California Results Expose Bad Polling

In the wake of Hillary Clinton's decisive victory Tuesday in the California Primary, one pre-election poll from C-SPAN/Zogby/Reuters that showed Barack Obama up by 13 points in the Golden State is getting a lot attention Wednesday morning.

On the morning of Super Tuesday, Zogby released its final tracking poll, showing Obama at 49 percent and Clinton at 36 percent, with 9 percent undecided, with a margin of error of 3 percent.

Pollster John Zogby said this as his poll was released:

In California, we have Obama polling into a 13-point lead. Monday was another big single day of polling for him there. What has happened here is that in addition to building leads among almost every part of his base of support, he has dramatically cut into Clinton's lead among Hispanic voters.

According to the Los Angeles Times on Wednesday, with 97 percent of the precincts reporting, Clinton won the Golden State with 51 percent of the vote, compared to Obama's 42 percent.

Andrew Sullivan called Zogby "Another Big Loser." Matthew Yglesias wrote about filtering out "the most-exuberant Zogby-filled dreams." Even the Brits are pointing out the error.

On Tuesday, in a post entitled "Somebody's Gonna Be Wrong," Pollster.com's Mark Blumenthal wrote about Zogby's lack of transparency about his methodology.

In his post, Blumenthal compared Zogby's results those of another pre-California poll from SurveyUSA that showed wildly different results.

Blumenthal writes:

If ever there was a case for better methodological disclosure by pollsters, this is it. If one poll, conducted entirely over the last 48 hours, shows a candidate leading by 10 points, while another conducted in the same state over the same time period, shows a another candidate leading by 13, and we cannot see enough of the details of how the polls were done to at least explain why they differ, why should we trust what any of these polls tells us?

On Wednesday, Zogby released a post-Super Tuesday analysis that includes this note about California:

Some of you may have noticed our pre-election polling differed from the actual results. It appears that we underestimated Hispanic turnout and overestimated the importance of younger Hispanic voters. We also overestimated turnout among African-American voters. Those of you who have been following our work know that we have gotten 13 out of 17 races right this year, and so many others over the years. This does happen.

John Zogby is a Huffington Post blogger.

Comments

[DRaymond]

Also adding to the problem is that the votes of 94,000 independents in Los Angeles County were not counted due to a bad ballot design.

That is about half of the independent ballots in Los Angeles, a group that Obama ran strong with.

http://www.latimes.com/news/printedition/california/la-me-vote7feb07,1,1380058.story

[SouthCentralDel]

5.5 million Absentee ballots. Sent in pre-Obama surge.

[Orikinla]

Polls are being fixed and rigged in favour of Barack Obama.

[vegasobserver]

Latinoids voted for Hillary because they are family-oriented and wanted a "mother figure" in the White House.

[Purple Girl]

It's become painfully obvious that pollster are not surveying the public- they are merely dispensing the mind set the Corporationist what US to fall into.
Having a Bach in Soc- I've learned (like statistician) you can 'prove anyting' if you use flawed methods.
Where were they standing (phoning) on Rodeo Drive? Or East LA, Or the Castro, Or Market Street???
They are not intersted in telling what your fellow Americans are thinking- they are interested in TELLING YOU WHAT TO THINK.
WEll I think Hillary has become as corrupt and arrogant as Cheney. I'm betting SHE IS HIS GIRL (Halliburton, KBR, Blackwater,Titan, Arab Em., Isreal's....
Worth catching "Iraq for Sale" On Encore. Right up there with Inconvenient Truth and Who killed the Electric Car
I've come to realize I owe those 'crazy' conspiracy theorist an apology. They aren't paranoid, We Do have entities out to get US.
I'm on Board
Cave Adsum!

[murz]

I thought Huffington Post monitored the comments. How could they allow an obvious lie to stay up?????

[Ides]

5.5 million Absentee ballots.

[FreedomLoverX]

If you are not going to count all the votes (20% not counted), how will you know?

[ljsfolly]

The polling process has been increasing screwed up with numbers beyond belief being put out for the sides and often wrong. There is a piece of the truth to those whosay they will vote one way then vote another. Whether its because one is female and one is black or becuase they don't want to explain their bias who knows? The other thing that happens increasingly is people do not answer the phone when they don't know the caller so avoid the calls. Those who will answer the calls from polls are a type of person who feels compelled to answer any call and might give a different slat to the polls. Now my beef is when the cable news heads declares a winner of a primary when there is 13% total votes. Then the talking goes on and on and gets more and more annoying as they explain their decision and act as if it's fact. I actually changed the channel and no longer watched the returns on super tuesday because all the cable channels were just a bad as the others. I like being informed but the way the cable news is providing mundane garbage and calling it reporting is not what I want out of them. Is it too much to give the races time to reach a majority of votes rather than projections, projections and declaring winners before 20 % votes are in?

[Chandidevi]

The polls got it wrong? Perhaps the polling PRECINCTS got it wrong. My paper ballot was thrown into a green plastic box which used a paper clip as its seal. The optical scanning machine was down. How many votes ended up that way, being carelessly handled and thrown into a hidden box. The Secretary of State fraud alert office said they were overwhelmed with reports from throughout California.

[tstreet]

Survey USA got this right on by showing Hillary winning by 10 points. Despite this, the media showed the Zogby poll all day and never mentioned the Survey USA poll. Survey USA shows their track record on their site. It appears they do a better job than most of the rest of the pollsters.

The fact that the media, including MSNBC never mentioned Survey USA makes me highly suspicious about their fairness and objectivity.

[genseric]

I find it curious that sometimes I can see the polling results mirror the pollster's own political bent. Zogby is a muslim and even though Obama was converted from islam, my guess is Zogby favors him for various ethno-centric reasons. Man, it's hard to find anyone objective these days.

[letsmakesense]

Isn't it about time to consider that maybe the polls were right and the difference lies in the vote count.
Who counts the vote?
See you don't know....
investigate
google while you still can
blackboxvoting
bradblog

[Elusive]

The reason Zogby's polling was so far off is because of the early voters. SurveyUSA took this into account and got an accurate prediction.

I have seen numbers from 1.5 to 3 million early votes were sent in. It makes sense that the majority of these were for Clinton because she was doing so well early on before Obama began campaigning there. Of course many Edwards supporters voted early before he dropped out.

Last night, the initial returns had Clinton at 55%, Obama at 33% and Edwards at 10%. These reflected the early votes.

As the same day votes were counted, it narrowed to Clinton 52% and Obama 42%.

The lesson, don't vote early people!

[vipersdad]

I think Polls are for "entertainment purposes only." Anyone who has spent even a modicum of time working in the social sciences and with statistical analysis knows there are many pitfalls and traps that conspire to produce such results.

We all wring our hands time and again about how some polls "got it wrong," but in every election cycle there are winners and losers among the major polls...goes with the territory.