Democracy Corp provides some data on the difference in candidate preference among people with different telephone habits (cell only, mostly cell, mostly landline, landline only). Democrats made the same argument in 2004 to argue that the polls were overstating Bush’s support. That was probably true then, but not by enough to matter. Eight years later, a lot more people have given up their landlines. Has the phenomenon become statistically significant?
Personally, I think it’s more important that Obama is leading in the polls in enough swing states to win reelection. But, for people wondering who will win the popular vote, the cell phone issue might help them form an informed opinion.