Poll: South Carolina Dem Race Tight

Comments (33)

Liberal Larry wrote on December 11, 2007 6:14 PM:

In other words, all one has to do to change the results of the poll is to change the formula for who is a likely voter.

Eric Kleefeld wrote on December 11, 2007 6:18 PM:

Not necessarily. The formula can stay the same, but people's responses to those questions will change.

Anonymous wrote on December 11, 2007 6:26 PM:

Is America ready for a Muslim president?

Dan wrote on December 11, 2007 6:28 PM:

Screw you, anonymous, if that wasn't snark.

dcshungu wrote on December 11, 2007 6:29 PM:

Interestingly, the internals show little change in terms of Obama's performance with white voters or Hillary's performance with blacks. The change has come in the demographic make-up of likely voters. A month ago, African-Americans made up 49% of likely voters — whereas now they are measured as taking up 54% of the likely Dem electorate, providing a net boost for Obama.

If I had to analyze this poll, I would say that it is highly skewed in favor of Obama, for 2 reasons: (1) The change in demographics as stated above to include more blacks right on the heels of (2) the Oprah "star power" effect, which had overlapped with the dates during which this poll was taken. Would the Oprah effect last? Time would tell, however, even then, a representative poll would try to survey the demographic groups in proportion to their relative make up of the state population. Such a poll, of course, might cost a bit more but would be more representative of the prevailing opinions in the state...

The race is tightening but Clinton remains the candidate to beat!

hello_world wrote on December 11, 2007 6:31 PM:

Anonymous@6:26

Why, do you know any Muslims running for President? Oh wait I see, you're one of those guys.

Greg DeLassus wrote on December 11, 2007 6:33 PM:

The race is tightening but Clinton remains the candidate to beat!

I am sure that is a fair summary of it.

gqmartinez wrote on December 11, 2007 6:57 PM:

dcs,

I'm a Hillary supporter, but I wouldn't call this poll biased for Obama. A poll is a poll and you have to read them with all the usual caveats. Obama bringing in Oprah may have given him a boost, but that's not bias in the poll, that's good campaigning and how you run a campaign to win. Obama is in a promising trajectory and Edwards seems to have bottomed out. If Obama can continue to avoid any critical coverage in the media, he'll be in a good position and he's doing the right things to ensure that position. Edwards can sneak his way back in since he can spin a good showing in Iowa as him "coming from nowhere" (despite being the favorite to win).

If you're a Clinton supporter, you don't poo-poo any poll just because it isn't what you want to see. You go out and try to change it.

M wrote on December 11, 2007 7:11 PM:

GQM,

Thanks for the reasonable analysis. Everyone has a favorite poll showing this or that, but regardless, the only thing that matters right now is momentum and Iowa.

I am an Obama supporter in the primary election, but a Democrat first and foremost. Smearing each other does no one any good, except for maybe pre-vetting our respective candidates for the general election.

Having noted that polls are changing, and also noting the general positive media reception to Obama, could another factor be involved in the polls; re: people are just now, in a general fashion, starting to really pay attention to the primary?

If that is the case, I think the theory would play out in the coming weeks by taking a look at polls in states such as Minnesota, California, and other Feb. 5 primary/caucus states.

One way or another, this will be plenty exciting.

Would anyone be all that surprised, given an Obama victory, of an Obama/Edwards ticket? Or is this too much to hope for from my progressive soul?

Thanks, GQM. I don't see enough rational Hillary supporters on blogs these days. I love your candidate too, just not as much as I love Obama and Edwards.

DTM wrote on December 11, 2007 7:22 PM:

As usual it pays to fact-check Eric.

Since the last poll, Obama's support among black voters went from 52% to 56%. His support among white voters went from 15% to 19%.

Meanwhile, it is true Clinton's support among black voters was unchanged at 39%. But her support among white voters went from 55% to 51%.

So, the change in likely voter mix was part of the explanation for the closing gap. But a change in the internals was also part of the explanation.

By the way, a likely voter model for the primaries set at the demographic portions in the state would make little sense, if nothing else for the obvious reason that Democrats and Republicans do not have the same racial distribution.

dcshungu wrote on December 11, 2007 7:34 PM:

GQM: Fair point but the Oprah effect alone would have been fine. However, that coupled with the increase in the number of blacks in the sample would skew the poll for obvious reasons. A poll, e.g. the Zogby internet poll, can fail if not conducted in a way that avoids bias... I stand by my analysis. This poll compared red v. red apples.

Posted in transit...

dcshungu wrote on December 11, 2007 7:38 PM:

oops! "Red v. green" apples, so it is not all that "bad"...

NCSteve wrote on December 11, 2007 7:40 PM:

Eric isn't really fudging the internal demographics. All the shifts toward Obama are within 4 points which is within the MOE. Which does not mean they aren't real, just that you can't say so with the requisite statistical certainty.

