Why the new PA Zogby poll should be taken with a grain of salt.

[usregimechange]

The most recent total for the two-day tracking is Clinton 48% to Obama 42%. However, they later quote Zogby who discloses that the one day tracking for today (04/20) was Clinton 53% to Obama 38%.

Here is the quote:

Pollster John Zogby: “A big one-day of polling for Clinton. If a 10-point victory is the
pundit-driven threshold she needs on Tuesday, it looks like she can do it. This does not look like a
one-day anomaly – undecideds dropped to only 5% in this latest single day of polling, and they are
breaking Clinton’s way. As I suggested yesterday, if white and Catholic voters, who still are the
biggest portion of undecideds, actually vote, Clinton will have her double-digit victory. Just today
alone, she polled 53% to Obama’s 38%.

If the above is true please note that yesterday Zogby stated that Obama "had a good today at 46% to Clinton's 44%."

Zogbys contention, that undecideds are trending to Clinton, in no way accounts for his 17% one day swing in her favor. Today Zogby press release notes that "the undecideds dropped by two points." In other words, Zogby has Obama up by 2% yesterday, down by 15% today, with the undecideds decreasing by 2%! So much for his theory that undecideds trending to Clinton will create a Clinton landslide. Something else accounts for such the swing given that only 2% of undecideds decided in that 24 hours.

This is what Zogby should say: I need to stop discussing one day survey results in press releases designed to report our two day tracking because we average the two days for a good reason - to increase validity.

[adoraz]

1. Its correct, do the 2 day math

the poll from the other day has Obama by 2.

I just did the math, its correct.

[usregimechange]

2. The math is correct but the conclusion that undecideds account for the swing is not nor is the

inaccuracy of drawing conclusions based on one day results apropriate. Again, they average them for a reason.

[adoraz]

5. yeah I just checked, you are right

the 5% thing is wrong.

I don't trust these polls one bit. there is no way there is a 17 point difference in 1 DAY. Thats impossible when there isn't any reasoning for it...

[usregimechange]

6. They disclose the sampling error for the two day tracking, I wonder what the one day would be?

[adoraz]

7. my guess-

for what they have now, its 4.1%, which is pretty big.

For the single day, I would like to say 8.2%, but I am not sure if its that simple (since its half).

Even if its not that amount, its still a pretty big MoE.

I think people should really focus on the 2 day and stop freaking out about the 1 day thing. 6% isn't bad at all.

If something had happened which could have made this kind of impact, then I would see a reason for worrying, but the fact is that its impossible for a 17% swing in 1 day when there is no news or anything which could have caused it.

I will be very interested in seeing tomorrow's polling data.

And I also really wish they showed the complete cross tabs.

[usregimechange]

10. Incidentaly a MOE of 8% or so for the one day could be in keeping with the two day average

15% lead minus 8.5% = 6.5% lead.

48 - 42 = 6% lead

[Condem]

3. The Bonneville Salt Flats, usregime.

[Aloha Spirit]

4. Check this out: it corroborates the Zogby Poll...

[Frank Booth]

22. Huh.

[madrchsod]

8. so hillary is going to win by 20%

i guess it`s over for barack....just when i thought he was going to get the nomination...oh well.....

[usregimechange]

11. He may lose by more than 10% but no way in hell 20%, I would bet money on it.

[usregimechange]

12. So, want to bet on it? How much?

[woolldog]

15. I think it's very possible he could lose by 15+.

I don't want to take the bet though, because I'm rooting for him.

Best chance he has is if weak HRC supporters--those who are more voting more against him than enthusiastically voting for her, stay home.

[madrchsod]

17. hillary to win 6 and under

[ORDem]

24. Who cares? 10%, 20%?? She'll maybe net 10 delegates. 10! She needs about 160 to catch Big O.

And this in a big state where she was supposed to do well. Time is running out for her. NC will more than sweep away what paltry net she gets from PA.

:dem:

[woolldog]

9. Good points useregimechange.

I don't know what to make of Zogby's Sunday polling...or Saturday's for that matter. (i was hoping to get some interesting discussion about that in my thread but that was a total failure). I think his support in PA is soft given I haven't seen him above 46 percent.

But that is quite an incredible swing and there's really nothing that happened in between Friday and Sunday to account for it.

[woolldog]

13. Notice that Obama received

40% on Friday 46% on Saturday and 38% on Sunday.

That suggests that Saturday was the anomaly, not Sunday.

[usregimechange]

14. Or that the accurate number is 42-43, just what the 2 day average shows...

[woolldog]

16. which still would result in a 15+ point Clinton win.

He just doesn't have much strength there against Clinton.

[PseudoIntellect]

20. The RCP average is under 6% difference in PA.

RCP average > a Zogby poll

[woolldog]

23. True, but Survey USA has Clinton way ahead too.

And they have an excellent track record.

[geek tragedy]

18. Zogby tries to predict both candidates winning. n/t

[usregimechange]

25. Tries to sell his polls but he doesn't have a crystal ball

[The_Casual_Observer]

19. Piss your pants?

[Hoof Hearted]

21. LOL! Ahhh well you know ONLY the polls that have O-Jesus walking on water

are valid.

:rofl:

[usregimechange]

26. Not yet, we will prevail in the end, I promise