Two new Survey USA polls...

[millwx]

VA:  Bush 50%  Kerry 45%
http://www.surveyusa.com/2004_Elections/VA040709president.pdf

MO:  Bush 48%  Kerry 46%
http://www.surveyusa.com/2004_Elections/MO040709president.pdf

*Updated for links

[Beet]

The first thing I thought was... damn should've picked Gep!!! ... but that's not true.

[ATFFL]

The first thing I thought was... damn should've picked Gep!!! ... but that's not true.

With Gep you can add 3% to Bush, likely in both polls.

Seriously, probably not in MO, but it is probably the same.

[Filuwaúrdjan]

The VA poll is interesting... weird choice of geographical regions tho'... what happend to SW VA?

[AuH2O]

That poll has Bush with 18% of the black vote,  46!% of the hispanic vote, and 29% of the asian vote. He should do better than 29% with the last, but if something like those numbers occured he win would by a lot more than 5%.

Plus, yeah, where is the SW? I don't see how Roanoke can be considered central VA (living here my whole life, that means Richmond area), let alone Danville, Blacksburg, and the Marion-Galax area. Leaving those out would explain Bush's small lead overall and 49-43% lead among rural voters.

Survey USA sucks...

[Filuwaúrdjan]

Actually missing out SW VA (eg: Buchanan county) would probably dent *Kerry's* numbers... unless they can't read maps and added it all into Central VA...
I'll agree that SUSA sucks...

[millwx]

Actually missing out SW VA (eg: Buchanan county) would probably dent *Kerry's* numbers... unless they can't read maps and added it all into Central VA...

I would presume that Survey USA didn't purposefully leave out SW VA.  My guess is that they're in there, but the sample from that rural region is so low that they can't, with any reliability whatsoever, report the results in SW VA separately.  That sample is likely included in the "rural" sample, but is simply not broken out separately.  So, I don't think anyone benefits from this region being "missing"... because it probably isn't "missing"... it's just not reported separately.

P.S.  I wouldn't go so far as to say Survey USA "sucks", but I agree, they do seem to be occasionally a bit "off".

[AuH2O]

It would depend on where they missed... the extreme Western part (Buchanan, Dickenson; Wise was a tie) went Gore, but most the others were comfortable Bush. Tazewell, Lee, Giles, etc., not to mention Roanoke area.

Regardless, if a poll has these kinds of questions, it's effectively worthless. There are enough variables to deal with already...

[classical liberal]

Both of these ar within the MoE of the POS polls last month.  A 3 point boost from Edwards in each state is not out of the question.

[AuH2O]

Why would Edwards affect VA? It's not like any part of VA is in a NC media market, besides which even North Carolinians don't like him that much...

Any Edwards boost would be a national phenomenon resulting from positive attention on the Dem ticket, not some sort of geographic appreciation...

[ATFFL]

Why would Edwards affect VA? It's not like any part of VA is in a NC media market, besides which even North Carolinians don't like him that much...

Any Edwards boost would be a national phenomenon resulting from positive attention on the Dem ticket, not some sort of geographic appreciation...

There are a few counties in south central VA that are hit by the Raleigh stations, but not enought o make a difference.

[The Vorlon]

Survey USA sucks...

[Beet]

Think they might have adjusted their methodology after consistently overpolling Democrats in '02? Are they the only firm that overpolled Democrats in '02? It seems like every firm that is substantially off makes that mistake.

[The Vorlon]

Think they might have adjusted their methodology after consistently overpolling Democrats in '02? Are they the only firm that overpolled Democrats in '02? It seems like every firm that is substantially off makes that mistake.

Actually, Survey USA is pretty darn close to "down the middle" - they do miss a fair bit here and there, but they are no mor elikely to miss left as they are to miss right.

I generally cut SUSA some slack - they use Robot calls as per Rasmussen, which I structurally have some doubts about, but on Balance a pretty decent firm.

When I poll screws up, it usually misses left a lot more oftent han it misses right.

This is due to the reponse rates of Republicans and Democrats.

Republicans tend to have higher incomes, and hence are away from home more, and also (increasingly) have things like call block / call display which tends to make it harder to contact GOP voters.

The good firms properly adjust for this, the bad firms do not.

There are, IMHO only three notable polling outfits that DELIBERATELY skew their results:

LA Times
CBS News
Minnesota Star Tribune

Survey USA misses the mark, but mostly in a random way

[AuH2O]

Was the methodology for those final polls the same as the ones discussed on this thread?

Maybe I shouldn't say they 'suck,' but that VA poll sure did...

[The Vorlon]

Was the methodology for those final polls the same as the ones discussed on this thread?

Maybe I shouldn't say they 'suck,' but that VA poll sure did...

As far as I know SUSA in 2004 is identical to SUSA 2004 in terms of methods.