http://www.rasmussenreports.com/...
....and FALLING everywhere else (even in FOX polls) since the beginning of October?
http://www.pollingreport.com/...
Why is the question even worth asking?
Several reasons:
- Rasmussen has had a historically decent record in predicting election outcomes. It cannot be dismissed out of hand, even if it is not a statistically "pure" poll.
- The TREND is different in the past few weeks - its not just a bias for/against Bush
- Rasmussen (I believe) sets quotas for the number of Ds and Rs that it includes in its sample, based on past electoral turnout .
- Over the last few weeks, fundamental changes in Bush's support levels have begun to occur among Republicans, (probably starting with the Miers thing).
I am leading up to a rather obvious possible explanation for all of this, but one which I have not seen raised among the poll-minded here.
Hypothesis:
People are increasingly unwilling to be identified as Republicans in Rasmussen polls, and Rasmussen (being a rightwing outfit) is not telling us about it.
This would have the effect of making people who are still willing to be called Republicans more Bushite as a group, as pollees who used to be Bush supporters are increasingly calling themselves Independents and non-Bush supporters.
Thus Rasmussen is increasingly oversampling the dwindling number of "Republicans" in their poll, causing the net support for Bush in the final tally to RISE - even as the real support for Bush among the population falls.
Based on this, I predict that Rasmussen is going to go down flaming in the next nationwide election, as the built in instability in their polling method finally turns around to bite them in the ass.
I am the first to admit that this hypothesis is a bit simple minded, but I can't help but think that it is part of the explanation of the recent weirdnesss of Rasmussen polling