People are buzzing about Barack Obama supposedly surging among Iowa Democrats. A Newsweek poll has Obama besting Clinton by four points and Edwards by six, but like most early polls, this is pretty worthless.
First, the poll asks people who they would back if the January caucus were held today - a huge if. The more accurate query is whether Iowans are backing a candidate yet. Large pluralities continue to say they're undecided, according to campaign sources, which is the norm for the Iowa Caucus.
Second, it's very hard to accurately poll the small, unpredictable "universe" of caucus attendees. This poll has a high margin of error at plus or minus seven points, (explained here), and it purports to identify "likely" caucus attendees by asking if they plan to caucus. But this is not like planning to vote. It takes over an hour to attend the caucus -- smack in the middle of Iowa winter -- so schedules and weather make turnout much more unpredictable than regular voting.
Third, no Iowa poll incorporates the defining feature of the Iowa Caucus: The required "viability" minimum of 15% that candidates must reach to have any votes count in a precinct. On caucus night, Iowans first gather into groups for their preferred candidate, and then any groups falling short of 15% must change to support a viable candidate. The final caucus tally refers to those second round totals. That means victory can depend on the second-place preferences of Iowans backing "second-tier" candidates. Field operatives get this dynamic. While the media chases worthless national polls and sloppy slices of state electorates, veteran strategists like Michael Whouley are tracking Iowans' second-place preferences. (Whouley is famous in Democratic circles for his field work for Bill Clinton, Al Gore and John Kerry, but he's likely to become a household name soon, thanks to Denis Leary's portrayal in HBO's forthcoming movie about the Florida recount.)
Finally, it is just too early to deduce anything meaningful from national or state polls. This time last cycle, Wesley Clark led national polls, as the Article 19 blog notes, while Democrats and pundits debated who were the "top three" in Iowa. There was the seemingly inevitable frontrunner, and that populist with deep roots in the state, and the new yet inexperienced candidate who wanted to bring America together. Sound familiar?
Now Hillary has Dean's frontrunner spot, Obama is the old Edwards, and the new Edwards is doing the Gephardt thing. The top three of 2004 had one thing in common: high poll numbers and losing campaigns.