I dislike trackers because their rolling average of past results (past 3 days) hides the larger variations in daily results, making it harder to know what's really going on.
Clark's recent move in the ARG New Hampshire tracker was a real trend, but the campaigns knew about it before we did because they were looking at daily polling results.
Other shifts in single day polling results turn out to be random error or bad polling that the three-day rolling average turns into a "trend." This happened several times in Gallup's Gore-Bush 2000 tracking, and each time the media bought into the nonexistent trend.
Rasmussen's tracker now contains either a day of bad data or a real change. Clark apparently equalled or edged out Dean in their 1/12 daily interviews -- that's the only way to explain the change from the 1/11 to 1/12 tracking results. Look for the 1/13 and 1/14 tracking to report further tightening in the Dean-Clark race.
1/9 1/10 1/11 1/12
Dean 25% 24% 24% 21%
Clark 15 16 16 17
Edwards 6 7 9 9
Lieberman 13 10 9 8
Gephardt 11 10 7 8
Kerry 10 9 8 8
Is 1/12 a real change? It's possible, given the continuing barrage of attacks on Dean. It could also be random; I have less confidence in tracking results from Friday, Saturday and Sunday interviewing than from weeknights, when there is more consistency in the types of people who are found at home. Politus and fladem have also questioned the reliability of Rasmussen's polling. They report that he is a right-wing-nut whose 2000 polls favored Bush.
I report, you decide.