« Manage the Seemann Campaign for a Day! | Main | Is Yucca Mountain Really a Big Deal? »

Wednesday, October 06, 2004

New Zogby Interactive Out

Posted by DavidNYC

How do ya like them apples:


Kerry retakes Ohio and Nevada, albeit by very narrow margins. But Kerry improved in 12 of 16 states since the last set of polls (released 9/20). In the four states in which Kerry lost ground, three (OR, NM, and MN) are where he's got some of his biggest leads.

Zogby thinks the debate had something to do with this, which would mean it had a fairly strong effect, because part of this poll was taken before the debate.

UPDATE: A correction: Ed in the comments observes that all of the polling was conducted after the first presidential debate - my goof. Perhaps that's why this release was a couple of days later than expected.

Posted at 07:07 PM in General | Technorati


How come no network even bothers to talk about edward's retaliation

"The vice president, I'm surprised to hear him talk about records. When he was one of 435 members of the United States House, he was one of 10 to vote against Head Start, one of four to vote against banning plastic weapons that can pass through metal detectors.

He voted against the Department of Education. He voted against funding for Meals on Wheels for seniors. He voted against a holiday for Martin Luther King. He voted against a resolution calling for the release of Nelson Mandela in South Africa."

Nobody seems to have refuted it either....

Posted by: mram at October 6, 2004 07:40 PM | Permalink | Edit Comment | Delete Comment

I wish the election was tonight.

Posted by: David Trinh at October 6, 2004 08:39 PM | Permalink | Edit Comment | Delete Comment

Zogby seems a bit optomistic from a Kerry point of view, but he is staking his reputation on this high profile WSJ poll. I mean, if he's way off on this, then he's going to really have egg on his face. Zogby might be best at figuring out who is actually going to show up and vote. His polls probably are picking up the 2-3% of Democrats who fly below the polling radar. Look for Kerry to pick up a few percentage from Democratic GOTV efforts on election day.

Posted by: Rock_nj at October 6, 2004 10:12 PM | Permalink | Edit Comment | Delete Comment

Actually, I believe that Zogby says in their "Overview" comments, that the polling started right after the end of the debate, so that *all* the polling represents post-debate opinion.

Posted by: Ed Fitzgerald (unfutz) at October 6, 2004 10:27 PM | Permalink | Edit Comment | Delete Comment

As for Zogby, I've made the same "staking his reputation" argument in the past, but there's just no way that both Zogby and SurveyUSA can be right in Tennessee. Zogby has Bush leading by a scant 0.9%, while SUSA (pdf) has Bush ahead by 19%.

That's a major difference, even given that SUSA uses likely voters.

Posted by: Ed Fitzgerald (unfutz) at October 6, 2004 10:32 PM | Permalink | Edit Comment | Delete Comment

I'd like to believe Zogby's interactive polls work (his national telephone polls are a different story and don't produce outliers as these do). But I have trouble believing that they do. I'm one of the regular respondents-- Zogby emails me and I take the poll, which collects the usual array of personal characteristics along with presidential preference, party affiliation, and Gore-or-Bush-in-2000. Zogby must then weight the hell out of them to produce anything like a representative sample. The trouble is that I'm not sure whether he can. Especially among less Internet-savvy groups and smaller subsets-- Latinos making under $50K a year who didn't vote last election (for example)-- it's hard to imagine that polling only Internet users willing to take a Web poll would not itself distort the sample. And remember that these are STATE polls, not national ones. How many Internet users would Zogby have to poll in order to get a representative set of Latinos from NH, or African-American New Mexicans? Do you think that Latinos from NH who are willing to take a Web poll are more likely to vote for Kerry than Latinos from NH who are not, all else being equal? I do. That's why I'm afraid I can't trust Zogby.

If Zogby's state polls turn out more accurate than SUSA I will rejoice. But if Zogby's more accurate I'll have lots of other reasons to rejoice, and so will most other readers on this blog.

Posted by: accommodatingly at October 6, 2004 11:03 PM | Permalink | Edit Comment | Delete Comment

That TN Zogby to SUSA spread might be the biggest I've seen to date. Bush at 58% in TN is also just about the highest I've ever seen him in any place that even nominally approaches swing state status.

I think Zogby has something of an out if he turns out to be badly wrong: "Hey, it was a noble experiment! We wanted to see if Internet polling could be done - and it turns out, it wasn't so good." On the other hand, we are pretty dismissive of Rasmussen for their disastrous B+9 national prediction four years ago. So it is a risk for the Zogster.

Posted by: DavidNYC at October 7, 2004 02:13 AM | Permalink | Edit Comment | Delete Comment

Zogby is experimenting, and right now they are testing their hypothesis. This is why one cannot take their figures seriously at this time, for their hypothesis has not yet been proven. I hope they are accurate, but I wouldn't bet the farm on their figures just yet.

