Mike points out at The Stump that a Clemson poll putting Barack Obama only two points ahead of Hillary in South Carolina has been smacked down by new evidence pegging Clinton to win by 23 points. Yowch.
Of course, this instance underscores the inconsistency of polling numbers, and the need for averages (RCP is now a full-on obsession), but further, it brings to mind a massive gap in even the most precise polling methodology: cellular phone users.
I haven't had a land line in four years. No, five. Sure, I don't live in Iowa (none of my resident states, past and present, has any hope of swaying the election), but I am as rabidly abreast of the issues as any--polled--voter. Why is no one calling me? In this day and age, how many more cell-phone-only voters are excluded from the conversation?
If someone knows more about whether mobile phone numbers are included in local polls, do tell. For now it seems obvious that a large sector of the electorate (probably young, but with a decent chunk at voting age) is being left out of the poll-driven national discussion. How to account for this? And, more importantly, which candidate can expect an unexpected surge among these unpolled masses?
Update: Noam points me to a Pew study that suggests the 'cell-only' public doesn't significantly alter poll results. I looked a little closer, and that's true, except for the issue of gay marriage--a metric with clear generational correlation. Mobile users approved at a margin of 51-37 over non-cell users(though the landliners carried the day; their 37% was the reported figure). If the exclusion of mobile users excludes all equally, it should have no effect on election-day results. But if an old/young divide in the electorate tracks onto a cell/no-cell divide, the candidate doing better among youth can expect a bump. Call me!
--Dayo Olopade