In what could be a huge problem for pollsters not only this year, but in all future elections, Gallup has found a statistically significant difference in support for Obama between cell phone only households and all other households. Pollster.com has the scoop:
While the inclusion of cell phone only households makes little difference in the Clinton-McCain contest, it benefits Obama by a net four points: Without cell phone interviews, and weighted using Gallup's usual likely voter model, McCain would get 49% to Obama's 46% (clarification: this result combines six Gallup/USAToday surveys conducted so far during 2008). With the cell-phone interviews included, the result is Obama 48%, McCain 47%.
Given that cell phone only households are only 16% of the population, a 4% shift toward Obama when those voters are included indicates a wide gap between cell-phone only households and all other households in terms of Obama vs. McCain preference. Keep in mind that this was tracked across six Gallup surveys, large sample sizes were available for the subsets in this study.
Tempting as it may be, I don't want to yet conclude that this means Obama is much further ahead of McCain than current polling indicates. Four years ago, I banked on the so-called Incumbent Rule, which holds that challengers receive the vast majority of late deciders in elections against incumbents, to make a similar pro-Kerry projection, and it fell flat. Once bitten, twice shy, I guess. Still, this is hopeful news, given that Obama already leads McCain anyway.