More on the Polls
Political junkies had poll whiplash this week. The N&O/Charlotte Observer poll had Richard Moore leading Bev Perdue by six points. Then Public Policy Polling sent out an email saying Perdue was ahead by eight points.
PPP’s Tom Jensen went farther. He blasted the media poll:
“The simple answer is that the (media) poll is wrong. It was done by a company that specializes in market research and is not known for political polling. You can read more information about what was wrong with the survey in two posts on our blog.”
You know politics is getting rough when the pollsters go negative on each other.
PPL has been causing whiplash itself. Once it showed Perdue 27 points ahead of
Ryan Teague Beckwith at the Dome blog responded: to Jensen (and my post)
“To be clear, our pollsters were not trying to figure out who will show up on May 6 and how they’ll vote, but rather to get the mood of the electorate on the issues.
“To that end, they did not use a list of people who have voted in past elections to determine likely voters — the method used by Public Policy Polling, among others, in tracking polls. Instead, our pollsters called random numbers, then asked if respondents were registered to vote and intended to vote.
“That may be one reason why the number of undecided voters is much higher than in some other polls. It also makes our ‘horse race’ numbers a little less reliable, which is why our reporters did not stress them in the leads of the stories.”
That is a perfectly logical and straightforward explanation. But it raises another question: If the horse-race numbers are less reliable because of the poll’s methodology, is the analysis of the electorate’s mood also less reliable?
The N&O ran an editorial on the poll that seemed defensive:
“Give the pollsters credit for honesty. Not all opinion surveys in political races are as up-front about how many people can’t commit.”
More on the Polls
Political junkies had poll whiplash this week. The N&O/Charlotte Observer poll had Richard Moore leading Bev Perdue by six points. Then Public Policy Polling sent out an email saying Perdue was ahead by eight points.
PPP’s Tom Jensen went farther. He blasted the media poll:
“The simple answer is that the (media) poll is wrong. It was done by a company that specializes in market research and is not known for political polling. You can read more information about what was wrong with the survey in two posts on our blog.”
You know politics is getting rough when the pollsters go negative on each other.
PPL has been causing whiplash itself. Once it showed Perdue 27 points ahead of
Ryan Teague Beckwith at the Dome blog responded: to Jensen (and my post)
“To be clear, our pollsters were not trying to figure out who will show up on May 6 and how they’ll vote, but rather to get the mood of the electorate on the issues.
“To that end, they did not use a list of people who have voted in past elections to determine likely voters — the method used by Public Policy Polling, among others, in tracking polls. Instead, our pollsters called random numbers, then asked if respondents were registered to vote and intended to vote.
“That may be one reason why the number of undecided voters is much higher than in some other polls. It also makes our ‘horse race’ numbers a little less reliable, which is why our reporters did not stress them in the leads of the stories.”
That is a perfectly logical and straightforward explanation. But it raises another question: If the horse-race numbers are less reliable because of the poll’s methodology, is the analysis of the electorate’s mood also less reliable?
The N&O ran an editorial on the poll that seemed defensive:
“Give the pollsters credit for honesty. Not all opinion surveys in political races are as up-front about how many people can’t commit.”