Gallup always thinking about changes, tweaks

People arrive to vote at a polling station in Eastern Market on Tuesday, November 6, 2012 in Washington, D.C.

President Obama set some interesting milestones with his re-election. No president since the Great Depression has been re-elected with unemployment this high and he's only the second to be awarded a second term with unemployment over seven percent -- the other was Ronald Reagan.

Looking back at those numbers, pollsters say it's easy to understand why some were quick to criticize or disbelieve the polls showing Obama with a lead going into election day. Yet after all was said and done, according to Frank Newport, editor-in-chief of Gallup, this election proved that "polling still works...old-fashioned polling across the board for most firms this time actually ended up being the best way to gauge what was happening out there."

In retrospect, Newport says pollsters should think about changing how they reach people with cellular phones. And perhaps most interestingly, Newport says his firm may rethink how it judges who a "likely voter" is, based in part on how successful President Obama was in getting people to the polls. "Gallup has a likely voter model that we've been using for many years," Newport says. "But I think Obama's people actually turned out in somewhat higher numbers than might have been anticipated based on the way that we've analyzed that historically."

"So, we'll analyze again how we isolate likely voters because that's a real key to try an figure out who's going to actually vote and how they are going to vote on Election Day," he adds.

About the author

Frank Newport, Ph.D., is the editor-in-chief at Gallup and appears regularly on Marketplace.
Log in to post3 Comments

Gallup had its biggest failure since 1948 (remember "Dewey beats Truman"?). Can't Marketplace interview somebody other than the incompetent Mr. Newport? Show a little pride, Marketplace, in your journalism.

@RX: agreed. Newport, as he did in yesterday afternoon's Marketplace interview, blithely maintains that polling works, barring some tweaks. Three weeks ago Newport was defending Gallup's 4% lead for Romney when nearly no one else projected this kind of advantage. Rather than showing professional curiosity and interest in how other pollsters have such accurate models, he continues to simply maintain that aging Gallup just needs a few small adjustments. This implies either a lack of humility or a lack of awareness--neither is trustworthy. And that's a damn shame, because I think someone with decades of experience could be an excellent contributor to the show.

I have always enjoyed Mr. Newport's observations on Marketplace. But, his response to Gallup's inaccurate prediction for the presidential elections was disappointing. Being a numbers guy, I thought that he would man up and simply say "well our model was wrong, hats off to Nate Silver, Linzer, Wang et al. for the way they utilized the data that was out there". Going off on aggregator pollsters was not the right response, I feel, and quite surprising for someone who claims to let numbers talk (which he usually does 90% of the time). My survey says "Gallup's tweaks left it out in the cold compared to everyone else" :-). Just shows that even seasoned survey specialists can fall under the all-pervasive influence of bias, I suppose.

With Generous Support From...