I don't believe these polls Steve Kornacki is talking about.

[karynnj]

11. I saw his coverage of the polls

I think that there is a real problem with polling that goes beyond landline vs cell phones. Many serious articles have explained that the pollsters are lucky if they got even 5% of people in their sample to respond - the articles referred to response rates of 2 to 3 percent. Years ago when repeat calling got response rates well over 50%, pollsters made the assumption that those who respond in a given demographic category are similar to those who don't. Obviously, this is a necessary assumption to make, but given the VERY low response rate, that assumption becomes harder to accept. In addition, the pollsters need to have population estimates for the demographic cells -- including Democratic, Republican, Independent. If fewer people now identify as Republican are fewer than those "population" estimates, estimates will be distorted. (I suspect that former Republicans would now identify as Independent - making the independent category more conservative.)