3 Daily totals released (Obama is doing quite well in Gallup Tracking)

[cthulu2016]

Gallup seldom releases individual daily numbers from their track polls, but we know from the WP that that Gallup Daily Tracking poll raw data for Oct 4,5 and 6 was 47-47 each day. (The three days after the debate.)
http://www.washingtonpost.com/blogs/election-2012/wp/2012/10/08/obama-loses-five-point-edge-in-gallup-poll/

Today's 7-day average is 50-45, covering Oct1-Oct7.

We know 4, 5 and 6 are each 47-47.

So the four days October 1, 2, 3 and 7 Obama averaged 52.25.

The tracking poll that included September 30-October 6 was 49-46. So the four days September 30, October 1, 2 and 3 Obama averaged 50.5.

The only day different there is Sept. 30 versus October 7, so Obama probably had a large day in Sunday polling. If I got this right, Oct 7 has to be 7 points higher than Sept 30. It could be a great 7th or a horrible 30th rolling off the average, but since Obama was doing well the first few days of October there probably wasn't a horrible day in the mix. (Someone good with algebra is welcome to find those two values, to the degree they can be calculated.)

Probably all a combination of jobs report and debate fade.

[reformist2]

1. I love this analysis.... I thought I was the only one trying to back out the dailies! :)

[cthulu2016]

2. It is so rare to get any daily sample data that

when I saw the Washington Post article I had to get my calculator out.

[reformist2]

3. I'm getting a Sunday number for Obama ranging from 54-60!

Here's my math: In order to get a 50 on the 7-day tracking poll, Obama needed at least a 49.5 average. So 49.5*7= 346.50.

346.50 - 3*47 - 3*50.5 = 54.

[woolldog]

5. Well done, sir.

[cthulu2016]

7. Is Oct 7th 7 points higher than Sept 30?

I get

30, 1, 2 3 average 52.25

1,2,3,7 average 50.5

So the difference between 30 and 7 have to be enough to move that 4-day average 1.75 points. 1.75 x 4 = 7

Seem right to 7ou?

[reformist2]

9. Because of rounding, it only needs to be 3.5 points higher...

...but by the same token, it could be as much as 10.5 points higher!

[cthulu2016]

11. hmmm

Presumably gallup uses raw data daily samples for each report, so there's no cumulative rounding error.

But yeah, I can see that.

[BumRushDaShow]

4. Yes, they spun with all they had this morning

comparing a 3-day to their more widely-reported 7-day to announce a "tie", and many on DU reacted in a fervor that all was lost.

And now, it's crickets from the pundits because the damage was done. How will they reconcile this? Are they thinking that by the time they try to start comparing the numbers apples to apples that somehow, MendaciousMitt will get a "bump" out of his rambling foreign policy speech that reversed many of the positions that he took during the primaries?

[DisabledAmerican]

6. You can't believe these polls! Doomsdayers said it's over Mitt Romney won

Stop the delusional these polls aren't true. Mitt Romney had a game changer moment even people on DU told us so that Obama lost the election because of the first debate.

[titaniumsalute]

8. One thing you have to remember...

The whole reason Gallup does a 7 day rolling average is to beef up sample size. The lower the sample size, the higher the standard error. That is why they don't typically release an individual day of polling. So the 52-45 might be a little out of whack but it certainly is in the right direction.

[woolldog]

10. Good point. nt

[dkf]

12. Or could it be a weekend effect?

I swear I remember a pattern like this from the last election but I can't remember which way it went if it was good or bad for Dems But I do remember personally telling myself not to be disappointed. Because we would rebound.