You wouldn't think we would mention the US elections on a site devoted to mobile devices, but there's an interesting and fast-growing phenomenon known as the "cellphone effect" which is potentially throwing polls off by several percentage points, reported over on the excellent fivethirtyeight.com. Essentially it boils down to this: some polling companies have a tendency to only interview people through their landline phones, so anyone without a landline simply isn't represented in their surveys. That means politicians with strong support among mobile-only households would do much better on election days than landline-only polls predict.