Hillary still had double-digit leads.
Posted by: Chris | February 06, 2008 at 05:45 PM
There was no mistake in the Zogby poll.
The skewed predictions where intentional Obama propaganda paid with the millions of dollars backing him. You can censor this statement out of the blog, but it will only further prove the point.
Posted by: Jim & Z. | February 06, 2008 at 06:51 PM
Zogby is either way off or spot on. He was wrong about New Hampshire. He was the one who started the whole Obama is leading in New Hampshire thing. He was right about South Carolina, but he was way off again in michigan with the Republicans. Too often he tries too hard to make a headline. We had a case with Zogby in last year's congressional race that showed the Republican incumbent up twelve points until the Democratic candidate pointed out a serious flaw in his methodology. He pulled the poll after some delay and did another poll which showed the incumbent up by two percentage points. The poll was never officially rerleased, however, just reprted on the day before the election. I got the impression that he would not have changed the poll if he didn't get called on it. He sometimes gets too busy rushing to make a headline.
Posted by: Ed Garfield | February 06, 2008 at 08:11 PM
Survey USA got the California results exactly right: Hillary 52% and Obama 42%. Zogby should blame the fact that he and his pollsters got caught up in Obama hype.
Posted by: Joseph | February 06, 2008 at 08:35 PM
The comments in this thread range from hysterical to outright insanity. Obama paid off the pollsters? The California voting system is off by 12% of votes? What are you people smoking. You people depress me, I can't believe how dumb comments can get. Here's a clue: observations lead to understanding. Nutty ideas, however, spring from basketheads.
Posted by: Bjorn Tipling | February 07, 2008 at 01:17 AM
Well, let's try again:
We've got the following prediction: Obama 49 % Clinton 36 % -> undecided 100 - 49 - 36 = 15 %
The margin of error of the poll is typically 3-4 % This estimate of the margin of error implies the following hypotheses:
1. A-select choice of interviewees (yeah, right)
2. and most important - it assumes the undecideds will distribute over the two candidates in proportional fashion !
However, that's not what they do ! Undecideds are just that - they choose at the last minute based on some detail that happened the last hour - they probably swing massively for the same person.
Hence the actual difference between poll and outcome can easily be 15 + 2 * 4 = 23 %.
Posted by: Toon Moene | February 07, 2008 at 12:24 PM
If you can't swing a vote by fiddling the polls then please don't worry.
In Florida we voted like California for Clinton. Now we are told the Obama campaign rejects the secret ballot and wants a caucus instead.
So you might find that you are actually not allowed to vote for Clinton. Perhaps you did not understand the way this change business works when Obama talked about change.
What you are supposed to do is to vote the way the polls tell you to.If you don't or if the Obama camp does not like the result, then you just have to come back again until the Obama campaign is satisfied with your choice.
Wow ! This is a new approach to Democracy. The polls tell you how to vote, then you get to keep coming back until you vote that way.
Posted by: david Hopcroft | February 07, 2008 at 01:04 PM
The polls were very close to being correct. When voted started in the first week of January Clinton was 20 to 25 points ahead.Hundreds of thousands voted in the early voting. The massive surge for Obama in the last week helped him to close the gap but was not enough to counteract the early voting. The late surge was what the polls were correctly indicating. Without that surge he would have been slaughtered in California. I hope he has learned that whenever there is early voting in a state he has to make time and get there.
Posted by: Marguerita | February 07, 2008 at 04:32 PM
"Most people think that they control things. There is the unseen factor whom you may call whatever It is. And this is the one that decides for the we are reaping now. Destiny prevails!" - miriam biteng @ February 06, 2008 at 05:40 PM
TRANSLATION? WTF! Learn English or review your garbled thoughts before posting. Oh, is it the 7th already? Alrighty, Miriam won't be back to see me kvetching about her moronic ramblings, but the rest of you get the fuck on board and make some gottamn sense.
Posted by: Joolio Downs-Blythe O'Skool | February 07, 2008 at 05:38 PM
These polls, like the news anchormen, are trying to control the outcome of the elections. They think they can dictate to people how to vote. Move aside you corporate dead weights, the people will decide who they want. Go Hillary!
Posted by: np | February 08, 2008 at 12:37 PM