Zogby

[DavidS]

Conservatives beware - Nate Silver (FiveThirtyEight.com: Electoral Projections Done Right) rips Zogby a new one.

Now, a lot of you have read my analysis on the polls and my strong defense of them. Many of you have also read how much I dislike the Zogby poll - not because he's partisan like GWU/Underground and Strategic Vision, but because his polls don't show any consistency whatsoever for more than 2-3 day period. One day McCain's up 2, the next Obama's by 7.

To use a metaphor, I minored in meteorology and studied the computer models (GFS, EURO, NAM, GFDL, JAM, NOGAPS, etc.). Weather models tend to give you information about weather up to 16 days (384 hours). Let's say I were to see a giant hurricane hitting North Carolina in 16 days from now on the GFS model. I'd make a note of it and look for it on the computer models to see if this model is catching onto something. If I see it on some of the other models, let's say 50%, for a period of time of 5 days or greater, then we have a trend and it's time to look for physical evidence to support this... i.e. sattelite maps.

So to are polls. If one poll, such as Pew, shows Obama up by 15, that to me is a massive hurricane about to hit NC. If none of the other polls show it, I make a note of it and look for a trend for a few days to see if the computer model is still picking it up. If not, then it was a blip. Ground noise. I mentioned the NOGAPS, because it is one of the most unreliabale outlier models in the country. It shows me blizzards in Miami because it's overdoing the amount of cold air that's being pushed down from a canadian high pressure system.

SO! What does this have to do with Zogby? Zogby is like the NOGAPS. It's all over the place. It's not showing me any consistency whatsoever. In addition, here is what Nate Silver, who specializes in these kinds of things says:

FiveThirtyEight.com: Electoral Projections Done Right: Trick or Treat

Trick or Treat

Do you spook easily?

Judging by the response in my inbox, some of you do.

Matt Drudge is touting the results of a one-day sample in a Zogby poll, which apparently showed John McCain ahead by 1 point.

There are a couple of significant problems with this.

Firstly, there is a reason that pollsters include multiple days of interviewing in their tracking polls; a one-day sample is extremely volatile, and have very high margins for error.

Secondly, the Zogby polls have been particularly volatile, because he uses nonsensical party ID weightings, which mean that his weighting process involves making numbers doing naughty things that they usually don't like to do.

Thirdly, Zogby polls are generally a lagging rather than a leading indicator. This is because he splits his interviewing period over two days; most of the interviews that were conducted in this sample took place on Thursday night, with a few this afternoon. The reason this is significant is because lots of other pollsters were in the field on Thursday night, and most of them evidently showed good numbers for Obama, as he improved his standing in 6 of the 7 non-Zogby trackers.

Finally, there was no favorable news for McCain to drive these numbers. Polls don't move without a reason (or at least they don't move much).

So go out to your Halloween parties and enjoy yourself, and we'll be back to covering the polls for you tomorrow.

 

[Jon]

So to are polls.

I was intrigued until this, then I stopped caring what this guy had to say.

 

[Razalas]

I was intrigued until this, then I stopped caring what this guy had to say.

As was I.

I think I stopped caring what Nate had to say when I read this on his FAQs..

" I vote for Democratic candidates the majority of the time (though by no means always). This year, I have been a supporter of Barack Obama. The other contributor to this website, Sean, has also been a supporter of Barack Obama."

..oh well.

 

[DavidS]

As was I.

I think I stopped caring what Nate had to say when I read this on his FAQs..

" I vote for Democratic candidates the majority of the time (though by no means always). This year, I have been a supporter of Barack Obama. The other contributor to this website, Sean, has also been a supporter of Barack Obama."

..oh well.

Don't like his political choice, no problem. Dispute his science.

 

[Jon]

Don't like his political choice, no problem. Dispute his science.

He does a pretty good job of disputing his own science. In his first point, he defends using running polling data, and in his third point, he criticizes Zogby for it.

 

[DavidS]

He does a pretty good job of disputing his own science. In his first point, he defends using running polling data, and in his third point, he criticizes Zogby for it.

Track Record: Zogby's telephone polls are not quite as bad a reputed, ranking just slightly below average (his internet-based Zogby Interactive polls are another matter altogether). However, they seem to be getting worse rather than better. Zogby was one of the few pollsters to call the popular vote for Gore in 2000, but he missed high on Kerry's numbers in many state-level polls in 2004, and then had an erratic primary season this year.

House Effect/Lean: For the reasons specified below, so far has leaned Republican by a point or two.

Features/Strengths: Zogby's write-ups are generally entertaining, and he reveals more than most about the progress of the candidates in individual days' results. One of two pollsters to publish results with a decimal place intact, which we like.

Zogby is the first pollster to go to press each day, publishing in the wee hours of the morning, although the way that he accomplishes this is to split his sample periods over two days. That is, a "day's" worth of polling consists of interviews from that afternoon plus the previous night.

Quirks/Concerns: There is one very, very significant concern with Zogby, which is that he has a longstanding rule to set his party weightings based on the exit polls from the most recent election. In this case, that means 2004, when a roughly equal number of Democrats and Republicans turned out. However, according to essentially every available poll, Democrats now have somewhere between a 5-point and a 10-point advantage in party ID.

 

[Jon]

How does that refute what I just said? Oh, right, it doesn't. I was pointing out how your original genius article's first and third points conflict, and you respond by talking about his second point.

Way to fail.

 

[Wade]

So, what's with all the Poll fascination? It's just a statistical guess. Heck, 10% of the sheep in this country change their mind, depending on whether their watching Oprah or Foxnews. The count is all that counts. Wait for Christmas morning, you'll know what santa brought you then, for sure.

 

[DiveCon]

So, what's with all the Poll fascination? It's just a statistical guess. Heck, 10% of the sheep in this country change their mind, depending on whether their watching Oprah or Foxnews. The count is all that counts. Wait for Christmas morning, you'll know what santa brought you then, for sure.

well put

 

[Red Dawn]

ZOGBY
Released: November 01, 2008

Obama 49.1%, McCain 44.1%



ZOGBY Electoral Map Predictions

 

[Jon]

I think that electoral map is pretty accurate. I believe McCain will win all those purple states, too. If he wants to win this election, he's going to have to wrestle away a couple of states Obama has a strong lead in. Pennsylvania, Virginia, and Colorado are his best bets.

 

[editec]

Polls are only as good as the pollsters who design them, that's for damned sure.