I don't really understand the value of so called "tracking polls." These polls take three days worth of polling, average them together, and give you a view of what the electorate is thinking over time. It's a fine idea of in theory, but in practice, these polls are like glaciers, they are very slow moving and don't respond quickly to jolts in the system, like a 500-point drop in the Dow or a contentious debate among Democrats. Take John Zogby's current South Carolina tracking poll as a perfect example. By using a three-day weighted sample, and by using one day which was before the Democrats debate, which has, as proved by Zogby's numbers from days following the debate, shaken up the race, just looking at the totals Zogby reports doesn't tell you nearly the whole story. In the past two days, John Edwards has seen a surge in South Carolina and in fact has, according to Zogby, overtaken Hillary Clinton for second place in the final day (Wednesday) of Zogby's polling. But, because the polling includes Monday and Tuesday in his sample, the number reported by the mainstream media is the three-day sample number, which shows Edwards in third place, trailing Clinton by five-points for second place behind Barack Obama. Let's take a closer look at the numbers.
In the first report of the poll, which was taken from Sunday to Tuesday, and thus only accounted for one day of polling after Monday's furious and game-changing debate, Obama lead Clinton and Edwards 43-25-15. From Monday-Wednesday, including two days of post debate coverage, Obama fell 4 points, Clinton 1, and Edwards gained 4. A huge jump, but in fact, the leap is even larger. Look at Zogby's report about just looking at Wednesday's total:
Edwards, meanwhile, has had his second good day since the Monday night CNN debate, in which he delivered a strong performance. He hit 19% support on Tuesday alone and then 27% support on Wednesday alone. And, on Wednesday alone, he pulled ahead of Clinton overall. He has pulled ahead among whites. Could he pull ahead of Clinton and finish in second place? Even with a strong showing here, where does he go next to take advantage of the momentum?
So, Edwards is really at 27% as of Wednesday, but because the poll counts a day even before Monday's debate, the reported number has him at only 19%. This is completely missing Edwards surge, and the collapse of both Clinton and Obama (though Obama remains in first place). Hillary Clinton may finish in third place, her second such finish of the campaign. And Edwards may get his second "silver medal" as Mitt Romney would say. And while this may come to a shock to the national media, who may say they never saw this coming, it's entirely predictable based on Zogby's poll.
In this era of 24-hour-news-cycles and when events like Monday's debate or Tuesday's stock-market mini-crash (and later recovery) have such a profound effect on people's views and opinion is so fluid that what people think on Monday may not be what they think Monday night, much less Wednesday, I think we need to ask how helpful these tracking polls are. We all asked how the polls could be so wrong in New Hampshire when Hillary Clinton seemingly made up a double-digit deficit overnight. Well, by using tracking polls, even if Clinton's comeback began two days earlier, it wouldn't have really registered to the amount it was in reality. We get fooled by not looking at the internal numbers of these tracking polls and thinking that a huge jump by Edwards, like over 15 points in two days, isn't meaningful before 48-hours before the primary.
We need more day-to-day polls, those are the numbers which need to be released and reported upon. They are far less likely to miss these moves in public opinion and tell you where people stand today, not yesterday.