Survey USA: Obama closing gap in NC, trails by 4 (was down 5 last month).

[Drunken Irishman]

McCain leads 49-45.

http://www.surveyusa.com/client/PollReport.aspx?g=c12aadc7-2830-4be0-8e30-a647afe3277d

[dsc]

1. that isn't closing the gap

a 1 point change is well within the MOE, those polls are statisticly the same. On the up side being down by only 4 is quite good for NC.

[Drunken Irishman]

2. It might be just a 1-point change, but it's +1 no matter how you look at it.

[dsc]

5. No it isn't

Any poll is basicly a guess at what the actual proportion of people who are intending to vote for Obama is. That guess is designed to be accurate within a certain range. That 1 percent is well within that range. If you were talking about an average of polls then it would be a big deal. This isn't.

[Drunken Irishman]

6. Ugh.

Then, under your definition, Obama could never close the gap unless he took a solid lead over McCain there, because the MOE is around +/- 3% and his margin is almost exactly like that.

[dsc]

8. No

If you took an average of polls and showed a movement of 1 point that would be proof of movement. Or, failing that, if you showed several polls over time showing a point or two movement, that would likely be movement. Or he could gain 6 or 7 in one poll.

[Drunken Irishman]

9. I'm talking this poll.

Averaging polls don't work as well with so few NC polls out there. I've never understood state averages because many of them average polls that were done a month and a half ago.

In this poll, Obama is +1. Yes it is an insignificant gain, but a gain nonetheless.

[dsc]

11. a 1 point gain in a poll with an MOE of 3 isn't a gain

sorry but it just isn't. A 95 percent interval is roughly 2 standard deviations which means that the standard deviation of this poll is about 3/2. A difference of 1 is then 2/3 of a standard deviation. There is over a 1 in 4 chance that the difference is explained by pure chance.

http://www.regentsprep.org/Regents/math/algtrig/ATS7/ZChart.htm

look up -0.67 and find the area

I know this isn't exactly intuitive but I am telling the truth here. Obama is clearly doing better than expected but the trend is flat not him gaining from the evidence so far.

[Drunken Irishman]

12. Like I said...

Unless Obama overtakes McCain, every gain he makes in this poll, under your definition, will not be a gain. Which you and I both know is bullshit.

If by next month Obama trails McCain there 49-46, the gain is in the MOE, yet it shows support gaining.

[dsc]

14. that is why I typed the words

if in a series of polls over time you see this, then you can say he is gaining. But yes, to have this one poll show a gain you would need a 6 point gain. Though a 5 point gain would be reasonably likely a gain as well. It should be noted that even the 6 percent gain would only be a 95% certain gain. If this had been a 4 point gain, the likely hood would probably be around 70 percent that he had gained. We expect too much precision from polls which just aren't that precise.

[Alexander]

3. Yes it is. McCain is now below 50%. That spells trouble for him.

There are now enough undecided voters to put Obama over the top, if he does really well there.

[Triana]

4. Dayum! Gooooooooooobama! n/t

[democrattotheend]

7. They undersampled African Americans in the poll

SUSA pegged them at 19% of the NC voting population...2004 exit polls had them at 26%.

[Bullet1987]

13. Blacks will likely make up about 30% this time around if not more

[Top Cat]

10. My home state.... Yeah