Apparently we can look forward to a story tomorrow about how most voters in North Carolina are undecided in the state's key races. That is true for the Democratic Senate race, but not for the Gubernatorial races on either side and certainly not for the Democratic Presidential race.
Here's the number one piece of information that needs to be released about this poll- of the folks they called who said they were registered to vote in North Carolina, what percentage of them said they intended to vote in the primary?
Whatever that figure is is essentially what the poll is pegging as voter turnout for the primary. We started out with a model that estimated about 20% turnout and switched it a few weeks ago to one that projects about 40% voter turnout.
I'm guessing that the call everyone in the phone book and simply ask them whether they intend to vote or not voter model yields a yes rate of more than 60%, maybe even more than 70%. That's the same as the McClatchy poll being based on a premise of 60-70% voter turnout, which is not even close to realistic.
It's also the reason why the undecideds in their horse race questions are skewed so inaccurately high. If there's little chance that half the people you interview for your poll are actually going to vote in the primary, they probably haven't been following the races closely and that's why they're undecided.
As I said yesterday, I think it's great for media outlets to commission polls. WTVD has it right by working with Survey USA, which is an outstanding political polling company. I look forward to their survey coming out every month. But it's pretty disappointing to see McClatchy not have the sense to hire a political pollster to do a political poll. Look at a google news search for Braun Research and you'll see they specialize in breakfast food and retirement planning, not calling elections. It's unfortunate the News&Observer and Charlotte Observer are splashing a deeply flawed poll all over their front pages.