by Bob Geary
Public Policy Polling, the Raleigh firm, is out with its first post-conventions survey of North Carolina voters. In NC, McCain leads Obama 48%-44%, with Libertarian candidate Bob Barr drawing 4%. In the U.S. Senate race, Democratic challenger Kay Hagan and incumbent Sen. Elizabeth Dole, the Republican, are in a virtual tie; ditto gubernatorial candidates Bev Perdue, the Democrat, and Pat McCrory, the Republican. The numbers: Hagan 43%, Dole 42%, Libertarian Chris Cole 6%; Perdue 41%, McCrory 40%, Libertarian Mike Munger 6%.
Says PPP analyst Tom Jensen:
The race in North Carolina is pretty much where it was before the two party conventions- close but with a definite advantage for McCain. A concern for Obama rising out of this poll is that undecided whites have gone from 10% three weeks ago to just 3% now, and two thirds of them have gone into the McCain column.
A Survey USA (SUSA) poll for WTVD in Durham yesterday put McCain up 20 in NC, with Dole and McCrory each up 8. A good time to remember that approximately one poll in 20 will be an "outlier." In other words, given the mathematical probabilities of sampling, you can accurately predict voting behavior about 95% of the time within a "margin of error" that gets smaller the more people you sample. But then there's that other 5 percent of the time, when the results of the sample will be inaccurate. So, if we assume for argument's sake that the PPP poll is one of the 95% and the SUSA poll isn't, then with a margin of error of 4%, Obama could be even with McCain in NC, or he could be 8 behind -- or anywhere in between. But he's not 20 behind.
On top of which the Perdue campaign is out with this on the SUSA poll, arguing that it over-sampled Republicans -- another trap-door for pollsters.
Pollster.com analyzed the recent Survey USA poll and suggests that there’s a major error in the Survey USA methodology: “In other words, in a survey conducted less than a month later, Democrats made up 6% less of the sample and Republican representation increased by 8%. Everything we know about partisanship suggests that such massive shifts over such a short period are highly unlikely.”
You can read the post here: http://www.pollster.com/blogs/about_that_north_carolina_poll.php