Got This From Intrade- Could Obama Have Been Up 56% -42% In Sunday's One Day Gallup Sample?

[regnaD kciN]

9. The problem with your calculations...

...is that they assume the pre-debate days were all equal, and that the three days following the debate were likewise all equal. There was probably quite a bit of variance within both sets of days.

That said, one thing you can extrapolate from tracking polls is the difference between the new day of data and the day that just dropped off the back end of the tracking period (in this case, the polling from last Sunday). The formula is, basically, the overall poll shift times the number of days in the tracker. Since the overall shift could be rounded to 2, that means it was somewhere between 1.5 and 2.49. Multiply that by the seven days of the tracking period, and you get the result that this Sunday's results were between 10.5 and 17.43 points better that last Sunday's.

Now, we don't know exactly what was in last Sunday's numbers -- they might have been poor for Obama. But here's another way of looking at it: pre-debate, it was O+5. Friday's release was still O+5, indicating that Romney still hadn't gotten much of a bounce post debate. Saturday's numbers, however, showed the lead shrinking to +3, which would indicate that Friday was a huge polling day for Romney (which is borne out in the other tracking poll shifts as well). After staying +3 on Sunday (Saturday's numbers) it went back to +5 today. That suggests that the numbers were roughly as favorable to Obama on Sunday as they were to Romney on Friday -- the big polling day for one canceled out by a big one for the other.