« Blue Swing State Polls | Main | WI, MI, AZ Residents: Did You Ever Vote GOP? »
Tuesday, September 21, 2004
Red Swing State Polls
Posted by DavidNYCMason-Dixon also did a companion set of polls for six red states: MO, AZ, NH, OH, WV & NV. The results are available here. All six show Bush with a lead, and all but one show Bush with a lower unfavorability rating than Kerry (West Virginia is the exception).
But here's something that's really very odd. In the blue state polls I posted yesterday, Bush not only led in unfavorables in every state, but his rating was no higher than 13 points in all but one state (New Mexico was the outlier). Why is that bizarre? Because in the red states, his lowest unfave rating was 37%.
Now that I look at it a bit more closely, I'm inclined to think that MSNBC fucked up and swapped Bush's neutral and unfavorable ratings for the blue state polls. I'm wondering what else might be wrong - like Mason-Dixon's likely voter model. I know in the past Chris has commented that M-D has a GOP bias, but I haven't seen as clear evidence of this as with Gallup. Anyone care to chime in?
Posted at 02:01 PM in General | Technorati
Comments
You are absolutely correct -- MSNBC has clearly transposed the labels for Bush's "unfavorable" and "neutral" in all blue state polls except New Mexico.
I was so utterly baffled when I first looked at those poll results yesterday...
Posted by: globecanvas at September 21, 2004 02:47 PM | Permalink | Edit Comment | Delete Comment
You're looking for bias in Mason-Dixon's data?
Well, have a look on Mason-Dixon's data for Oregon then, which they state to have collected during September 14th through 16th. They claim a Bush vs. Kerry result of 47:43. (Which is quite astonishing for Oregon, actually...)
Research 2000 has collected data in Oregon starting just one day earlier, ending the same day, and they call 44:51 in favor of Kerry...
Although I'm really pesimistic about Kerry's chances lately, I don't think anything special happened on September 13th in Oregon that could explain a Kerry-loss of 8 %... :-/
On the other hand: Survey USA just released a Maryland poll showing Bush and Kerry tied at 48 % both.
I guess, quite a lot is possible at the moment. We just can't tell any more.
Posted by: Bornheimer at September 21, 2004 06:09 PM | Permalink | Edit Comment | Delete Comment
I've been watching the polls for about six months. Has anyone checked out the WSJ Battleground State Poll (Using Zogby). It's a totally different picture (one that I like, anyway) WTF Mate? (By the way, my family voted absentee yesterday -- three Tennessee votes for Kerry)
http://online.wsj.com/public/resources/documents/info-battleground04-frameset.html
Posted by: Ed Wagner at September 21, 2004 07:14 PM | Permalink | Edit Comment | Delete Comment
Oh, this is simple. If you like the results, they are unbiased. If you don't like the results, it must be those dirty, corrupt, money-making pollsters. It's very simple, really.
If you want to be serious about it, then start with the "likely voter" models. They are only accurate in hindsite, and with the voter registration waves taking place, some being rejected etc., who knows? Not one person on earth. Maybe the guess is good, maybe not. In an election as close as this one it could influence the outcome. Oh well.
Posted by: Mark Olsen at September 22, 2004 12:46 PM | Permalink | Edit Comment | Delete Comment