Despite ourselves, it can be hard not to get elated or depressed over movements in Rasmussen's tracking poll, since it's the only one to provide a daily fix. Along with others, I have more than once cautioned against taking it to have any signal. Specifically, for weeks on end, both candidates had been within 3 points of 45%, which means that their individual movements were within the margin of error. But the conclusion that it is all just sampling noise turns out to have been incorrect, as the plot below reveals.
In this figure, the jagged curve gives the daily difference between Kerry's and Bush's percentages (positive scores indicate Kerry ahead) with a seven-day running average applied (this is on top of the three-day average Rasmussen uses to produce each day's data points). This is for the period from February 3 until now. The regression line, that is, the best fitting linear trend, is shown in blue.
Clearly there is a steady trend over this period. The slope of the curve is .013%. That is, on average, Kerry has gained roughly an eightieth of a percentage point on Bush per day.
Of course, all the usual caveats regarding statistical analyses apply. Nevertheless, it was informative to me to find that there is support for a systematic trend in the numbers to date. And the trend that is there is entirely consistent with the methodical but relentless approach that has seemed to characterize Kerry's gameplan, often to the frustration of those of us who are less prone to be patient.
Update [2004-7-22 13:14:33 by nedog]: Corrected to say that a seven-day running average was used on the diferences rather than a five-day average as originally stated.