A new study by Pew Research says that cell phone only households are divergent enough in their political views to screw up most polls. While a few pollsters do include cell phone only households, most don't because it's too expensive. This contradicts earlier studies that claimed a good poll could ignore people without landlines. Thus, Pew figures Obama's support is under-reported by 2-3% consistently.
If you want to read the Pew release on this, you'll find it here
http://pewresearch.org/...
The basics of the piece are summed up in this paragraph
The Pew Research Center for the People & the Press has conducted three major election surveys with both cell phone and landline samples since the conclusion of the primaries. In each of the surveys, there were only small, and not statistically significant, differences between presidential horserace estimates based on the combined interviews and estimates based on the landline surveys only. Yet a virtually identical pattern is seen across all three surveys: In each case, including cell phone interviews resulted in slightly more support for Obama and slightly less for McCain, a consistent difference of two-to-three points in the margin.
And the conclusion is pretty telling
Even though the omission of cell phones from election polls does not currently make a large difference in the substantive results, Pew's surveys this year suggest at least the possibility of a small bias in landline surveys. Such a bias could be consequential in an election that appears to be very competitive right now, especially if significant numbers of young people turn out to vote. In particular, the research suggests that estimates of the candidate preferences of young voters may be biased if cell phone interviewing is not included in the survey.
These problems are all the more pressing as the number of Americans who are reachable only by cell phones increases. U.S. government surveys estimated that about 15% of adults were "cell only" in the fall of 2007 and the rate of increase since 2004 has been at least 2% a year, meaning that the number may be as high as 17% in this election cycle.