Thursday, September 12, 2013

Is N.C.'s Public Policy Polling biased?

Is Public Policy Polling biased? The Raleigh-based firm is a prolific poller, and has established a strong record of accuracy over the years. But it's also described as Democratic-leaning and now finds itself in a debate about whether it concealed poll results that were unfriendly to a Democrat.

PPP polled in Colorado's state Senate recall elections last weekend. When it found that Democratic state Sen. Angela Giron was down by 12 points, it decided not to make the results public. PPP said it was concerned that something was off with its methodology -- it was unusual that a Democrat would be losing so badly in a heavily Democratic district. Perhaps respondents didn't understand the question?

Turns out they understood the question just fine. Giron did in fact lose in Wednesday's recall vote by the precise 12 points that PPP's survey predicted. And now PPP's decision to keep its results private is lighting up the Twitterverse, including from data phenom Nate Silver, formerly of the New York Times.

"VERY bad and unscientific practice for @ppppolls to suppress a polling result they didn't believe/didn't like," Silver tweeted Wednesday afternoon.

"Nate I'm sorry but that is absurd," PPP tweeted back. "You're saying you would put out a model if you had serious concerns it was wrong?"

"I'm especially skeptical when a pollster puts its finger on the scale in a way that matches its partisan views," Silver later tweeted.

This is more than just a spat between wonks. Alleging that a polling firm is suppressing poll results because it doesn't like them is an extremely serious charge. Silver's accusation would have more credence if PPP didn't a) routinely release polls that included bad news for Democrats and b) have a long track record of accuracy and c) blow the whistle on themselves in this case.

Another tweeter pointed out that Silver himself withheld poll data on election day that he thought was suspicious, and included video to prove it. (Watch in the 5:40 to 6:10 range.) No word from Silver on that.  

If a polling firm sees red flags that suggest its data may be off, it makes sense for it to revisit its methodology rather than just throw results out there that it doubts are true. Silver is a smart guy, but he's wrong on this one.

-- Taylor Batten


 

5 comments:

WashuOtaku said...

It's only an issue if they make it a habit out of it. If it was just this one time, understandably so; but if they hid other results in similar fashion, then yes, their credibility is in question.

kantstanzya said...

Mr. Batten is exactly correct. The Democrat polling organization PPP is certainly no more biased than The Charlotte Observer and NBC, CBS, etc. are in reporting...or not reporting....the news.

As Dan Rather and his producer Mary Mapes (who spent FIVE years digging into Bush's National Guard records!) famously said after CBS was caught co-ordinating with the Democrat Party and intentionally using forged documents to run a false story about George Bush to try and influence the 2004 election that they ran it anyway because, despite the lack of real evidence, they "just knew it had to be true anyway".

That the liberal media exist in a cocoon where other world views except their own are inconceivable is well known. There is Paulene Kael's famous comment after Nixon's landslide in 1972 that it was an impossible result because she didn't know a single person who voted for him. One poll by Freedom Forum found 89 percent of Washington Bureau Chiefs supported Al Gore and 7% George Bush. Not much has changed.

We should not be surprised that PPP did not believe their own polls or that they refused to publish them and thought surely they had to be flawed. Afterall they are still shocked the NC Constitutional Marriage Amendment which was supposed to be a close call got 61% of the vote.

Someone probably got fired in this case because the question wasn't written in a manner sufficient to get the answer they really wanted.In polling it is all about how the questions are worded. And as they thought when they got an "impossible result" either the question was written poorly or alternatively "perhaps respondents just didn't understand the question." Regular people, after all, aren't as smart as pollsters, pundits and journalists.

Wiley Coyote said...

...and today, they say the Earth is cooling faster than they thought.

Imagine that.

You can't use "liberal" and "fact" in the same sentence. It's like trying to attach two magnets together using the north end of each.

silicon28 said...

So Can't stand you thinks that the "liberal media exists in a cocoon?"

Then I guess Karl Rove's election night meltdown on Fox News was just "distrusting the model?"

Lotta words used to say nothing here...

john barker said...

PPP has always been a left leaning organization, they just got caught this time! I cannot count the times they have tried to influence elections to go to democrats by not reporting good news for Republican candidates. Look at their pre-election polls and after election results in NC.