Today, I finished up the final of the six panels I was on with an excellent session on polling featuring Tom Matzie (of MoveOn.org), Stan Greenberg (of Democracy Corps and GQR), Mark Blumenthal (Mystery Pollster of Pollster.com) and Joel Wright (of Wright Consulting, with whom MyDD produced the Democratic Congressional Challenger's Strategy Memo last year). One of my favorite parts of the panel was when Mark put up two graphs from his website that have been receiving a lot of play lately, and then kept rapidly flipping back and forth between them to show how they match up.
Check it out for yourself. First, here are the trends in national Democratic primary polls in 2003-2004:
Second, here is a chart on how closely people pay attention to campaigns, graphed according to how far away the election actually is:
The key point here is how these charts match up. Specifically, the rapid change at the end of the 2004 Democratic primary campaign occurred at the same moment when people began to pay far more attention to the campaign. Smaller changes that occurred before Iowa also corresponded with major media moment in the campaign, such as Clark's entry into the race and Gore's endorsement of Dean. The point here, which should have been obvious to me all along, is that the campaign won't really change much until the level of coverage of the campaign changes. Only major events that receive truly massive amounts of news coverage have any possibility to alter the shape of the campaign in a statistically significant manner.
The point here is this: don't expect any long-term, gradual improvement for any candidate. National changes in campaigns like this will happen only in large chunks, and as the result of major events. Otherwise, expect the campaign to stay pretty much as it is, and pretty much the way it has been for the past four months, until such an event takes place. . Expect small, weekly changes away from the status quo to reverse themselves in only a week or two. Basically, unless something major happens, the horserace isn't going anywhere for a while.