Nate Silver at fivethirtyeight.com has a really interesting piece today on the relative effect of including cell phone users in polling. He compares pollsters who have been including them against those that have not. There is a net 2.2 point difference in Obama's favor, and it's fairly dramatic in some places. Most interesting is the electoral college map that flashes between the two scenarios. I encourage you to take a look..

http://www.fivethirtyeight.com/