Let's do ourselves a favor and just agree to ignore Rasmussen. His poll is lousy. So is Survey USA's polling. The reason for this is that they are done by an automated caller. Respondents do not speak to a real person. This skews results because people are more inclined to hang up. More hang ups mean that the ultimate sample that they take is not respresentative of the national public.
A common sense test of this hypothesis that something smells fishy with Rasmussen can be seen by the fact that his is the only poll whose results have been consistently within the margin of error. In other words, Bush and Kerry each have always ranged from about 44 to 50. In other words, the difference between any given Rasmussen poll with any other given Rasmussen poll can be explained entirely by statistical differences. Rasmussen's poll indicates that there has been no statistically noticeable movement in the electorate in nearly nine months. His is the only poll where this is the case. Anybody believe that?
Nice observation