In a past article I know that I've referred to and reposted a survey posted by the Daily Kos from a group called "Research 2000". The Daily Kos has called the the polls from Research 2000 into question based on additional statistical research and comparison to other polling data. Research 2000 had enjoyed a good reputation in political circles for polling data, but Markos Moulitsas points out that Research 2000 even when given the opportunity to validate their polling data has refused to do so, and therefore he doesn't believe he can trust any of their data.
I know that I've posted Research 2000's poll concerning Republican voter. I will, in light of this, retract my trust in that polling data.
Here is more on the issue from the Daily Kos: