Wednesday, March 21, 2007

The latest from the Civitas Institute

I attended my first Civitas Poll luncheon this afternoon to get a first hand look at their latest poll results. If you’ve never been, I recommend it. It's in downtown Raleigh at the Clarion Hotel and for only $10 you get lunch, a first peak at the results, and some great discussion.

One interesting finding in their poll was the impact of labels on the gubernatorial primaries. Jack Hawke, President of the Civitas Institute, explained that in previous polls they had only read the names of candidates and not job titles, because there are no titles written on the ballot.

So for Civitas’ latest poll they added titles and what happened? On the Democratic side their findings almost exactly mirrored ours.

Moore 25%
Perdue 34%

Jack Hawke thought that the “Lt. Governor” label gave Perdue a boost of about 10 percentage points. That’s possible. It definitely makes Perdue’s lead seem delicate. But PPP’s tracking polls have never included titles and we still have Perdue with a ten point lead.

On the Republican side the results (with titles) were:

Graham 8%
Orr 14%
Smith 11%

Hawke thought that describing Fred Smith as “State Senator and businessman” gave him a significant jump. It probably did, but we’ve always had Bill Graham with a lead in our polls.

I bet the differences in our findings are attributable to sampling. Our polls are only of Democrats and Republicans with a history of voting in primary elections. Civitas’ poll sampled general election voters. That’s a good sample for their statewide issue questions, but it may not get the best results for a primary contest.

I’ll have more on the latest Civitas poll in the next few days.


Will Cubbison said...

You cant say it. So I will. The difference between your polls is that you use actual questions and actual methodology. The civitas polls are laughable.

Next time you go to a lunch ask for the questions they ACTUALLY use. They are rarely what they say they are.

Justin Guillory said...

Haha. It is true that the language of the Civitas poll often leaves something to be desired, and they don't provide you with the exact question wording (like we do).

But I don't like to harshly criticize other pollsters methods, since our methodology is so often criticized.

Web Statistics