Elon has its new poll out- I think it is even more flawed than usual, but the results generally make sense.
What's wrong with it? Incredibly small sample sizes for one thing. No professional polling company would ever release a survey for the Republican Governor's race that included just 233 respondents. All of their results for election matches have much higher margins of error than anything we (or most outfits) would ever release.
They also said they only weighted for gender. My guess would be that means racial minorities and young people are probably seriously under represented in the poll.
Finally, they polled the primaries not by asking people whose party they would cast a ballot in if they voted in the primary but by asking which party respondents planned to vote for in the fall. So someone who said they were going to vote Republican for President and Democratic for Governor was asked who they would support in each of those primaries- even though that's not a real option in the voting booth. And of course as usual they make no effort to identify who is actually likely to vote in the primary (though they are honest about that.)
So with an enormous grain of salt, here are the results:
Barack Obama 45
Hillary Clinton 31
I tend to think that is about right, maybe a little charitable to Obama. We had Obama up two points two weeks ago but I am sure that has expanded as the race has moved his way nationally.
John McCain 62
Mike Huckabee 24
I doubt that it is really this lopsided. I think that's a function of not polling likely primary voters.
Bev Perdue 40
Richard Moore 28
I think the margin is probably correct or at least in the ballpark. The percentage of undecideds is too high but that's a function of polling people who will not vote in the primary and haven't given the race any thought.
Pat McCrory 28
Fred Smith 12
Bill Graham 10
Bob Orr 6
I don't think McCrory is really up by that much but the order seems about correct.
You can see the full poll here.
I don't mean to be too tough on Elon, I'm sure doing the poll is a good learning experience for the students. But I think they're fully capable of producing a much better poll.