Gallup's latest Poll Over samples Southern Voters by 50%

[berni_mccoy]

In Gallup's latest Poll, they claim Obama is losing to Romney 52-48% on Likely voters sampled between Oct 9-15. See: http://www.gallup.com/poll/158048/romney-obama-among-likely-voters.aspx

NOTE: The headline reads 50-46, which is also a 4 point gap, but that leaves out 4% who either are undecided or not voting for either candidate. The data in the chart is re-calculated to only count voters who gave an answer (so 52-48% of the 96% who picked a candidate).

In the chart, they break down the percentages by region:

East: O(52) - R(48)
Midwest: O(52) - R(48)
South: O(39) - R(61)
West: O(53) - R(47)

So, forgetting the south for a moment, Obama is winning every where else in the country by a 4-5 point lead.

I repeat: in every region of the U.S. EXCEPT THE SOUTH, Obama is winning by a 4-5 point margin! This is great news for Obama because that means he should win by a large margin of electoral votes. But that is a side note for this discussion, albeit, an important one (we all know Obama is winning the real campaign, the electoral vote campaign, this just confirms it).

So how does Gallup get Romney in the lead by 4 points? If you equally weight the regions, The total should be O(49)-R(51), which seems consistent with other national polls lately (Rasmussen, IDB/TIPP, etc). That means there must be a greater weighting to the South in order to show O(48)-R(53). A simple calculation shows that the breakdown of weighting must be:

East: 23%
Midwest: 23%
South: 32%
West: 22%

That means the South is given 9-10 more points than the other regions of the U.S. In other words, the South is being weighted 1.5 more times than the other regions, a factor of 50% more than what it would have been to produce an overall total of 49-51. '

How do we know this poll is completely bogus? Because they did not weight the regions from 2008. Yes, on the same web page, they show the same data from 2008. And the totals from 2008, Obama 53.5 to McCain 46.5 add up exactly to an equal weighting by region.

Gallup. You're BUSTED.

[Cali_Democrat]

1. Why is the South so against Obama?

Could it be because.....um.....he's....ummm......

Naahhh....couldn't be.

[truebrit71]

3. Gee I wonder.....what could it be...

...

[MADem]

2. Wha' em ah nawt sup-prized??

They're bullshitters.

It's like they are trying to game the masses with bullshittery...or maybe they're trying to create a number trail to justify the theft?

[berni_mccoy]

4. The kicker here is that they only weighted the 2012 numbers. They left the 2008 numbers untouched.

Which shows they are truly manipulating the data to tell a manufactured story.

[MADem]

10. This has been my sense all along--they'd better not try to paint the USA as RMoney-loving peoples...

I think there will be a revolution if they try to steal this one. I'm too old for that shit.

GOBAMA, damn it--I want him to win it decisively!

[cthulu2016]

5. The South is 37% of the POPULATION

As I said to you in the other thread, the US Census region the South has 37% of the population of the US.

But that doesn't mean Gallup under-weighted the south at 32%. There are reasons we would expect the south to have fewer voters relative to population.

[berni_mccoy]

6. So why didn't they weight the 2008 data then?

And as I said before, it depends on what you mean by "South". The "South" is not a well-defined term. Does it include MD, VA and DC? The population figure you cite does. If so, the spread is wrong.

[starroute]

7. I did find a listing that supports their breakdown

It lists:

South: 34.4% of Total U.S. Population
Northeast: 20.4% of Total U.S. Population
Midwest: 24.0% of Total U.S. Population
West: 21.2% of Total U.S. Population

But if you look at the map that goes with it, it has the south including everything south of the Mason-Dixon line -- meaning Delaware, Maryland, Virginia, and Washington, DC.

Given these, and that Florida and North Carolina are swing states at the very least, I would suspect that the real problem with the figures is that the Deep South is far more heavily Romney than just 61%, which skews the entire region.

Not that it wouldn't be a problem if Obama won in the Electoral College while losing the popular vote, but given the trend of the other polls, I can't see that happening.

http://dmc122011.delmar.edu/socsci/rlong/intro/usmap.htm

[berni_mccoy]

9. Yes, I am aware of what is typically considered the southern population.

That doesn't change two facts:

1. Winning by a large margin in the south won't help Romney win the Electoral Votes needed overall.

2. Gallup didn't weight their poll proportionately by population for the 2008 numbers.

[MADem]

11. POPULATION....v. VOTING population.

The northeast is chock-a-block full of OLD people--I know, I drive them to the polls.

My friends down south have four to six kids, most of 'em. They don't vote, those kids...

[durablend]

8. Well what's so shocking?

They're just saying only people who would vote for Romney matter....whocares about those America-haters who would vote for "that other guy"

[Indykatie]

12. I Hate How the Media Focuses on National Polling

This analysis is a reason why they can be so mis-leading and can paint an inaccurate picture of the race which may in fact be the reason for keeping them front and center. The pundits repeat these numbers and seemingly never bother to provide caveats even the more liberal leaning ones.

[berni_mccoy]

13. Yes, national polls are meaningless unless the underlying breakdown is known.