I’ve been waiting to post something on the “cellphone problem” — the fact that almost all pollsters don’t include cellphone numbers in their sample, and that cellphone-only voters tend to be different from landline voters. But Mark Blumenthal has covered pretty much everything that I might want to say, and so I’ll direct you over there for the detail. It’s a real issue, and one reason that, even on the morning of the election, there is going to be a larger margin of error than the pollsters advertise.
, at