An in depth look at the Gallup poll (National and NC)

[King of New Orleans]

and their implications for the fall election.

First, the National Poll

http://online.wsj.com/public/resources/documents/info-b...

Kerry/Edwards leads 50-46 among LIKELY voters head to head with Bush/Cheney

Kerry/Edwards leads 51-44 among REGISTERED voters head to head with Bush/Cheney

Add Nader/Camejo to the mix and it's slightly different

K/E leads 50-45 among LIKELY voters in the 3-way race N/C gets 2

K/E leads 50-42 among REGISTERED voters N/C gets 4.


How's does Gallup determine LIKELY voters? Basically by asking if the voter has voted in the past and how certain they are of voting this fall. Its a SCREENING process. How accurate? it's open to debate (which we'll get to in NC)

However, it's clear that the higher the turnout this fall the better it is for Kerry/Edwards as REGISTERED voters more likely to vote K/E than Gallup's perceived LIKELY voters.

A second oddity is that this is the only poll I've seen where Nader actually slightly nicks GeeW instead of Kerry (though small numbers are always open to question)

Now, the NC Gallup poll

http://www.usatoday.com/news/polls/tables/live/2004-07-...

For brevity sake, I'll just look at the 3-way breakdown

Among LIKELY voters Bush is ahead 54-39 (Nader 4)

Among REGISTERED voters Bush is ahead 49-43 (Nader 4)

So, in NC there's a 9 POINT swing between LIKELY and REGISTERED voters while there's only a 3 point swing among National voters?

This should have run up a big red flag at Gallup (which probably explains why they included the 49-43 number in the USA Today story when they normal don't mention such things).

Part of the problem may be the turnout formula Gallup apparently used which says "based on past voting history in North Carolina turnout is assumed at 50%"

The problem with this?

Turnout for the last two Presidential elections in NC have been 58.9% and 58.8%

Available in PDF here

http://www.sboe.state.nc.us/index_data.html


Bottom line? Gallup admits they don't have experience polling in NC and it showed in this poll. They're adjusting of registered voters into likely voters looks highly suspect at best, incompetent at worst (we'll give them the benefit of the doubt and not accuse them of being malicious)

Second Bottom line? It's not just important to register people to vote, it's important to get them out and vote. Higher turnout=Better results for Kerry/Edwards

Thank you for wading thru this.

[MallRat]

1. Good job on the NC numbers.

Now it makes sense. Bush by 6 sounds closer to the truth than 14.

-MR

[Media_Lies_Daily]

2. Excellent post...about what I suspected, Thanks for the work you did.

[papau]

3. Excellent analysis - thanks for posting!

:-)

[NewYorkerfromMass]

4. You rock! Thanks! nt

[blm]

5. Good job. Thanks.

.