17

Post-election, a polling conundrum

U.S. President Barack Obama stands on stage after his victory speech at McCormick Place Nov. 6, 2012 in Chicago, Ill.

Today -- or rather, last night -- was a big day for pollsters across the country and that includes Gallup. We talk to Frank Newport, editor-in-chief at Gallup, for our Attitude Check segment.

Early predictions from Gallup had suggested Mitt Romney would win the election. In its last poll before votes were tallied, Gallup predicted the popular vote would be tight, but that Romney would win by one point. Newport points out that some of the people  who predicted Obama would win, including Nate Silver, don't do their own polling. They aggregate polls from many organizations, and Newport says this could have long-term effects on the polling industry that could result in fewer firms conducting their own polls.

Asked what he thought was most surprising about the election, Newport said turnout among non-white voters was one. But perhaps the biggest issue (though not necessarily a surprise) was that the candidate who was seen as being most trusted on the economy didn't win. Romney wasn't able to translate his success as a businessman into an election win.

About the author

Frank Newport, Ph.D., is the editor-in-chief at Gallup and appears regularly on Marketplace.
Log in to post17 Comments

Pages

PLEASE! Marketplace, you seem like nice thoughtful people. Stop using this hack. He missed in 2008 by 4 points, in 2010 by 8 and this year by 7, the WORST among all major pollsters. One simple reason, kids. Frank Newport continues to believe that the electorate is whiter and older than it is. He projected an electorate that was not only Whiter than 2008 but Whiter than 2004, and older than not only 2008 but than 2004. This despite the census and voting facts that the electorate has gotten lss white and younger for ever election since 1980!!! Lose this hack

What a whiner! Gallup was complete trash this cycle and Silver outed them. Frank Newport isn't fooling anyone.

Time to cut Frank and this segment loose!

I admit I'm a bit late to this party, but I agree that Marketplace can do better than Gallup ... if it feels the need for any relationship with a pollster at all. The question isn't just 'why did Gallup do so poorly in this election.' The real question is how many other non-electoral polls it has gotten wrong. The Newport segments have often left me scratching my head, or bored to tears.
And if you want to know just how terrible Gallup did check out Silver's Nov 10 scorecard in which the once venerable polling firm came in dead last among two dozen pollsters who conducted at least five surveys in the last 3 weeks of the campaign.
http://fivethirtyeight.blogs.nytimes.com/2012/11/10/which-polls-fared-be...

Newport's comments left me baffled. When Kai said (paraphrasing), "The headline says Nate Silver got it right and Gallup got it wrong. How do you react?", Newport said, "I don't agree at all." His *next* sentence is that Nate got it right. So where, exactly, is his disagreement?

Newport goes on to say that no polls would exist if everyone aggregated. This ignores the key part of Silver's commentary on Gallup--and national polls in general--which is that they aren't very accurate. Accuracy comes from aggregating STATE polls. If the national polls dried up and died, there would still be significant utility in the state polls. There's no argument that they would disappear.

Finally, Newport clings to the idea that Gallup was statistically correct when they predicted a very close race separated by a point or two. Um. Yes, true. But you predicted that Romney would win. Do you see the problem, Frank? With elections, most people have a single question they're looking to answer. Who will win? Your poll consistently said "Romney". Through this brutal binary lens, Gallup was 100% wrong.

I'd like to see Marketplace move away from their relationship with Gallup in pursuit of thought-leaders in the polling community. And I think there's an interesting story to be explored on the decline of Gallup and other national pollsters.

There are the actual polling statistics.

There is the interpretation of those statistics.

There is the reporting on the interpretation of the statistics.

Statements like... "Gallup shows Obama trailing by five points nationally." from newspapers like the Examiner are problematic for me because they're not meaningful. This is not how our elections work.

I do think that their polls and the coverage they received in the news motivated a lot of Democrats to go out and vote though. So for that I thank Gallup.

Marketplace should seriously consider revamping this regular segment, or dropping it altogether. Polling is a fascinating subject, but interviewing only a single practitioner, the one with arguably the least trustworthy record in the political sphere, is extremely problematic. Why exactly should Marketplace be supporting the poorest performer in this space?

Will this guy ever admit that Nate Silver may, in fact, know what he's doing? How can Newport still insist there is no logic in Silver's statistics after Silver correctly predicted the election outcomes of every state in the union?

Newport just seems like a sore loser at this point.

"These aggregators don't exist without people who are out there actually doing polls."

Yes, yes, we all understand the Tragedy of the Commons.

Frank is so focused on chastising Nate for using data that pollsters make publicly available that he's missed the point completely.

538 took a bunch of information that was out there for the asking, rolled it up, and presented it. That's EXACTLY what Gallup does, except with different datacrunching methodologies. Nobody's calling Gallup out for single-mindedly focusing on methodology that led to the wrong answer, even if it was 1 point.

538 owes Gallup a "thank you," but Gallup owes everyone an explanation.

Dr. Newport's comments had a tone of professional jealousy when he discussed Mr. Silver's incredibly accurate work. His excuses for the inaccurate Gallup analyses demonstrated that his organization is a dinosaur unable to adapt to more scientifically supportable statistical analysis. Maybe it's time to look at inviting Mr. Silver in place of Dr. Newport.

Pages

With Generous Support From...