Correction: Obama loses ground in SC according to Mason-Dixon poll

[antiimperialist]

Although Obama has closed in in Iowa and New Hampshire, contrary to an analysis in another thread, statistics from this new poll compared to the previous Mason-Dixon poll in June show the following trend in South Carolina, with the previous numbers in parenthesis.


Clinton 28 (25)
Obama 25 (34)

Different polls have different methodology, therefore can't compare to each other.
We could say that in this poll Obama lost ground if compared to the latest ARG, for example, where Clinton is up by over 20%. But we cold say that Clinton won 1% compared to the latest Rasmussen conducted in December 03-04.


http://www.foxnews.com/story/0,2933,284022,00.html

[Zueda]

1. Are you for real? nt

[antiimperialist]

2. the numbers are for real

...

[Zueda]

3. Let me try this again .. Are YOU for real? nt

[mtnsnake]

9. What is it with you? Why don't you try offering something concrete

instead of running from thread to thread asking irrelvant questions or posting goofy one-liners that have no follow-up explanations about WTF it is you're trying to say?

[antiimperialist]

4. Let me try again. The numbers are for real

IF the numbers are for real I am for real. Are you on denial?

[Frances]

5. Personally I think we will see ups and downs for

each of the candidates over the next month.

Many people still haven't made up their minds. That includes me.

[ClarkUSA]

6. Oh, please, he's within the MOE in IA, SC, and NH now.

And the newest Iowa poll is from Newsweek (12/5-12/6)... and it has Obama ahead by SIX points, outside of the MOE.

[antiimperialist]

8. I'm talking about SC only

I admit in Iowa and NH he's gained some ground in the Mason-Dixon trend.

[ClarkUSA]

12. He's been gaining in SC all along - a statistical blip means nothing - wait until the "Oprah bump"

;-)

[Imagevision]

7. Bridgeport Post CT. Polling has Obama @ 33 - Hillary @ 26

[jefferson_dem]

10. As compared to the numbers from *June*.

Ok. Technically true, I guess.

[antiimperialist]

11. And around June almost all polls except for Mason-Dixon had Clinton ahead

This shows that Mason-Dixon in June had a big Obama bias. Not saying Mason Dixon was inaccurate, but that it deviates from the average. This puts in doubt your assumption that it is fair to compare Mason-Dixon to the average of all polls.

http://www.realclearpolitics.com/epolls/2008/president/sc/south_carolina_democratic_primary-234.html

[suston96]

13. South Carolina? 12-6-07

Looks like Hillary is headed up, Obama down. Trends, you know. LOL.

http://www.pollster.com/08-SC-Dem-Pres-Primary.php

[antiimperialist]

14. That's not a Mason-Dixon graph. It's a graph of all polls

And like I said, around June, most polls had Clinton up at the same time that Mason-Dixon had Obama by 9%. And we are discussing Mason-Dixon.

[suston96]

15. A graph of all polls?

Yeah, I should have slept through Statistics 201 where I was taught to never rely on one set of stats - one poll. I was taught to make an average, throwing out the outliers - the highs and the lows - and then averaging the rest to get a better indication of what's really been polled.

But hey, people like to orgasm on their favorite one day poll. More joy to you and yours.

[antiimperialist]

16. That's my point. You compare averages to averages. Mason-Dixon to Mason-Dixon

The person who originated the other Mason-Dixon poll thread said Obama had gone up in SC, when in fact he did worse in this Mason Dixon poll than in the last poll.

Back in June, a graph would have shown hillary ahead in SC, while Mason Dixon had Obama up by 9% at the same time. Therefore I am right in arguing that we can't say that Obama is not moving up in SC based on this Mason-Dixon poll. He's moving up in the average, but different polls have different methodologies.

[suston96]

17. Your point is well taken....

But no single poll should be used for anything other than comparing it with other similar polls with similar methodologies and so on.

And even then, outliers must be considered and removed to get any kind of feel of what is happening.

In this case, South Carolina, a single poll even when comparing it to a previous version of itself, will not help much because its methodology could have been flawed earlier.

[antiimperialist]

18. Well, we agree then

Because some were drawing the conclusion that this Mason-Dixon poll proved Obama was surging in SC, when it doesn't necessarily do so.