Why doesn't Gallup believe voters who say they will vote for Obama?

[Alhena]

From Gallup.com:

In the traditional model -- which identifies likely voters based on current voting intentions and self-reported past voting behavior -- Obama's support has ranged narrowly between 49% and 51%, while McCain's support remains in the 44% to 47% range. Today's 50% to 45% Obama advantage matches the average for this model to date.

The expanded model -- which identifies likely voters based on current voting intentions only -- has generally shown Obama doing slightly better, with his support averaging 52% and ranging from 50% to 53%. McCain has averaged 44% in this model, and his support has been between 42% and 46% since early October.

- end quote -

Poll respondent: I just registered to vote and am going to vote for Obama!

Gallup: No you're not.

Shouldn't the record number of Democrats voting in states like North Carolina and Nevada definitively establish that the "expanded" likely voter model is the correct one? Obama is up by *9 points* today on this model- one more than yesterday. Why does Gallup feel compelled to keep pandering to the right wingers with the "traditional" poll?

[Thrill]

1. Again. They don't believe Black and Young voters are going to vote

Even though Early voting is proving otherwise.

They should just come out and say it

[still_one]

4. They might also believe in the so-called Bradley effect. They are wrong

on both counts

The so-called Bradley effect occurred because Armenians were underrepresented in the polling numbers at that time, and that is why dukemajen won

Does not apply at all in this case

[Bicoastal]

2. Don't worry about it--when your tightest model shows a 5% lead,

the rest is gravy. Clinton won 1992 with 5% on Bush--and that was with Perot helping out.

[Jackpine Radical]

3. Obama has Barr helping. He'll help more than the Nader/McKinney types will hurt.

A lot of discouraged, pissed-off anti-McCain Republicans who might otherwise do a hold-your-nose-for-McNasty gambit.

[crispini]

5. It's right here: "self-reported past voting behavior"

Traditionally, the newly registered vote in MUCH lower numbers. That is why political wisdom tends to discount voter registration drives -- they may register, but it takes an average of seven touches from there to get them to go and vote.

Note that I said "traditionally." This year is different, and I think Gallup has their head up their ass for not recognizing that.

I have access to a "voter turnout model" from my county party. This rates people according to the liklihood of their turning out to vote, 0-100. Of course we're seeing a buttload of the 90-100 people. BUT we're also seeing a spike down there at the bottom at the 0-10 people.

[pnwmom]

6. Because the "expanded" model is new and relatively untested. Gallup is testing the

new model openly, by putting it out there with the "traditional" model. Good for them.