They don't appear to have changed the poll or the methodology. More black people answered "yes" to the question "are you planning on voting?" If you don't say "yes" to that, you never get to whatever other factors they use to determine whether you actually will.

Maxwell wrote on December 11, 2007 7:42 PM:

If anything the poll is skewed to Hillary. Look at the male to female assumptions. This poll assumes that 60% of the primary electorate will be female, and 40% male. That's a tremendous disparity. Is there any real precedent for that, or is it just a speculative guess that this is a "movement" election for women?

Given that Obama is drawing 53% of male voters to Clinton's 26%, even a slight change closer to gender parity would shift the result toward Obama's favor.

dcshungu wrote on December 11, 2007 7:48 PM:

LOL. This was a Dem primary poll...at least it was the last time I'd checked.

Extreme case : Who would benefit if you increased the number of mormons in a sample? Duh. The key would be to include mormons in proportion to their relative number in the population at large. This is elementary stuff...

DTM wrote on December 11, 2007 7:48 PM:

NCSteve,

The problem is that it simply isn't true, as Eric claimed, that "the change has come in the demographic make-up of likely voters."

In fact, if my math is correct (and that is admittedly a big if), out of Clinton's three point drop, only about 1/10th of a point is attributable to the demographic shift.

Obama did benefit more than Clinton lost (thanks to his greater differential). But still, out of his seven point gain, only about two of those points is attributable to the demographic shift.

So, regardless of whether you want to view the shifts in the internals as statistically significant (and that is actually a very complex subject), those shifts had much more to do with the change in the overall numbers than the demographic shift Eric highlighted.

random wrote on December 11, 2007 7:55 PM:

dcshungu said: "This is elementary stuff..."

Yep. True of all of your posts. Time to tack a break... until 2009?

dcshungu wrote on December 11, 2007 8:02 PM:

Not make too fine a point but it would be quite trivial to weight the sample according to relative number of R v. D in a demographic group, as the mormon example shows, failure to make the sample look like the population that it is supposed to be representative of would bias the results -- another elementary point.

Maxwell wrote on December 11, 2007 8:08 PM:

In 2004, CNN exit polls estimated:

57% to 43%, female to male
51% to 49%, white to black

So this likely voter model assumes a greater female turnout, and a greater African-American turnout.

DTM wrote on December 11, 2007 8:08 PM:

dcshungu,

Actually, the demographics of the parties change over time.

But more fundamentally, you seem to be forgetting the poll isn't supposed to be representative of all Democrats. It is supposed to be representative only of likely voters in the Democratic primary. THAT is the population they are trying to sample.

And of course it is entirely possible that the best available model of that likely voter population will change during the campaign, including with respect to demographics.

Anonymous wrote on December 11, 2007 8:26 PM:

If anything the poll is skewed to Hillary. Look at the male to female assumptions. This poll assumes that 60% of the primary electorate will be female, and 40% male. That's a tremendous disparity. Is there any real precedent for that, or is it just a speculative guess that this is a "movement" election for women?

The 2004 SC Democratic primary exit polls showed:

57% female, 43% male
51% white, 49% non-white

As African American turnout increases, the gender disparity in favor of female voters also increases due to the huge gender gap in voting in the African American electorate.

A 60%/40% female/male split in democratic primary voting this time around is certainly plausible and not really out of the ordinary at all. The overall general election voting in 2004 was 54%/46%. The Democratic primaries tilt much more heavily female.

Steve wrote on December 11, 2007 9:35 PM:

There seems to be some inconsistency in how results are portrayed. For example, Clinton comes out ahead in this poll despite a few recent ones in which she's behind, but the she's portrayed as losing support--because of a change from the same poll a month ago. Reasonable enough, I guess.

But I feel like candidates (including Hillary) are sometimes also perceived to have momentum based on week-to-week changes, even from different polls. Certainly news outfits have said things like "Hey, look last week Obama led by two points in an InsiderAdvantage poll--Clinton is surging and has gained six points on him since then!" before.

Meanwhile, a national poll showing her 30 points ahead (I can't believe it's really that big) is also described as not much movement, though it's quite different from what others have shown.

Maybe we should skip forcing everything into a narrative and just report the numbers (which probably don't mean that much anyway)?

phidda wrote on December 11, 2007 9:43 PM:

I prefer my own nonscientific, non poll tested method for measuring the zeitgeist:

http://www.google.com/trends?q=clinton%2C+obama%2C+edwards&ctab=0&hl=en&geo=US&geor=all&date=mtd&sort=0

Debashish Sinha wrote on December 11, 2007 10:14 PM:

Screw you Anonymous.

Debashish Sinha wrote on December 11, 2007 10:18 PM:

Barack has been on a tear in Iowa and SC, but I don't see him beating HRC for the nomination. The establishment is too strong, she's been too far ahead nationally, for the idealists in this country to beat her.