Posted by: pepe at October 7, 2004 06:51 AM | Permalink | Edit Comment | Delete Comment

Closer inspection reveals the apples aren't as tasty as they could be:

Kerry leads in Arkansas by 0.2
Kerry leads in Ohio by 0.3
Kerry leads in Florida by 0.4

virtually a dead heat

Posted by: anon at October 7, 2004 07:50 AM | Permalink | Edit Comment | Delete Comment

No other polling firm, except for one firm that I can not remember, came as close to predicting the 2000 presidential race than Zogby. I have faith in Zogby.

Posted by: Peter at October 7, 2004 10:10 AM | Permalink | Edit Comment | Delete Comment

Zogby tries to guage what the compostion of the turn out will be like. That's his little secret for coming up with accurate polling. Most polling firms just randomly call people. Zogby tries to do representative samples, and obviusly it's working quite well. What good is a random sample that polls 30% Republicans 25% Democrats and the rest indies, when on election day the actual turnout is more likely to be something like 40% Democratic, 35% Republican and the rest indies. That's Zogby's secret, and I'm sure other polling firms will soon emulate him.

BTW, Gallup oversamples Republicans, which is why their polls are so biased towards Bush. They are essentially worthless polls.

We'll see how accurate Zogby's interactive polling turns out.

Posted by: Rock_nj at October 7, 2004 10:24 AM | Permalink | Edit Comment | Delete Comment

Zogby is not staking his reputation on the Interactive polling. His reputation relies on regular polls, which I believe he is doing correctly. The interactive polling should be considered very unproven right now -- we just don't know if his model is OK. Als, the self selection used gives too much room for poll manipulation.

And frankly, TN is unbelievable. Native Prince Al Gore couldn't carry the state, and a Yankee liberal is within 1% ? Impossible.

SUSA is probably off as well. Last time Gore lost by 3-4 points, so I would expect a 10-11 point victroy for W here (assume that Gore's nativeness gave him 6 points or so beyond Kerry).

I would consider the Zogby Interactive polls valid only for trends, if that.

Posted by: erg at October 7, 2004 11:22 AM | Permalink | Edit Comment | Delete Comment

It's not impossible for Kerry to win Tennessee. Probable? Hell no. But still possible. For one, Tennessee, despite being a Republican leaning state, is far more moderate than most Southern states. Democratic governor, legislature, and majority Democratic delegation to the House. Of course, this is tempered by the fact that it's also the state with Rhea County and Bill Frist. But a strong showing on Election Day, even if it's a loss, could be good news for long term Democratic prospects.

Posted by: Dale at October 7, 2004 12:17 PM | Permalink | Edit Comment | Delete Comment

Yes, but if TN rejected its native son Al Gore, it's hard to imagine it will cozy up to a Yankee from MA!

Posted by: Pepe at October 7, 2004 02:16 PM | Permalink | Edit Comment | Delete Comment

OK -- some bad news out of FL. Quinipiac shows a 7 point Bush lead. Betty Castor is tied with Mel Martinez.

ARG is th eonly company that has shown a Kerry lead recently. My guess is that the President got a lot of free air time because of his visits after the hurricanes. Kerry really could not do that, because it would have looked like policiticking.

Still, Kerry has done better than the last Q * poll. Fl is still within reach.

Posted by: erg at October 7, 2004 02:47 PM | Permalink | Edit Comment | Delete Comment

Yes, Bush has been helped by the hurrican relief effort. The polls will tighten up as the election approaches. Look for a nail biter in Florida.

Posted by: Rock_nj at October 7, 2004 03:52 PM | Permalink | Edit Comment | Delete Comment

What, still no Colorado on the Zogby site?

They picked their battleground states early on. Could NJ, AZ also be considered battlegrounds?

Posted by: Dory at October 8, 2004 01:46 AM | Permalink | Edit Comment | Delete Comment

watch West Virginia!

Posted by: flocal at October 8, 2004 07:14 AM | Permalink | Edit Comment | Delete Comment

Other Zogby polls:

VA: Bush 49.6%, Kerry 47.4%
CO: Bush 48.0%, Kerry 49.2%
NC: Bush 50.3%, Kerry 47.8%

Posted by: DFuller at October 8, 2004 08:46 AM | Permalink | Edit Comment | Delete Comment

At the end I think it will come down to Florida again. Ohio will stay Republican, and Pennsylvania will stay in the blue column. Long term the upper Midwest concerns me but I think advances in Colorado and places like Tennsesse and Virginia will counteract any Midwest erosion. I don't see any emerging democratic majority but I think over time the dems situation should slightly improve.

Posted by: Brett at October 10, 2004 06:17 PM | Permalink | Edit Comment | Delete Comment