Following up on my award winning diary on Zogby's internals yesterday, I know apply my vaunted intellect to his Iowa tracking results
today. (Originally posted in comments)
That ARG polls 200 people a day, rather than 167. Their poll is considerably more stable, but that may just be NH.
I ran the numbers a bit in a diary. It's pretty obvious that Sunday was a Dean "high", Monday we know was 18, and Tuesday was somewhat above 18.
Best guess? Sunday's numbers were at least 33, and probably a bit higher. (The Sat/Sun average was 33, and the tracking poll acts like Sundays number was higher than Saturdays).
Assuming 33, however, means Tuesday's number was 21 (33 + 18 +21). If it was higher, than Tuesday's number was lower.
Anyways, now that Sunday's been removed, we have 18 + 21? + x. Since the average is 21, we're looking at Wednesday's number to be 24, or close to it. (Zogby implies this, by stating "Dean is not in a free fall").
Dean's "drop", like Kerry's rise, seems to be entirely an artifact of Monday's polling.
I don't think Iowa's race is this fluid. I think Zogby's daily sample is too low, and thus the spikes are magnified.
I also think, perhaps less charitably, that Zogby's comments are geared less towards informing the electorate, and more towards hyping the race to sell more polls.
Tracking polls are often volatile, and Zogby's daily sample is only 167...I wonder if that's why his tracking poll is more volatile than ARG's NH tracking poll, with it's daily sample of 200?