Keith wrote on December 11, 2007 11:15 PM:

THe problem with using the 2004 exit polls is that there was not an African American candidate in the race. If you look at SurveyUSA's caveat, they are not sure how many African-Americans will turnout when a viable African-American candidate is on the ballot.

Desider wrote on December 12, 2007 6:19 AM:

Looking at this for 1.3 seconds, it seems like Obama got quite a bit of support at Edwards' expense, and Hillary's dropped slightly but mostly that blacks are more enthused about the election now. This is to be expected from the Oprah-thon and the big question is whether it really means more people vote. The second question is what Hillary and Barack+Oprah will do in South Carolina in the next 1 1/2 months. Star power typically fades over time.

Anonymous wrote on December 12, 2007 6:44 AM:

Desider,

What Oprah helped do is focus voters on the campaign and on Obama as an appealing candidate. No one I know thinks voters will vote for a candidate because Oprah says to. What they will do, and what Oprah has asked them to do is take a good look at Obama. Oprah says when she did that she decided he was exactly what America needs in a leader at this time.

The trend in polls seems to be -- IA, NH, SC ... and inklings already in CA -- that when voters focus on the election and speciric candidates, the Bill-effect fades for Hillary, people don't like her as a candidate, and other candidates gain support. The recent WashPo/ABC poll showed that nearly half of Hillary's supporters are there because of Bill, not because they feel Hillary is the most qualified. It also showed that now less than half of Hillary support is "Stronly committed to her," so at least open to considering other candidates.

Oprah has helped Obama build monmentum, and has helped to focus voters still deciding on him as an candidate. She has used her celebrity appropriately and well. She's not a decider, or meant to be, just more fuel for building excitement and momentum for Obama as a candidate.

Concerned in Iowa wrote on December 12, 2007 6:54 AM:

dcshungu says: "But Clinton remains the candidate to beat!"

I agree totally. Bill Clinton is the "candidate to beat" in this campaign. Hillary is just not gaining any traction on her own. It's all about Bill.

But more progress is being made on beating Clinton every day.

As voters take a good look at Hillary as a candidate for president, not just as Bill's wife, they don't like what they see, support is eroding fast. The Huckabee phenomenon shows how quickly another candidate can rise. There's a lot of profanity, back biting and plate throwing in Hillary Land. The coronation is in seriouls jeopardy.

random wrote on December 12, 2007 7:01 AM:

Hillary's appeal seems to be "Nominate and elect me, I am the expert at politics as usual. I can do war with the Republicans and win."

What Obama is saying and Oprah is echoing is: "We need to break out of politics as usual, it is destroying us as a nation. I want to bridge divides, inspire, persuade and lead a large new majority with constructive change for the future."

I don't want more of the Clintons!

savvy wrote on December 12, 2007 9:12 AM:

ERIC you write:

"The change has come in the demographic make-up of likely voters. A month ago, African-Americans made up 49% of likely voters — whereas now they are measured as taking up 54% of the likely Dem electorate,"

What does this mean? Are you saying that the total number of likely voters shifted to having a larger percent of AA voters or are you saying that a larger percent of AA voters shifted to voting for Obama? i.e. did the pool of AA voters become larger or the percent of the pool of AA voters voting for Obama increase?

It is very unclear.

You further state:

"Hillary Clinton has a statistically insignificant 44%-40% lead "

If this is true the entire blurb you wrote is not valid. Statistically insignificant means that these results are LIKELY to have occured or PROBABLY the results of statistical variance as opposed to HIGHLY IMPROBABLY being the results of the selection process.

In short, this data is meaningless in terms of statistical accuracy.

DRinOH wrote on December 12, 2007 9:34 AM:

You can find this on Sullivan, or here...
http://www.southernpoliticalreport.com/storylink_1210_74.aspx
It's an Insider Advantage poll, so I don't know why they aren't posting it here at TPMEC (well, other than the fact that Eric Kleefeld works here...)

December 10, 2007 — An InsiderAdvantage/Majority Opinion survey conducted December 8-9 of likely voters in South Carolina’s presidential primary shows Barack Obama extending his lead over Hillary Clinton. The telephone survey of 480 likely voters in the primary was weighted for age, race, and gender. It has a margin of error of 4%

Q. If the South Carolina Democratic Presidential primary were held today, would you vote for:

Obama: 28%
Clinton: 22%
Edwards: 14%
Biden: 10%
Richardson: 2%
Kucinich: 1%
Dodd: 0%
Gravel: 0%
Undecided: 23%

votenic wrote on December 12, 2007 5:18 PM:

2008 Presidential Election Weekly Poll

www.votenic.com

New YouTube Video!
The Only Poll That Matters.
Results Posted Weekly Tuesday Evening at Midnight.

Post a comment

Name:
Email Address:
URL:
	Remember personal info?
Comments: (you may use HTML tags for style)