Monday, May 05, 2008


Are there any polling trends in either Indiana or North Carolina?

Sometimes polling trends are useful in predicting a winner.... if candidate A is gained in the polls, then chances are that candidate will do better than the polls predict.

There has been a trend in this primary season that in those states where Clinton leads, she picks up most of the undecideds. This is especially true if she is gaining in the polls. Unfortunately, there is no trend in the polls that one can pick up on in Indiana. If there is a slight trend, it is in Obama's direction. PPP has Clinton's lead down from 7 to 5 (but Clinton passes the 50% threshold). Insider Advantage has Clinton's down from 7 to 4 (Clinton doesn't pass the 50% threshold). Zogby has them both below 50% and shows no real movement. American Research Group (who is shady as hell... and I don't trust) has Clinton moving from a 7 to a 9 point lead (and above 50%). And the real wrench is Surveyusa which has Clinton moving from 9 to 12 (and over 50%).

Looking at the data, I can come to the conclusion that Hillary Clinton is the overwhelming favorite tomorrow, but there is no way to reasonably guess the exact final outcome. I went with the 3% + the current average lead as we saw in Ohio and Pennsylvania (5%). See if that works.

North Carolina is a state that Obama leads in. His lead stands at 7% in the polls. Here's a real whoozer. Early voting has Black turnout at 40%. Now, research suggests that Black turnout is usually higher during early voting than on election day. So we know it won't be 40%. Barack Obama won early voting in Texas but lost the primary.

Those polls that take early voting into account say Obama won it by 15-20 percent. Election day voting should be much closer.

What is the ultimate turnout of African-Americans and Whites will determine the election. If its 40% African-Americans, Obama will roll like a big wheel through a delta cotton field. If however, the turnout is 10% lower, Clinton has a shot. I split the difference at 35%... and thus give Obama his 7 point lead.

As I said, I feel a lot more comfortable about Indiana than North Carolina.

Just to complicate things however, I looked at the polling trends in North Carolina.

In every poll, Clinton has improved from the prior poll except one (where its the same).....

PPP has gone from a 12 point lead for Obama to a 10 point lead. SurveyUSA is the same 5. Insider Advantage from +5 to +3. Rasmussen from 14 to 9. Zogby from 9 to 8 (UPDATE SHOT BACK UP to 14... I think its garbage). ARG from 11 to 8.

What does that mean? Does it mean that Clinton will ultimately do better on election day? What about early voters?

Now, I also want to discuss a trend certain people believe they see. People claim that Barack Obama did better that polls indicated in South Carolina and Virginia (also in Georgia, but I want to concentrate on the other two). Thus they say he'll do better in North Carolina than polls indicate as the states are similar. I decided to look back at some of the polls and found things were not as easy as a, b, c.

South Carolina:
Zogby nails every candidates percentage except for one.... Barack Obama. What happened? There were a ton of undecided in the race.... They were overwhelmingly Black. He grabbed every single one.

Surveyusa: overestimated Edwards (remember him?) black vote. And the undecided White vote went to Obama (it was 2%). Further, "other" received two points in the poll. And those votes went to Obama.
Rasmussen: although we don't have full access to the numbers... all the undecided Black voters went Obama's way.

What does this tell us? Well it tells us that most of Obama's supposed "doing better than the polls" was nothing more than undecided Black voters going his way for the most part. The polls did a pretty decent job at judging Clinton's vote, but they did a poor job at guessing Obama's.

A little more interesting... but once again Obama's support is greater than the polls indicate, while Clinton is a little less (around 2-3%).
Surveyusa: Two reasons... one Clinton was said to get 13% of the Black vote. she got 10 (9 to 1 split anyone?). Two, her white vote was 3 points high. Obama grab white undecideds and the rest is history.
Mason Dixon: over 25% Black undecideds... all went Obama. And as with Surveyusa, Clinton did a little worse among White voters.
Rasmussen: Gave Clinton 22 percent of the Black vote... wrong. Give Clinton 10% and Obama the undecided Black vote... and whalaa.

So are these polls wrong because they didn't take into account high black turnout? No, they were wrong because they were wrong. In South Carolina and in Virginia, the polls overestimated Clinton's Black vote. In Virginia, it was a combination of that and pinning her white vote a little too high (by around 3).

In North Carolina, the dynamics have changed. I believe the election will be much more along racial lines. Clinton will gain in the white vote, and Obama will be 9 to 1 in the Black vote.

And do you want to know something interesting? Surveyusa has Black voters making up 32% in North Carolina. They said Blacks would make up 29% in Virginia, they made up 30%. They said Black would make up 55% of the vote in South Carolina, they made up 55%.

You think they might know something we don't?

I'm going 35% on the Black turnout.... maybe I'm a little high.

And if the turnout is 32%..... watch out because Clinton really does have a shot.

Comments: Post a Comment

Subscribe to Post Comments [Atom]

<< Home

This page is powered by Blogger. Isn't yours?

Subscribe to Posts [Atom]