Replies

Read earlier in the thread. Party affiliation breakdown appears to be sentiment based, not registration based, because many states don't have party registration at all.

I guess I will wait and see what the internals are. I would be careful about repeating anything posted as a comment as factual. The poster is making a mistake. The Party Identification is based on what the Respondent TELLS the pollster. I take part in Zogby's weekly interactive survey where they test how on line results correspond to their telephone surveys. One of the 1st questions is which party are you registered. They then WEIGHT the poll based on past turn out. Because the Democrats have more registered voters nationwide, self-identified Democrats tend to get a large % of responses recorded in the polls. For Example, one poll gave the breakdown as 39% Democrats, 35% Republicans 21% Independents. One of the reason to be cautious with "News Media" polls is they game the numbers. In Sept when they wanted a Bush lead so they oversampled Republicans. When they wanted to show a tight race, they over sample Democrats. Like most scientific measurements, polls are mainly a reflection of the assumptions the pollster makes before measuring.

The point I am making, is the internal numbers on the poll all give Kerry HIGHER numbers on the ISSUES but give Bush the lead overall. That would suggest that Kerry is losing some from his base because these people are saying that while they agree more with Kerry on most issues, they are picking Bush for Pres.

Yes, I do not doubt that Zogby words his question that way, but in the other newsweek thread there is an extensive quote from Gallup saying that they, too, get their party affiliation data by "self identification".

In all cases it is self identified because the official registration lists may not have phone numbers on them, and because some states register people with no party identification. That's not the issue. The issue is how is the question worded:

"Are you registered Rep, Dem or Ind?" A person from a non partisan registration state would not be able to answer this.

or

"Given politics today, do you most identify with Dems, Reps or Inds?"

All can answer this second question. And it would be a valid explanation for why partisan mix varies so widely in these polls. People change their attitudes more readily than they can go down to the county courthouse and change their registration.