A reader writes in, wondering about what the tracking polls found yesterday afternoon and last night:

Looking at the 2004 tracking polls, note that around 10/28, Bush was leading about 50-45. However, on Halloween, Kerry surged to a tie, eventually leading in 3 polls.
I have been pushing that polls should be ignored on and after 10/31 unless there is significant movement in one direction or another, because Halloween is a time when parents are out, or married couples are otherwise occupied.

It’s a possibility, and makes the Zogby result from yesterday even more unusual — you would figure parents of small children, usually a somewhat GOP-leaning demographic, would be busy and not answering Zogby’s calls. On the other hand, would young voters be more likely to be out at parties, etc., last night? And since Halloween was on a Friday this year, did the tracking polls miss the high school football crowd as usual?

While I’m always pleased to see a poll showing the guy I prefer ahead, I noted not too long ago that we’ve seen quite a bit of back-and-forth movement in Zogby’s numbers. His numbers seem to be eye-opening, “hey-look-at-this-trend, the-race-is-changing!” every couple of days or so. But even if this is just another swing… there’s only three full days until Election Day. Zogby’s tracking numbers could, presumably, swing back to Obama over the next 72 hours. But even if you assumed bad faith and felt that he tweaked his numbers to generate the most dramatic result, one would presume that he wouldn’t want his final call to be too far off the final numbers we see on Election Night. Could Zogby say that he saw a swing of 10 points to Obama in the final 72 hours? I guess, but without some October November Surprise or trigger, people would react with typical skepticism.

In other words, even if Zogby’s 48 to 47 McCain lead is off, how far off can it be?