Thursday, September 17, 2009

Public Policy Polling is biased

Article quoting a PPP employee who admits that the polling company is indeed biased in its political opinions:

“We’re absolutely rooting in the race. We don’t want Richard Burr to get reelected. We wanted Obama to win last fall,” said Jensen. “But our reputation is predicated on getting it right, and we’re not going to cook the numbers just to tweak Richard Burr’s nerves. They are what they are.”

Recently, a poll of likely New Jersey voters was put on the blog. What kind of reputable polling company would ask people questions like, "Do you think Barack Obama was born in the United States?" and, "Do you think Barack Obama is the Anti-Christ?" Probably one that's looking for a few cheap laughs.

Definitely some of the most useless information I've ever read.

1 comment:

Evan said...

PPP also asked if people believed Bush caused the 9/11 attacks, something that generally makes some Democrats look very stupid. Just because this poll made New Jersey residents on balance look pretty stupid doesn't mean it's automatically invalidated. Unlike the poll Lachlan posted earlier, PPP's poll was phrased in a responsible way and was in no way a push poll, moving respondents to be forced to a conclusion. The sample size--500--is about right for a state like New Jersey, with a margin of error of 4.5% on the whole sample. As a result the MoE on some of the subsamples can be larger--for instance, the birther/anti-Christ polling on conservatives or the polling on truthers--but the gist of the poll remains the same. Some people in New Jersey believe crazy things.

FiveThirtyEight says that PPP is slightly more accurate than the average pollster - see here.

Just because PPP is a Democratic polling organization doesn't mean they're inherently bad. Even if they have a rooting interest, that doesn't by default make them prone to giving bad outcomes. It's in the interest of all pollsters to be as accurate as possible; if you blow a poll, it's embarrassing, and people won't take you seriously.

The results of this poll kind of do suck for New Jersey. But there's no reason to suspect they're wrong.