Here's the Edwards campaigns' internal take on Register poll, which shows them in third.
Is the poll accurate? There are good reasons to think it is NOT.
-The poll was conducted during the holiday AND over the weekend. There is plenty of evidence that either of these would make it more difficult to obtain a representative sample. The combination makes the problem of obtaining a valid sample GEOMETICALLY [sic] worse.
-The poll is at odds with history. The poll says that 60% of Democratic caucus participants will be first timers. Usually, the number of first-time caucus goers is no more than 20%.
-The poll also says that 45% of those at Democratic caucuses will be Independents or Republicans.
The poll is at odds with other polls. Other polls show a close race with other candidates leading.
-Yepsen himself highlights the fluidity of the results rather than the horserace.
What does the poll really say-The poll says the race is close. With a margin of error of +/-5, any of the top 3 Democrats could be in any of the top 3 positions.
-The poll says the race is yet to be decided. 34% say they could change their minds, and 6% do not express a preference. This means 40% are yet to decide.
-As Yepsen points out, 21% of those in the 2004 entrance poll [said they had made their decisions in the last 3 days -- something this poll cannot capture.
Marc Ambinder is a contributing editor at The Atlantic. He is also a senior contributor at Defense One, a contributing editor at GQ, and a regular contributor at The Week.