While most polls show Obama with a lead over McCain, Rasmussen Reports has for some time suggested that McCain is leading in several key states as well as I think overall leading nationally.
Although it's possible Rasmussen is on to something other pollsters are missing, common sense suggests that his results - out of the statistical probability ranges for the other polls - are the result of a polling error.
I want to see their methodology but I'm wondering if the cell phone factor is an issue. Young voters are more likely to support Obama but also *far* less likely to be polled by traditional means, which often exclude cell phone calls. I'm assuming some polls are trying to factor in this complex cell phone variable but assume that factor may create a lot of variability.