I actually posted this as a comment but I felt the need to do a short diary on the subject. There is a fair bit of "Edwards Up!" "Edwards down!" going on due to the release of two conflicting poll numbers - one from InsiderAdvantage showing Edwards leading the field in Iowa by a 30-27-24 lead over Obama and Clinton, respectively. Another poll was released from ABC News showing Edwards lagging behind at 20%, with Obama leading the field in Iowa with 33% and Clinton a close second at 29%. So who's more reliable? Which one is the most recent snapshot? Which one can you trust?
First of all, you can't trust any damn poll right now. Iowa is fluid right up until caucus day. If you are actively working on a campaign, you have no time to bask in the glory of your candidate's lead (mine's Edwards), or sob at your candidate's fall. Each one of those polls can turn upside down, inside out, and every other way you can imagine before - or even on - caucus day.
Second, I want to provide a short analysis of the two polls. They are being released at the same time, but there are significant differences in their methods and timelines:
Timelines and Snapshot
The ABC News poll was conducted over 5 days, Dec. 13-17, whereas the InsiderAdvantage poll was conducted over only Sunday and Monday (the 16th and 17th). Under normal circumstances, the longer period would add to the degree of confidence, but given the rate at which Iowans change their minds, those ABC polled on the 13th, 14th and 15th may well have changed their minds by the 17th. Given Iowa's situation, InsiderAdvantage has a better chance of getting it right, simply because it started later.
Sample Size and Time
The ABC News poll surveyed 656 likely caucus goers over 5 days, whereas InsiderAdvantage surveyed 977 over 2 days. Any statistician will tell you this makes the InsiderAdvantage data more reliable.
Margins of Error
InsiderAdvange has an MOE of +/- 3% and ABC News has an MOE of +/- 4% (that may not seem like much, but ABC News has admittedly a 33% higher error possibility than InsiderAdvantage).
Polling is not a perfect science. All it gives you is a snapshot. You should treat it as a snapshot. And in Iowa's case, the picture is changing by the day. But comparing these two snapshots, in my judgment, the InsiderAdvantage poll is a better snapshot of more recent activity. That means that Edwards' surge is more likely real than not. And that is great news for those of us who support Edwards. Because I am an Edwards supporter, the conclusion above is convenient for me. But I assure you that it was not a foregone one. I looked at both surveys, the data, the timeline, and made a judgment about which one is best. It happens to be a convenient truth for me, but it is the truth.