Wednesday, September 22, 2004

Zogby and ARG Polls

As one commenter pointed out, the Zogby state polls are done interactively through the web, unlike traditional phone polls. Perhaps Zogby's methodology is wrong, but I think this is a reasonable attempt to avoid some of the problems of phone polls--non-responses, more cell phones, etc. And the results of these interactive polls are not wildly different from those of traditional phone polls. For example, look at today's ARG polls of all 50 states. The have the national race tied at 47-46 for Bush, and the electoral college is about the same. Bush and Kerry each have leads outside the MOE in states with 133 and 132 electoral votes, respectively. When you toss in leads within the MOE, Kerry has 270, Bush has 253, and WI and WV are dead-even. Are these polls correct? I have no reason to think that they are any less correct than any other polls with similar timing and methodolgy. Perhaps they overstate Kerry's totals or perhaps Bush's. I don't know. But they do reinforce my impression that this is a close race, nationally and in the Electoral College, and that neither candidate has put this thing away. If anyone is so sure of the outcome, then put your money where you mouth is and give me odds.

5 comments:

Anonymous said...

I'm not sure why folks have been geting all hot and crying about "partisan polls" just because
Bush's lead isn't holding up. He got a nice convention bounce but events on the ground in Iraq since have not been positive. Kerry has actually had a good couple of weeks and seems to have turned a corner.

No one should be suprised if the race is tightening. It's been tight all year. I think the only anomoly was the seperation Bush had in the weeks after his convention.

Thomas said...

I certainly wouldn't say that Zogby's methodology is unreasonable, and I hope that wasn't what I communicated. Rather, I meant just to suggest that his methodology--a reasonable attempt to adapt polling to this modern world of ours--is untested.

In that scenario, I'd give greater credence to the ARG polls than to Zogby's.

I don't doubt that this is a close race, and I don't believe that Bush has put the thing away. But the evidence, conflicting as it is, would lead most observers to think that Bush has a real and significant lead both nationally and in the electoral college.

Palooka said...

Listen, nobody knows which poll is more accurate, which methodology is promising. Is normalizing the sample for party affiliation based on 2000 results sound, etc.

But my point is this--there have been many polls showing a pronounced Bush lead, and now there's some that show it tied. Which to believe?

To put too much faith in Zogby, despite his untested methodology, is unjustified.

It's astonishing that Zogby thinks he can acheive less bias by doing an "interactive poll" than traditional phone polling. Does traditional polling have biases? Sure, but I strongly suspect that is significantly less than, say, an "interactive poll."

Suggestion: Why don't you talk about party normalization and which polls are following that methodology and which are not. Is Zogby normalizing based off of 2000?

Seems like Gallup and others are normalizing based off of recent, not historic, trendlines. Which is more sound?

Again, I only pointed out that reason the electoral map was looking better for Kerry is because of all of the Zogby polls, which are clearly out of the norm by around 6 points.

MWS said...

Who knows and who cares? Just vote. It is impossible for a lay person such as me to have any idea which polls are valid and which are not. I hope it's a close race whoever wins because I think presidents often go bananas after landslides and think they have been annointed by God.

Marketing man said...

You have a great blog here! I will be sure to book mark you.
I have a automatic in lead lead lead leads.com mlm opt responder responsive surveyed site. It pretty much covers automatic in lead lead lead leads.com mlm opt responder responsive surveyed related stuff. Check it out if you get time :-)