Saturday, October 18, 2008

Research 2000 NC Poll

Sometimes in polling two wrongs make a right.

I think that today's new Research 2000/Daily Kos Poll showing Kay Hagan leading by four and Barack Obama up by two in North Carolina is right on the money.

How they got there, however, does not match up with reality. In 2004 the electorate in North Carolina was about 79% white, 18.5% black, and 2.5% other. Although I'm sure it will be more colorful this year, Research 2000's projection of 69% white, 24% black, and 7% other is not realistic. 74% of registered voters in North Carolina are white, 21% are black, and 5% are other races. Typically non-white people here vote at a much lower rate than whites- if minority voters simply match the proportion of registered voters they comprise that would be a huge movement in and of itself, but they're not going to turn out at such a disproportionately high level that whites get pushed down to 69% of the electorate when they're 74% of registered voters.

(Before someone puts the demographics from the first few days of early voting in the comments to try to show me I'm wrong, keep in mind that over 40% of early voters during the primary were black but the election day voters were so disproportionately white that it pushed the total final African American share of the vote down to 34%.)

So if the Research 2000 poll significantly over samples minority voters, why aren't its numbers overly friendly to Obama and Hagan?

Because they find Obama getting just 27% of the white vote. This was the same problem with the R2K poll of North Carolina last month, which showed McCain with a 17 point lead. Simply put, Obama has zero chance of winning the state if he really only gets that percentage from white voters. He needs 35-36% at a bare minimum, and we've found him as high as 38% there the last few weeks.

I believe this poll is wrong in that it projects minorities to make up far too large a segment of the electorate, and it is wrong in that it gives Obama far too little of the white vote. But those two errors offset to produce topline numbers that are very reasonable.

12 comments:

  1. On their NC-Gov numbers, its obvious that they accidentally switched Perdue and McCrory's numbers. McCrory is supposedly winning 80% of Democrats and 87% of African-Americans, while Perdue supposedly wins 84% of Republicans and 60% of whites. So if you adjust for this, Perdue is actually 5 points up, which seems reasonable.

    ReplyDelete
  2. Tom: Any idea on when new FL, VA, or NC numbers will be coming? Thanks.

    ReplyDelete
  3. We're not due for Virginia this week, I believe you are thinking of West Virginia? I am hoping to get that out tomorrow.

    NC will more than likely be Monday and Florida Tuesday.

    ReplyDelete
  4. Okay, excellent, WV tomorrow? Already? You guys are fast! Willing to drop us any hints?

    ReplyDelete
  5. Tom,

    Sounds great, can't wait for WV. And yes, hints and teases are really great... even if they are good for McCain!

    Also, has Indiana been scrapped? (Not that I expect you to do all four...)

    ReplyDelete
  6. My clue for WV is that I hope our poll is wrong.

    We're doing Indiana too, that will probably come out along with Florida on Tuesday.

    ReplyDelete
  7. Oh no (for WV)!

    Well, I never fully bought that WV was that much in play. Also, maybe your survey is showing more Republican numbers than other pollsters because of when it was taken. Most tracking polls are showing that the race is tightening - and that is probably first and foremost taking place in red leaning states.

    ReplyDelete
  8. Any chance you guys are going to poll NC-8 again? You guys havent polled in awhile and its a top level race.

    ReplyDelete
  9. I hope to get in one more NC-8 poll. It would probably be in the field Sunday, Monday, Tuesday before the election with a release six days out.

    ReplyDelete
  10. In case you missed it, R2000 has confirmed that they got the Gov race flipped. They do show Perdue up by 5%.

    ReplyDelete
  11. Five points is still within the margin of error. I just watched the debate from last week in the NC governor's race. I am not sure Perdue has a good understanding of of the roles of the Legislative Branch even though she has been the Lt. Governor for the past eight years. She repeatedly said on my watch or when I become governor I will enact laws.... She is not bright. Perdue makes Palin look knowledgeable. I was impressed with Munger. He is smart and he understands the issues. My advice to the party leadership, push Hagan and spend the next four years finding a candidate..

    ReplyDelete
  12. I'm a democrat from Charlotte and am not surprised by the Hagan and Obama poll, but am surprised by the Perdue poll. Someone outside of North Carolina who does not understand the niceties of our politics may not know that there is a big perceived divide between urban Charlotte and rural North Carolina. Going to a state university (UNC Chapel Hill), has very much heightened the awareness that I have for this issue. Many Charlotteans--democrats and republicans--feel very much ignored and disdained by the state government. As a result, there are many democrats in Charlotte who will for Obama and Hagan, but will reject Perdue and vote for McCrory for governor because he is the mayor of Charlotte and is perceived there as someone who will fight for Charlotte's interests in the state government. Something to think about.

    ReplyDelete