4.29.2008

More polls: Clinton continues to gain ground in North Carolina, mixed numbers from Indiana

I reported this morning that the latest polling news brought some good news for everybody -- with the biggest question mark the North Carolina primary which is showing some major movement without endangering Obama just yet. And indeed, the latest Survey USA suggests that Clinton's NC investments are starting to really pay off:

  • SUSA finds Obama's lead down to five percent, 49% to 44%. Last week, Obama was up by 9%.
  • The movement is primarily among white voters, among which Clinton has opened a 31% lead; she tightened the race particularly in the Research Triangle area.
This week's other two NC poll had bigger Clinton improvements (+12% for PPP, +9% for Rasmussen) but also had Obama staying put with a bigger lead (+12% for PPP and +14% for Rasmussen). Considering the importance of looking at trendlines, however, there is little doubt that there is considerable movement in Clinton's direction. And the narrative that could be taking shape is exactly what Clinton is hoping for -- white voters moving away from Obama in the context of renewed controversy over Reverend Wright.

Whether this is due to a post-Pennsylvania bounce that will subside by May 6th or whether it indicates a more meaningful change of momentum is still unclear. It is worth pointing out that the Gallup tracking poll shows no sign of Obama picking up the ground he lost following the PA primary; the two candidates are in a statistical tie for the fifth day and Clinton even edged Obama out today, 47% to 46%. In the general election, Clinton now runs 4% better than her rival, with Obama losing 2% to McCain in one day. However, Rasmussen's tracking shows little progress by Clinton, though today is the first day in quite some time that both Democrats have a narrow lead against McCain.

Naturally, keeping NC close would do Clinton little good if she cannot carry Indiana comfortably. And two new surveys from Indiana show two different leaders in the all-important May 6th primary:

  • After showing a Clinton comeback in NC, PPP now finds Clinton to be leading comfortably in Indiana, 50% to 42%; as always, Obama is benefiting from the open primary as he has a 15% lead among independents. The result and internals are similar to yesterday's SUSA poll, suggesting (just like yesterday) that PPP changed its turnout model.
  • A Howey-Gauge poll, meanwhile, shows Obama very narrowly ahead, 47% to 45%.
It has been obvious for quite some time now that Indiana is the first state in quite some time that could truly go either way; polls with Clinton leading tend to show a slightly bigger margin than polls with Obama ahead, but the Illinois Senator's ability to massively outspend (and thus outorganize) Clinton gives him an added bonus.

Update, in response to Mark: While I fully agree that it is virtually certain that Obama will pick up much much more than 73% of the black vote in Indiana, I disagree with the rest of your analysis of the internals of these polls: (1) A gender breakdown of 55% women to 45% men does not over-represent the female vote; in fact it is a 3% underrepresentation of the female vote, which should be good for Hillary. If this was a general election, you would be right; but the female vote constitutes 57% to 59% of the electorate in most Democratic primaries (58% in PA, 59% in OH and IL).

(2) The partisan breakdown is also in line with what we have seen in previous open primaries. About 70% of registered Democrats participating sounds about right, even accounting for the fact that Republicans don't have a competitive primary. (3) The "already voted" numbers you refer to in the NC poll by SUSA represents 2% of the sample. That's about 10 to 12 people... These numbers have no statistical significance whatsoever, the margin of error in that small a sample is humongous and it was irresponsible of SUSA to include numbers from such small subsamples.

Labels: ,

2 Comments:

  • Thanks for the good write-up. As always, the devil is in the details:

    Indiana:

    PPP Internals: shows Obama with only 73% of the black vote. Highly unlikely. More likely will be 90%. That's a 17% difference of 10% of the population of Indiana, which means a possible (or better put, likely) error of 1.7% against Obama right now in this poll. Which makes it more like 48.3 to 43.7 - if we are realistic about this poll. Obama has polled consistently higher and higher in the black vote in each successive race and there would be no reason in the world to think that that pattern should suddenly change now. Something must be not right with the statistics here.

    The polls is also comprised of 55% women / 45% men. This means that women are overrepresented by 3 points. It is also based on a representation of 74% democrat / 14% republican / 12% other - which will be hard to gauge do to the open primary nature of this state and no statistical way to know, from voter registration, how many republicans will actually go vote in the DEM primary. But in this state, according to the poll, she is winning in the republican vote. Obama is swamping her in the "other" vote.

    So, the way I see it, it really depends on whether 2,000,000 GOPers in Indiana wake up on primary day and say, "Gee, I think I will go vote for Hillary". Because, if more independents end up voting than in this poll model and surely Obama will pick up more than just 73% of the black vote, then the race will be much closer than this. It could look a lot more like the Howey-Gauge poll.

    But I still think she will win IN by a lean margin. And they will split the delegates about 50-50.

    ----
    SUSA, North Carolina, internals:

    Under the category "already voted", it's Obama 57, Clinton 39 (margin: Obama +18) - and this is the more likely spread we will see than just 5 points. Strangely enough, under "already voted", there is a subcategory for likely voters. Kind of dumb. All voters who have already voted ARE likely voters. Oy. They are not only likely, they are CONFIRMED. Oy. In the SUSA poll, Clinton has pulled ahead among the affiliateds, a statistic we have not seen in any other NC poll yet. So, we will have to delve into the next polls and see if this holds or if it is an outlier.

    SUSA pegged Ohio right on the money.

    It was off in PA.

    So, I doubt seriously that the race in NC is in single digits.

    By Blogger Statistikhengst, At 29 April, 2008 16:01  

  • Good information but I cannot believe that CLinton will improve much in the Triangle (RTP) area as this is such a demographic goldmine for Obama - students, African Americans, whites making >>>$50,000 a year etc. Clinton country is going to be eastern NC and northern NC.

    By Anonymous Anonymous, At 29 April, 2008 16:42  

Post a Comment

Subscribe to Post Comments [Atom]



<< Home