|
Good question. If you've been following these tracking polls, you've probably noticed that Gallup's tracking poll (not to be confused with its USA Today poll) generally (no, not always) shows a wider gap between the candidates in Obama's favor than does Rasmussen's tracking poll. Here's why. Gallup's tracking poll measures REGISTERED VOTERS. Rasmussen's measures LIKELY VOTERS.
Wait, you say, but doesn't that mean Rasmussen's tracking poll is more accurate?
No. There's no evidence a poll of LIKELY VOTERS is any more accurate than a poll of REGISTERED VOTERS when conducted more than a week before Election Day.
AND then there's the Obama Effect, which further complicates the picture. One of the screens to determine whether someone qualifies as a LIKELY VOTER and/or how to weight their poll response is how often they've voted in the past. As we all know, there's one candidate who is bringing millions of NEW (thus UNlikely, by this criteria) registered voters to the table. They're being undercounted in polls of LIKELY VOTERS.
That's how Gallup's USA Today poll of LIKELY VOTERS and Rasmussen's tracking poll of LIKELY VOTERS massively under-represent Obama's strength while Gallup's daily tracking poll of REGISTERED VOTERS and other polls of REGISTERED VOTERS reflect more support for him. Obama's REGISTERED VOTERS, as we've seen by his ability to capture the first-time vote, are LIKELY to vote for him, but aren't counted as such.
My money is one the registered voter model for its predictability--not for every election, but certainly for this one. And, by the way, there's a chance that that model might even under-represent Obama's strength. He'll be continuing to register hundreds of thousand of more supporters (who aren't being included in anyone's poll results at the moment), and McCain won't be.
|