Talk Elections

Lets see, some new survey usa presidential polls came out later today.  Lets chew on em

Oklahoma

Bush-58%
Kerry-37%

No surprise here

South Carolina
Bush- 52%
Kerry 41%

A bit better than last time for Bush

California
Kerry 49%
Bush 46%

Seems a little too close for me.

California is obviously out of whack, but the other two appear to be right on.

It would be hilarious if Bush lost state after state and then somehow managed to win with California.

Yeah, Kerry will win California by at least 10%.  Unlike most Republicans, I think Bush only has a chance of winning maybe two states he didn't win in 2000.  Iowa and New Mexico.  Bush will lose New Hampshire and the only way i see him winning is to hold onto the rest of his 2000 states.  

I obviously agree with the others that Bush is not 3 points behind Kerry in CA. After the GOP convention he might be about 6 or 7 points behind but, 3 points? No way.

Party ID in CA poll was:

GOP = 40%
Dem = 37%
Ind = 22%

2000 Exit polls in Ca:


Dem = 43%
GOP = 35%
Ind = 22%

SUSA correctly and honestly flagged this. (Note to LA Times, this is how intellectually honest polling organization behave)

This is the 1 in 20 polls that blows up :)

Is SUSA a decent "Vorlorn Approved" firm, in general???

I have them as a "second tier" firm.

They are honest, not biased, and doing the best they can in a good faith way.  I cut these guys some slack. When something "blows up" (ie this CA poll) they put a big note in the .pdf and say... "Hey.. this sample blew up"

(Note to LA times.....)

B to B+ overall.

Not a Mason-Dixon or a POS, but a clear step above say an ARG. Roughly on par with say Research 2000.

They crank out a few more "blips" but only because they just do so damn many polls.. :)

What I like about them is the publish ALL the data so you can go in and see how a sample looks like and what went wrong if it blew up.

Even in a "bad" sample (like this CA poll) you can still extract useful information.

This is a pretty good firm overall.

Well, if the polling organization knows it's one of those twenty that's innaccurate, what's the point in publishing obviously flawed material?   I can see cases where a polling firm could simply oversample Democrats who are voting for Bush or something, but when something clearly doesn't match the cold, hard facts, I'd be tempted to throw it out.

Frequently they're under contract to a media outlet which won't give them the time (or allow them to) clean up the survey (weight, resurvey, etc.).

Most of the public is unaware of the tension between getting it now/on budget versus getting it right but a little bit latter and perhaps with a little more expense.

What I like about SUSA is the demographic and regional stuff they have in their PDF's... sure the MoE's are very high, but it's interesting to see that in NC (as an example), Bush has a huge lead in the suburbs, while the rural areas are very close.

Public Opinion Strategies just did the battleground states (Aug 16-18) and their Iowa sample "blew up".

They didn't include the results in the presentation and said.. "hey we're doing it again... we will get Iowa to you next week..."

The media, on the other hand wants the poll NOW.

Dare I be cynical, but they LIKE a bad poll.  "California a Battleground State !!?!!" is a better headline than "Huge Kerry lead, race is effectively over"

Where do you find the POS polls, Vorlon? I can't see them on their website. Another firm's polls I can't find is Hart/Teeter (NBC).

reweight to 2000 turnout puts Kerry +12 in CA