Wednesday, September 17, 2008

Diverging Virginia Polls

In response to several inquiries, I just sent this e-mail to our Virginia press list:

I have received several questions about the disparity between our Public Policy Polling survey today showing Barack Obama with a two point lead in Virginia and a Christopher Newport University survey showing John McCain leading by nine points in the state.

While only the election will answer who is right or wrong, the demographic composition of the polls can easily be compared to the state’s demographics.

About 20% of Virginians are black and according to the 2004 exit polls for the state, 21% of voters that year were black. PPP projects blacks to make up 22% of the electorate, while Christopher Newport projects blacks to be less than 10% of the electorate.

To put this into voter turnout figures, for the Christopher Newport survey to be correct there would have to be 55% turnout from non-black eligible voters, while only 25% of black eligible voters would have to turn out. PPP thinks that black voters and those who are not will turn out at roughly equal levels.

There is also an age disparity. We project that 44% of the electorate will be under 45 years of age, while Christopher Newport anticipates only 12% of the electorate being under the age of 40. According to the 2004 exit poll 49% of voters were under 45.

A poll that significantly undercounts young and black voters will also significantly undercount support for Barack Obama. It is vitally important to look at the demographic compositions of polls and how they compare both to the demographics of the state and those of past electorates when trying to sift through disagreements like the ones found in today’s polls.

6 comments:

Anonymous said...

I hate to tell you this but I think you're going to find another credible poll released tomorrow with similar results to CNU.

Anonymous said...

If the CNU poll's demographics are as advertised, it's no wonder there's a slant.

To anonymous: we may see a similar result, but that is no credit to how CNU (apparently) got there.

Anonymous said...

Another poll similar to CNU. So I guess SUSA and Rasmussen are similarly bad poll results? Seriously anonymous --- do you really think that African American turnout in VA will be half of what it was in 2004? Because you have to in order to believe the Newport poll results. I can think of no serious political anaylst who believes that to be the case, but if you want to live in la la land go right ahead.

Anonymous said...

Tom,

First of all, this is not a criticism, but a real question which I hope you can answer:

I've heard that the exit poll in Virginia (much like the ones in North and South Carolina) may have been overstated to a degree (according to Larry Sabato, African Americans made up something like 15%-16% of the vote. Being from neighboring North Carolina, I was wondering if you've heard anything about this or not?

Tom Jensen said...

Elliot,

I wouldn't be surprised if that's the case. The 2004 exit poll for North Carolina definitely overestimated black turnout. With Barack Obama at the top of the ticket though I would expect the composition of the electorate this year to more closely mirror the demographics of the state, which would put the black vote in Va. at about 20%.

Statistikhengst said...

would be thankful if you could send me the poll internals for my statistical research (I mean, the CNU poll) - the exact internals, if you have them, to:

bonncaruso@gmail.com

thanks.

here my electoral statistics blog. Feel free to add me to your blogroll if you like what you see.

 
Web Statistics