Thursday, September 11, 2008

Research 2000 NC Poll

Look the Tar Heels kick off in less than an hour and I already wasted one night this week debunking a cruddy North Carolina poll so this is all I'm going to say:

-Research 2000 is a good company, but the last time it and PPP had a major conflict about a North Carolina contest, this is what happened. They missed the margin by 25 points, we missed it by two.

-There is no way Democrats will do as badly with white voters as that poll indicates. And don't throw the 2004 exit poll at me- because the exit poll way over represented black voters, it means Democrats did much better with white voters than the exit poll indicated. Although the demographic breakdown of the poll generally passes the smell test, it's the results for that group that are throwing off the whole thing. I think Obama will end up with at least about 33% of the white vote while Kay Hagan and Bev Perdue will get 40% or more.

I guess North Carolina is a hard state to poll. I hope Research 2000 and Survey USA do a better job of it the next time they survey the state. All their previous general election polls here had looked fine this cycle.

10 comments:

Anonymous said...

Wow, don't hold back or anything. Please tell us how you really feel about Research 2000.

Truthfully, I think they "threw" this poll in order to show that they aren't always biased toward Democrats.

Tom Jensen said...

I don't think that's something polling companies really do.

I don't have a problem with Research 2000- usually they're just fine, but it is frustrating to see two different polls go so far off in the state in one week.

Anonymous said...

Why can't their polls be "right" and yours be wrong?

The election hasn't happened yet so to call other people out like that is rather childish.

RS said...

R2K has a smaller Democratic advantage (+9) than either PPP or Civitas (+13).
But more striking is the difference in how the Independent support splits:
NC close: PPP 44-43; Civitas 46-42
NC far-apart: SUSA 59-34; R2K 62-31

I guess we'll find out who's right... No disrespect, Tom - though since you are based in NC, I tend to trust your poll!

(I read someplace Obama pulled his staffers out of GA and into NC - I guess their internals follow PPP/Civitas!)

Anonymous said...

Tom,

Why do you continually denounce other polling agencies, when they conflict with your own company.

Tom Jensen said...

I received lots of e-mails this week asking me why the polls in North Carolina are so divergent, and who I think is 'right.' I am just giving my honest answer to that question based on poll demographic breakdowns and companies' prior history polling in the state.

Anonymous said...

Just ask yourself in any election in NC have independents EVER (excepting 1984 election which was a landslide) broken 60-30 in favor of one candidate? If not, that should answer your question about which NC polls are closer to the mark and which are off base.

Anonymous said...

Well, something is happening in North Carolina. More and more, it looks like it might be a bigger win for McCain.

One Key aspect of Charolette area was interesting. McCain is doing extremely well. If the numbers reflect that, there is a good chance for McCroy to be the next governor. I am biased for Republicans but, more and more data points to that direction.

Anonymous said...

Here are some data points that DO NOT point well for Reps. JUst this WEEK in NC, Dems registered over 7,000 new voters, Indies registered almost 6,000, and Reps 2,500 voters. It is this discrepancy between where people are moving with their feet (or better said with their party id) and where SUSA is showing it (a tossup with an even split between the number of Dems and number of Reps showing up) that just does not jive.

If ranjit wants to follow SUSA, fine by me, but I will follow hard, objective numbers that cannot be spun or disputed and those numbers do NOT show a wide swing to the Reps.

Anonymous said...

Sheesh. OF COURSE he thinks he's right! Why is this a shocking development. He's created complicated models, I presume, based on certain demographic trends, turnout assumptions, etc. If he doesn't believe in his own numbers ... I can't even finish this sentence.

 
Web Statistics