AP poll: Obama leads Romney by 1% nationally

[Marzupialis]

Interview dates: August 16– 20, 2012
Number of interviews: 1,006
Number of registered voters: 885

Lead was 3% in June poll.

[mac56]

1. 1,006 interviews? 885 registered voters?

Oh yeah, that one's a lock.

[longship]

9. ^This^ nt

.

[cthulu2016]

16. Rejecting math, science and reason doesn't help

Do you think this poll would be much more accurate with a sample of 10,000? It wouldn't.

The results of a random sample of 1,000 people will not vary much from the results using a random sample of ten million people.

Start flipping a coin and recording the results. You will, within a thousand flips, be quite close to 50-50.

Flipping another million times isn't going to make your results much more accurate. You will simply move closer and closer to 50-50.

[mac56]

18. Gosh, thanks, Mr. Wizard.

[CabCurious]

30. That's not how statistical polling works

First of all, this isn't a blind, random sample. These are stratified samples.

Secondly, no national poll is very meaningful as a snapshot. The only time national polling tells us anything is when we see strong trends across time, across many polls.

[Walk away]

2. I wish I could be as patriotic as our First Lady but I am not proud of this country.nt

[DURHAM D]

3. What?

[Walk away]

32. Apparently half the people in this country are idiots. What is there to be proud of?...nt

[liberallibral]

4. How The Hell Is It Not 20%???

[sinkingfeeling]

5. NBC and WSJ say it's Obama by 4%.

http://www.slate.com/blogs/the_slatest/2012/08/22/nbc_wsj_poll_obama_leads_mitt_romney_by_4_points_.html

[dtom67]

6. This is the problem...

How can anyone doubt that the system is broken after reading this? Romney should not have any chance of winning, yet there it is...

[WI_DEM]

7. Again not much of a bump so far from Ryan

[johnnyrocket]

8. There's a LOT of pissed off Americans, it's a shame they take it out on the President.

[Liberal_Stalwart71]

10. Democrats on this very forum have been blaming Obama since day one.

It's no wonder that the larger American populace does too.

[ProudToBeBlueInRhody]

11. 1000 people....

Not really a great sample.

[Marzupialis]

12. Gallup tracking uses 3000+ sample

Is that a great sample in your view?

[ProudToBeBlueInRhody]

13. For a national poll, not really

In a state poll....it's good.

Then again, I don't care about the national polls....and I really don't care if people they polled were in states like Texas and Alabama, or Vermont and California.

[Marzupialis]

15. Can you mention a poll with a larger sample than Gallup's?

?

[ProudToBeBlueInRhody]

17. I don't know and I don't care

National polls are pretty worthless, and they aren't the goal of winning an election.

[Marzupialis]

19. First it was small sample vs. big sample. Now it's national vs. state

Your argument changed.

[ProudToBeBlueInRhody]

20. No it didn't....

It can be both....1000 people in a national poll is not a lot IMHO when we had 169 million vote last time out, and it doesn't matter anyway because it's the individual state polls I care about. The electoral college is my only concern.

Don't really know why you are so interested in coming at me on this. We disagree. Move on.

[Marzupialis]

21. You said 3050 people isn't a lot either

What do you consider enough?

[ProudToBeBlueInRhody]

24. I doubt there is any number you reasonably could sample in a national poll....

....that would be acceptable.....50,000 maybe? I know that's impossible....but 169 million voted in last election....

That's why those national polls don't mean a thing to me. Again, don't know why you're tearing at me on this. I'm certainly not going to celebrate a 1% point lead in a national poll. It's unacceptable.

[Hippo_Tron]

35. The central limit theorem says otherwise

Once the ratio of sample to population becomes smaller than 5%, it doesn't matter how much smaller it gets the margin of error is still the same. 1000 is still less than 5% of registered voters even in the least populated state Wyoming. Thus, it doesn't matter if it's a state poll or a national poll. A sample of 1000 still gets you the same thing.

[mzmolly]

26. Gallup has Rmoney ahead by two. RCP shows the Prez. ahead by 1.5 on average.

Though state polls, look good for Obama, regardless.

http://www.realclearpolitics.com/epolls/2012/president/us/general_election_romney_vs_obama-1171.html

[Sugarcoated]

14. That's nice.

[Mr.Bill]

22. Popular vote polls are meaningless.

It's the electoral vote that counts. Just ask Al Gore.

[Marzupialis]

23. Let's say national polls had Obama by 11%

Would that be meaningless? Would you have to wait to see state polls in order to figure out who will win?

[Mr.Bill]

25. 11%

That would mean more than 1%, but only in the sense that it would be more likely to have some bearing on the electoral votes. And if I were either candidate, yes, I would still want to see the state by state count, if for no other reason than to know where to focus my efforts. For instance, I live in Northern California, and there is almost no advertising by either candidate here so far. I suspect this is because Obama is thinking it's a sure thing and Romney thinks it's a lost cause.

[mzmolly]

27. Let's say the difference is 1%.

As that is what it is. If the difference was 11%, surely that would impact swing state polls.

[mzmolly]

28. Adding an RCP comparison of 2008 to 2012

Election '12 vs. Election '08 Four Years Ago Today

http://www.realclearpolitics.com/epolls/2012/president/election_2012_vs_election_2008_four_years_ago_today.html

[gabeana]

29. Thanks for the sanity

I chuckle when I see all the doomsedayers/sky is falling crowd here on DU

the polls are going to go up and down

people need to buckle up

[CabCurious]

31. Excellent post. Thanks :)

[RBInMaine]

33. AP is corporate right wing SHIT. They are the same as FOX. Other polls have Obama further ahead.

[juajen]

34. They're freakin' stealing it! No way is this accurate.